diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/README.md b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/README.md new file mode 100644 index 000000000..3cadea3bc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/README.md @@ -0,0 +1,171 @@ +# 电梯内电瓶车入室检测 + +## 内容 + +* [项目说明](#项目说明) +* [安装说明](#安装说明) +* [数据准备](#数据准备) +* [模型选择](#模型选择) +* [模型训练](#模型训练) +* [模型导出](#模型导出) +* [检索库生成](#检索库生成) +* [检索库部署](#检索库部署) + + + +近年来，电瓶车进楼入户发生的火灾事故屡见不鲜，针对该现象推出了相应的电瓶车入室检测模型，旨在从源头减少这一情况的发生。 针对室内摩托车模型可能会发生的误报情况，采用了额外的图像检索方式实现更为精确的识别。 本案例使用了飞桨目标检测套件PaddleDetection中的picodet模型以及图像识别套件PaddleClas中的轻量级通用识别模型。 + +![demo](docs/images/result_5.png) + +注:AI Studio在线运行代码请参考[电梯内电瓶车检测全流程](https://aistudio.baidu.com/aistudio/projectdetail/3497217?channelType=0&channel=0)(配置gpu资源) +## 2 安装说明 + +##### 环境要求 + +* PaddlePaddle = 2.2.2 +* Python >= 3.5 + + + +## 3 数据准备 + +本案例中picodet的模型数据集为VOC格式(使用labelimg制成)，包括21903张电梯中的图片，其中训练集17522张，测试集4381张，皆来自日常的电梯场景中，共有14715个摩托车的框，23058个人的框，3750个自行车的框，由于picodet使用的是coco格式，所以需要将VOC格式转换成coco格式。 生成VOC数据集：使用python的labelimg图像标注工具为原始图片生成对应的标注xml文件作为原始的VOC格式数据集，生成的xml文件格式如下图所示，其中每个object代表框出的每一个对象，object中的name表明对象的名字而bndbox中包含框的具体坐标（左上角以及右下角）。 + +![label_img][docs/images/label_img.png] + +![xml_content](docs/images/xml_content.png) + + +生成VOC数据集： 完成图片标注后，下一步就是生成数据集，将每个图片与其xml对应起来按比例生成对应的训练集以及测试集. + +``` +├── classify_voc.py +├── picodet_motorcycle +│ ├── Annotations +│ │ ├── 1595214506200933-1604535322-[]-motorcycle.xml +│ │ ├── 1595214506200933-1604542813-[]-motorcycle.xml +│ │ ├── 1595214506200933-1604559538-[]-motorcycle.xml +| ... +│ ├── ImageSets +│ │ └── Main +│ │ ├── test.txt +│ │ ├── train.txt +│ │ ├── trainval.txt +│ │ └── val.txt +│ └── JPEGImages +│ ├── 1595214506200933-1604535322-[]-motorcycle.jpg +│ ├── 1595214506200933-1604542813-[]-motorcycle.jpg +│ ├── 1595214506200933-1604559538-[]-motorcycle.jpg +│ | ... +├── picodet_motorcycle.zip +├── prepare_voc_data.py +├── test.txt +└── trainval.txt +``` + +VOC数据集 [下载地址](https://aistudio.baidu.com/aistudio/datasetdetail/128282) +检索库数据集 [下载地址](https://aistudio.baidu.com/aistudio/datasetdetail/128448) +将VOC格式的数据集转换为coco格式（使用paddle自带的转换脚本）： +仅举例说明，使用时请修改路径 +``` +python x2coco.py --dataset_type voc --voc_anno_dir /home/aistudio/data/data128282/ --voc_anno_list /home/aistudio/data/data128282/trainval.txt --voc_label_list /home/aistudio/data/data128282/label_list.txt --voc_out_name voc_train.json +python x2coco.py --dataset_type voc --voc_anno_dir /home/aistudio/data/data128282/ --voc_anno_list /home/aistudio/data/data128282/test.txt --voc_label_list /home/aistudio/data/data128282/label_list.txt --voc_out_name voc_test.json +mv voc_test.json /home/aistudio/data/data128282/ +mv voc_train.json /home/aistudio/data/data128282/ + +``` + + +## 4 模型选择 + +本案例选择了PaddleDetection中提出了全新的轻量级系列模型PP-PicoDet + +PP-PicoDet模型有如下特点： + + - 更高的mAP: 第一个在1M参数量之内mAP(0.5:0.95)超越30+(输入416像素时)。 + - 更快的预测速度: 网络预测在ARM CPU下可达150FPS。 + - 部署友好: 支持PaddleLite/MNN/NCNN/OpenVINO等预测库，支持转出ONNX，提供了C++/Python/Android的demo。 + - 先进的算法: 在现有SOTA算法中进行了创新, 包括：ESNet, CSP-PAN, SimOTA等等。 + + + + +## 5 模型训练 + + +首先安装依赖库 +``` +cd code/train/ +pip install pycocotools +pip install faiss-gpu +pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple +``` + +导出为serving模型的准备 +``` +pip install paddle-serving-app==0.6.2 -i https://pypi.tuna.tsinghua.edu.cn/simple +pip install paddle-serving-client==0.6.2 -i https://pypi.tuna.tsinghua.edu.cn/simple +pip install paddle-serving-server-gpu==0.6.3.post102 -i https://pypi.tuna.tsinghua.edu.cn/simple +``` + + + +## 6 模型导出 + + +导出为serving模型 +``` +cd code/train/ +python export_model.py --export_serving_model=true -c picodet_lcnet_1_5x_416_coco.yml --output_dir=./output_inference/ +``` + +``` +cd code/train/output_inference/picodet_lcnet_1_5x_416_coco/ +mv serving_server/ code/picodet_lcnet_1_5x_416_coco/ +``` + +启动服务 +``` +cd /home/aistudio/work/code/picodet_lcnet_1_5x_416_coco/ +python3 web_service.py +``` + +请求结果如下图所示: + +![infer_result](docs/images/infer_result.png) + + + +## 7 检索库生成 + +在目标检测模型部署完毕后电瓶车入室检测的功能便可投入使用，但为了提高整体的准确度减少误报则还需要一个额外的检索方式。这里采用了PaddleClas下图像识别中的轻量级通用识别模型general_PPLCNet_x2_5_lite_v1.0_infer + +首先从paddle下载解压模型并导出为serving模型 +``` +cd code/ +wget -P models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/rec/models/inference/general_PPLCNet_x2_5_lite_v1.0_infer.tar +cd models +tar -xf general_PPLCNet_x2_5_lite_v1.0_infer.tar +python3 -m paddle_serving_client.convert --dirname ./general_PPLCNet_x2_5_lite_v1.0_infer/ --model_filename inference.pdmodel --params_filename inference.pdiparams --serving_server ./general_PPLCNet_x2_5_lite_v1.0_serving/ --serving_client ./general_PPLCNet_x2_5_lite_v1.0_client/ +cp -r ./general_PPLCNet_x2_5_lite_v1.0_serving ../general_PPLCNet_x2_5_lite_v1.0/ +``` + +解压数据集后根据路径修改make_label.py,随后生成索引库. +``` +cd code +python make_label.py +python python/build_gallery.py -c build_gallery/build_general.yaml -o IndexProcess.data_file="./index_label.txt" -o IndexProcess.index_dir="index_result" +mv index_result/ general_PPLCNet_x2_5_lite_v1.0/ +``` + + + +## 7 检索库部署 +``` +cd /home/aistudio/work/code/general_PPLCNet_x2_5_lite_v1.0/ +python recognition_web_service_onlyrec.py +``` + +在实际场景中请求结果如图所示。 + +![index_infer_result](docs/images/index_infer_result.png) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/build_gallery.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/build_gallery.cpython-37.pyc new file mode 100644 index 000000000..0127dba01 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/build_gallery.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/det_preprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/det_preprocess.cpython-37.pyc new file mode 100644 index 000000000..5d7e3c23a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/det_preprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/postprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/postprocess.cpython-37.pyc new file mode 100644 index 000000000..48953d4a7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/postprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/predict_rec.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/predict_rec.cpython-37.pyc new file mode 100644 index 000000000..661ed04d7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/predict_rec.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/preprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/preprocess.cpython-37.pyc new file mode 100644 index 000000000..4369f8ab4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/__pycache__/preprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_gallery.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_gallery.py new file mode 100644 index 000000000..5a7d82fb5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_gallery.py @@ -0,0 +1,213 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import faiss +import numpy as np +from tqdm import tqdm +import pickle +from predict_rec import RecPredictor + +from utils import logger +from utils import config + + +def split_datafile(data_file, image_root, delimiter="\t"): + ''' + data_file: image path and info, which can be splitted by spacer + image_root: image path root + delimiter: delimiter + ''' + gallery_images = [] + gallery_docs = [] + with open(data_file, 'r', encoding='utf-8') as f: + lines = f.readlines() + for _, ori_line in enumerate(lines): + line = ori_line.strip().split(delimiter) + text_num = len(line) + assert text_num >= 2, f"line({ori_line}) must be splitted into at least 2 parts, but got {text_num}" + image_file = os.path.join(image_root, line[0]) + + gallery_images.append(image_file) + gallery_docs.append(ori_line.strip()) + + return gallery_images, gallery_docs + + +class GalleryBuilder(object): + def __init__(self, config): + + self.config = config + self.rec_predictor = RecPredictor(config) + assert 'IndexProcess' in config.keys(), "Index config not found ... " + self.build(config['IndexProcess']) + + def build(self, config): + ''' + build index from scratch + ''' + operation_method = config.get("index_operation", "new").lower() + + gallery_images, gallery_docs = split_datafile( + config['data_file'], config['image_root'], config['delimiter']) + + # when remove data in index, do not need extract fatures + if operation_method != "remove": + gallery_features = self._extract_features(gallery_images, config) + assert operation_method in [ + "new", "remove", "append" + ], "Only append, remove and new operation are supported" + + # vector.index: faiss index file + # id_map.pkl: use this file to map id to image_doc + if operation_method in ["remove", "append"]: + # if remove or append, vector.index and id_map.pkl must exist + assert os.path.join( + config["index_dir"], "vector.index" + ), "The vector.index dose not exist in {} when 'index_operation' is not None".format( + config["index_dir"]) + assert os.path.join( + config["index_dir"], "id_map.pkl" + ), "The id_map.pkl dose not exist in {} when 'index_operation' is not None".format( + config["index_dir"]) + index = faiss.read_index( + os.path.join(config["index_dir"], "vector.index")) + with open(os.path.join(config["index_dir"], "id_map.pkl"), + 'rb') as fd: + ids = pickle.load(fd) + assert index.ntotal == len(ids.keys( + )), "data number in index is not equal in in id_map" + else: + if not os.path.exists(config["index_dir"]): + os.makedirs(config["index_dir"], exist_ok=True) + index_method = config.get("index_method", "HNSW32") + + # if IVF method, cal ivf number automaticlly + if index_method == "IVF": + index_method = index_method + str( + min(int(len(gallery_images) // 8), 65536)) + ",Flat" + + # for binary index, add B at head of index_method + if config["dist_type"] == "hamming": + index_method = "B" + index_method + + #dist_type + dist_type = faiss.METRIC_INNER_PRODUCT if config[ + "dist_type"] == "IP" else faiss.METRIC_L2 + + #build index + if config["dist_type"] == "hamming": + index = faiss.index_binary_factory(config["embedding_size"], + index_method) + else: + index = faiss.index_factory(config["embedding_size"], + index_method, dist_type) + index = faiss.IndexIDMap2(index) + ids = {} + + if config["index_method"] == "HNSW32": + logger.warning( + "The HNSW32 method dose not support 'remove' operation") + + if operation_method != "remove": + # calculate id for new data + start_id = max(ids.keys()) + 1 if ids else 0 + ids_now = ( + np.arange(0, len(gallery_images)) + start_id).astype(np.int64) + + # only train when new index file + if operation_method == "new": + if config["dist_type"] == "hamming": + index.add(gallery_features) + else: + index.train(gallery_features) + + if not config["dist_type"] == "hamming": + index.add_with_ids(gallery_features, ids_now) + + for i, d in zip(list(ids_now), gallery_docs): + ids[i] = d + else: + if config["index_method"] == "HNSW32": + raise RuntimeError( + "The index_method: HNSW32 dose not support 'remove' operation" + ) + # remove ids in id_map, remove index data in faiss index + remove_ids = list( + filter(lambda k: ids.get(k) in gallery_docs, ids.keys())) + remove_ids = np.asarray(remove_ids) + index.remove_ids(remove_ids) + for k in remove_ids: + del ids[k] + + # store faiss index file and id_map file + if config["dist_type"] == "hamming": + faiss.write_index_binary( + index, os.path.join(config["index_dir"], "vector.index")) + else: + faiss.write_index( + index, os.path.join(config["index_dir"], "vector.index")) + + with open(os.path.join(config["index_dir"], "id_map.pkl"), 'wb') as fd: + pickle.dump(ids, fd) + + def _extract_features(self, gallery_images, config): + # extract gallery features + if config["dist_type"] == "hamming": + gallery_features = np.zeros( + [len(gallery_images), config['embedding_size'] // 8], + dtype=np.uint8) + else: + gallery_features = np.zeros( + [len(gallery_images), config['embedding_size']], + dtype=np.float32) + + #construct batch imgs and do inference + batch_size = config.get("batch_size", 32) + batch_img = [] + for i, image_file in enumerate(tqdm(gallery_images)): + img = cv2.imread(image_file) + if img is None: + logger.error("img empty, please check {}".format(image_file)) + exit() + img = img[:, :, ::-1] + batch_img.append(img) + + if (i + 1) % batch_size == 0: + rec_feat = self.rec_predictor.predict(batch_img) + gallery_features[i - batch_size + 1:i + 1, :] = rec_feat + batch_img = [] + + if len(batch_img) > 0: + rec_feat = self.rec_predictor.predict(batch_img) + gallery_features[-len(batch_img):, :] = rec_feat + batch_img = [] + + return gallery_features + + +def main(config): + GalleryBuilder(config) + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_general.yaml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_general.yaml new file mode 100644 index 000000000..5b83ea4d4 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/build_general.yaml @@ -0,0 +1,36 @@ +Global: + rec_inference_model_dir: "./models/general_PPLCNet_x2_5_lite_v1.0_infer" + batch_size: 32 + use_gpu: False + enable_mkldnn: True + cpu_num_threads: 10 + enable_benchmark: True + use_fp16: False + ir_optim: True + use_tensorrt: False + gpu_mem: 8000 + enable_profile: False + +RecPreProcess: + transform_ops: + - ResizeImage: + size: 224 + - NormalizeImage: + scale: 0.00392157 + mean: [0.485, 0.456, 0.406] + std: [0.229, 0.224, 0.225] + order: '' + - ToCHWImage: + +RecPostProcess: null + +# indexing engine config +IndexProcess: + index_method: "HNSW32" # supported: HNSW32, IVF, Flat + image_root: "" + index_dir: "./images/index" + data_file: "./images/motorcyclebike_label_all_02.txt" + index_operation: "new" # suported: "append", "remove", "new" + delimiter: "\t" + dist_type: "IP" + embedding_size: 512 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/det_preprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/det_preprocess.py new file mode 100644 index 000000000..65db32dc3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/det_preprocess.py @@ -0,0 +1,216 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import cv2 +import numpy as np + + +def decode_image(im_file, im_info): + """read rgb image + Args: + im_file (str|np.ndarray): input can be image path or np.ndarray + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + if isinstance(im_file, str): + with open(im_file, 'rb') as f: + im_read = f.read() + data = np.frombuffer(im_read, dtype='uint8') + im = cv2.imdecode(data, 1) # BGR mode, but need RGB mode + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + else: + im = im_file + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + im_info['scale_factor'] = np.array([1., 1.], dtype=np.float32) + return im, im_info + + +class DetResize(object): + """resize image by target_size and max_size + Args: + target_size (int): the target size of image + keep_ratio (bool): whether keep_ratio or not, default true + interp (int): method of resize + """ + + def __init__( + self, + target_size, + keep_ratio=True, + interp=cv2.INTER_LINEAR, ): + if isinstance(target_size, int): + target_size = [target_size, target_size] + self.target_size = target_size + self.keep_ratio = keep_ratio + self.interp = interp + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + assert len(self.target_size) == 2 + assert self.target_size[0] > 0 and self.target_size[1] > 0 + im_channel = im.shape[2] + im_scale_y, im_scale_x = self.generate_scale(im) + # set image_shape + im_info['input_shape'][1] = int(im_scale_y * im.shape[0]) + im_info['input_shape'][2] = int(im_scale_x * im.shape[1]) + print(0000000000000000000000000000000000000000) + print(im) + print(im_scale_x,im_scale_y,cv2.INTER_LINEAR,self.interp) + im = cv2.resize( + im, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + print(im) + im_info['im_shape'] = np.array(im.shape[:2]).astype('float32') + im_info['scale_factor'] = np.array( + [im_scale_y, im_scale_x]).astype('float32') + return im, im_info + + def generate_scale(self, im): + """ + Args: + im (np.ndarray): image (np.ndarray) + Returns: + im_scale_x: the resize ratio of X + im_scale_y: the resize ratio of Y + """ + origin_shape = im.shape[:2] + im_c = im.shape[2] + if self.keep_ratio: + im_size_min = np.min(origin_shape) + im_size_max = np.max(origin_shape) + target_size_min = np.min(self.target_size) + target_size_max = np.max(self.target_size) + im_scale = float(target_size_min) / float(im_size_min) + if np.round(im_scale * im_size_max) > target_size_max: + im_scale = float(target_size_max) / float(im_size_max) + im_scale_x = im_scale + im_scale_y = im_scale + else: + resize_h, resize_w = self.target_size + im_scale_y = resize_h / float(origin_shape[0]) + im_scale_x = resize_w / float(origin_shape[1]) + return im_scale_y, im_scale_x + + +class DetNormalizeImage(object): + """normalize image + Args: + mean (list): im - mean + std (list): im / std + is_scale (bool): whether need im / 255 + is_channel_first (bool): if True: image shape is CHW, else: HWC + """ + + def __init__(self, mean, std, is_scale=True): + self.mean = mean + self.std = std + self.is_scale = is_scale + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + im = im.astype(np.float32, copy=False) + mean = np.array(self.mean)[np.newaxis, np.newaxis, :] + std = np.array(self.std)[np.newaxis, np.newaxis, :] + if self.is_scale: + im = im / 255.0 + print(im) + im -= mean + im /= std + return im, im_info + + +class DetPermute(object): + """permute image + Args: + to_bgr (bool): whether convert RGB to BGR + channel_first (bool): whether convert HWC to CHW + """ + + def __init__(self, ): + super().__init__() + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + #im = im.transpose((2, 0, 1)).copy() + print("detprossssssss") + print(im) + im = im.transpose((2, 0, 1)).copy() + print(im) + return im, im_info + + +class DetPadStride(object): + """ padding image for model with FPN , instead PadBatch(pad_to_stride, pad_gt) in original config + Args: + stride (bool): model with FPN need image shape % stride == 0 + """ + + def __init__(self, stride=0): + self.coarsest_stride = stride + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + coarsest_stride = self.coarsest_stride + if coarsest_stride <= 0: + return im, im_info + im_c, im_h, im_w = im.shape + pad_h = int(np.ceil(float(im_h) / coarsest_stride) * coarsest_stride) + pad_w = int(np.ceil(float(im_w) / coarsest_stride) * coarsest_stride) + padding_im = np.zeros((im_c, pad_h, pad_w), dtype=np.float32) + padding_im[:, :im_h, :im_w] = im + return padding_im, im_info + + +def det_preprocess(im, im_info, preprocess_ops): + for operator in preprocess_ops: + print(operator) + print(im) + print(666) + im, im_info = operator(im, im_info) + print(im) + return im, im_info diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/postprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/postprocess.py new file mode 100644 index 000000000..d26cbaa9a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/postprocess.py @@ -0,0 +1,161 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import copy +import shutil +from functools import partial +import importlib +import numpy as np +import paddle +import paddle.nn.functional as F + + +def build_postprocess(config): + if config is None: + return None + + mod = importlib.import_module(__name__) + config = copy.deepcopy(config) + + main_indicator = config.pop( + "main_indicator") if "main_indicator" in config else None + main_indicator = main_indicator if main_indicator else "" + + func_list = [] + for func in config: + func_list.append(getattr(mod, func)(**config[func])) + return PostProcesser(func_list, main_indicator) + + +class PostProcesser(object): + def __init__(self, func_list, main_indicator="Topk"): + self.func_list = func_list + self.main_indicator = main_indicator + + def __call__(self, x, image_file=None): + rtn = None + for func in self.func_list: + tmp = func(x, image_file) + if type(func).__name__ in self.main_indicator: + rtn = tmp + return rtn + + +class Topk(object): + def __init__(self, topk=1, class_id_map_file=None): + assert isinstance(topk, (int, )) + self.class_id_map = self.parse_class_id_map(class_id_map_file) + self.topk = topk + + def parse_class_id_map(self, class_id_map_file): + if class_id_map_file is None: + return None + + if not os.path.exists(class_id_map_file): + print( + "Warning: If want to use your own label_dict, please input legal path!\nOtherwise label_names will be empty!" + ) + return None + + try: + class_id_map = {} + with open(class_id_map_file, "r") as fin: + lines = fin.readlines() + for line in lines: + partition = line.split("\n")[0].partition(" ") + class_id_map[int(partition[0])] = str(partition[-1]) + except Exception as ex: + print(ex) + class_id_map = None + return class_id_map + + def __call__(self, x, file_names=None, multilabel=False): + if file_names is not None: + assert x.shape[0] == len(file_names) + y = [] + for idx, probs in enumerate(x): + index = probs.argsort(axis=0)[-self.topk:][::-1].astype( + "int32") if not multilabel else np.where( + probs >= 0.5)[0].astype("int32") + clas_id_list = [] + score_list = [] + label_name_list = [] + for i in index: + clas_id_list.append(i.item()) + score_list.append(probs[i].item()) + if self.class_id_map is not None: + label_name_list.append(self.class_id_map[i.item()]) + result = { + "class_ids": clas_id_list, + "scores": np.around( + score_list, decimals=5).tolist(), + } + if file_names is not None: + result["file_name"] = file_names[idx] + if label_name_list is not None: + result["label_names"] = label_name_list + y.append(result) + return y + + +class MultiLabelTopk(Topk): + def __init__(self, topk=1, class_id_map_file=None): + super().__init__() + + def __call__(self, x, file_names=None): + return super().__call__(x, file_names, multilabel=True) + + +class SavePreLabel(object): + def __init__(self, save_dir): + if save_dir is None: + raise Exception( + "Please specify save_dir if SavePreLabel specified.") + self.save_dir = partial(os.path.join, save_dir) + + def __call__(self, x, file_names=None): + if file_names is None: + return + assert x.shape[0] == len(file_names) + for idx, probs in enumerate(x): + index = probs.argsort(axis=0)[-1].astype("int32") + self.save(index, file_names[idx]) + + def save(self, id, image_file): + output_dir = self.save_dir(str(id)) + os.makedirs(output_dir, exist_ok=True) + shutil.copy(image_file, output_dir) + + +class Binarize(object): + def __init__(self, method="round"): + self.method = method + self.unit = np.array([[128, 64, 32, 16, 8, 4, 2, 1]]).T + + def __call__(self, x, file_names=None): + if self.method == "round": + x = np.round(x + 1).astype("uint8") - 1 + + if self.method == "sign": + x = ((np.sign(x) + 1) / 2).astype("uint8") + + embedding_size = x.shape[1] + assert embedding_size % 8 == 0, "The Binary index only support vectors with sizes multiple of 8" + + byte = np.zeros([x.shape[0], embedding_size // 8], dtype=np.uint8) + for i in range(embedding_size // 8): + byte[:, i:i + 1] = np.dot(x[:, i * 8:(i + 1) * 8], self.unit) + + return byte diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_cls.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_cls.py new file mode 100644 index 000000000..cdeb32e48 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_cls.py @@ -0,0 +1,140 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import numpy as np + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from python.preprocess import create_operators +from python.postprocess import build_postprocess + + +class ClsPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"]) + + self.preprocess_ops = [] + self.postprocess = None + if "PreProcess" in config: + if "transform_ops" in config["PreProcess"]: + self.preprocess_ops = create_operators(config["PreProcess"][ + "transform_ops"]) + if "PostProcess" in config: + self.postprocess = build_postprocess(config["PostProcess"]) + + # for whole_chain project to test each repo of paddle + self.benchmark = config["Global"].get("benchmark", False) + if self.benchmark: + import auto_log + import os + pid = os.getpid() + self.auto_logger = auto_log.AutoLogger( + model_name=config["Global"].get("model_name", "cls"), + model_precision='fp16' + if config["Global"]["use_fp16"] else 'fp32', + batch_size=config["Global"].get("batch_size", 1), + data_shape=[3, 224, 224], + save_path=config["Global"].get("save_log_path", + "./auto_log.log"), + inference_config=self.config, + pids=pid, + process_name=None, + gpu_ids=None, + time_keys=[ + 'preprocess_time', 'inference_time', 'postprocess_time' + ], + warmup=2) + + def predict(self, images): + input_names = self.paddle_predictor.get_input_names() + input_tensor = self.paddle_predictor.get_input_handle(input_names[0]) + + output_names = self.paddle_predictor.get_output_names() + output_tensor = self.paddle_predictor.get_output_handle(output_names[ + 0]) + if self.benchmark: + self.auto_logger.times.start() + if not isinstance(images, (list, )): + images = [images] + for idx in range(len(images)): + for ops in self.preprocess_ops: + images[idx] = ops(images[idx]) + image = np.array(images) + if self.benchmark: + self.auto_logger.times.stamp() + + input_tensor.copy_from_cpu(image) + self.paddle_predictor.run() + batch_output = output_tensor.copy_to_cpu() + if self.benchmark: + self.auto_logger.times.stamp() + if self.postprocess is not None: + batch_output = self.postprocess(batch_output) + if self.benchmark: + self.auto_logger.times.end(stamp=True) + return batch_output + + +def main(config): + cls_predictor = ClsPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + batch_imgs = [] + batch_names = [] + cnt = 0 + for idx, img_path in enumerate(image_list): + img = cv2.imread(img_path) + if img is None: + logger.warning( + "Image file failed to read and has been skipped. The path: {}". + format(img_path)) + else: + img = img[:, :, ::-1] + batch_imgs.append(img) + img_name = os.path.basename(img_path) + batch_names.append(img_name) + cnt += 1 + + if cnt % config["Global"]["batch_size"] == 0 or (idx + 1 + ) == len(image_list): + if len(batch_imgs) == 0: + continue + batch_results = cls_predictor.predict(batch_imgs) + for number, result_dict in enumerate(batch_results): + filename = batch_names[number] + clas_ids = result_dict["class_ids"] + scores_str = "[{}]".format(", ".join("{:.2f}".format( + r) for r in result_dict["scores"])) + label_names = result_dict["label_names"] + print("{}:\tclass id(s): {}, score(s): {}, label_name(s): {}". + format(filename, clas_ids, scores_str, label_names)) + batch_imgs = [] + batch_names = [] + if cls_predictor.benchmark: + cls_predictor.auto_logger.report() + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det.py new file mode 100644 index 000000000..0b9c25a5a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det.py @@ -0,0 +1,195 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from det_preprocess import det_preprocess +from preprocess import create_operators +from utils.draw_bbox import draw_bbox_results + +import os +import argparse +import time +import yaml +import ast +from functools import reduce +import cv2 +import numpy as np +import paddle +import requests +import base64 +import json +class DetPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["det_inference_model_dir"]) + + self.preprocess_ops = create_operators(config["DetPreProcess"][ + "transform_ops"]) + self.config = config + + def preprocess(self, img): + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': np.array( + img.shape[:2], dtype=np.float32), + 'input_shape': self.config["Global"]["image_shape"], + "scale_factor": np.array( + [1., 1.], dtype=np.float32) + } + + im, im_info = det_preprocess(img, im_info, self.preprocess_ops) + print(111111111111111111111) + print(im) + inputs = self.create_inputs(im, im_info) + return inputs + + def create_inputs(self, im, im_info): + """generate input for different model type + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + model_arch (str): model type + Returns: + inputs (dict): input of model + """ + inputs = {} + inputs['image'] = np.array((im, )).astype('float32') + inputs['im_shape'] = np.array( + (im_info['im_shape'], )).astype('float32') + inputs['scale_factor'] = np.array( + (im_info['scale_factor'], )).astype('float32') + #print(inputs) + return inputs + + def parse_det_results(self, pred, threshold, label_list): + max_det_results = self.config["Global"]["max_det_results"] + keep_indexes = pred[:, 1].argsort()[::-1][:max_det_results] + results = [] + for idx in keep_indexes: + single_res = pred[idx] + class_id = int(single_res[0]) + score = single_res[1] + bbox = single_res[2:] + if score < threshold: + continue + label_name = label_list[class_id] + ''' + results.append({ + "class_id": class_id, + "score": score, + "bbox": bbox, + "label_name": label_name, + })''' + results.append({ + "bbox": bbox, + "rec_docs": "background", + "rec_scores": score, + }) + return results + + def predict(self, image, threshold=0.5, run_benchmark=False): + ''' + Args: + image (str/np.ndarray): path of image/ np.ndarray read by cv2ps + threshold (float): threshold of predicted box' score + Returns: + results (dict): include 'boxes': np.ndarray: shape:[N,6], N: number of box, + matix element:[class, score, x_min, y_min, x_max, y_max] + MaskRCNN's results include 'masks': np.ndarray: + shape: [N, im_h, im_w] + ''' + inputs = self.preprocess(image) + print(str(inputs)) + np_boxes = None + input_names = self.paddle_predictor.get_input_names() + print(input_names) + for i in range(len(input_names)): + input_tensor = self.paddle_predictor.get_input_handle(input_names[ + i]) + input_tensor.copy_from_cpu(inputs[input_names[i]]) + print("!!!!!!!",inputs[input_names[i]]) + t1 = time.time() + print(self.paddle_predictor.run()) + output_names = self.paddle_predictor.get_output_names() + boxes_tensor = self.paddle_predictor.get_output_handle(output_names[0]) + + np_boxes = boxes_tensor.copy_to_cpu() + t2 = time.time() + + print("Inference: {} ms per batch image".format((t2 - t1) * 1000.0)) + + # do not perform postprocess in benchmark mode + results = [] + if reduce(lambda x, y: x * y, np_boxes.shape) < 6: + print('[WARNNING] No object detected.') + results = np.array([]) + else: + results = np_boxes + + results = self.parse_det_results(results, + self.config["Global"]["threshold"], + self.config["Global"]["labe_list"]) + return results + + +def main(config): + det_predictor = DetPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = det_predictor.predict(img) + print(output) + draw_bbox_results(img, output, image_file) + + return image_file,output + +def cv2_to_base64_img(img): + data = cv2.imencode('.jpg', img)[1] + return base64.b64encode(data.tostring()).decode('utf8') + +def solve_output(output,image_file): + print(image_file) + img = cv2.imread(image_file) + + for bbox in output: + left,top,right,bottom = int(bbox["bbox"][0]),int(bbox["bbox"][1]),int(bbox["bbox"][2]),int(bbox["bbox"][3]) + print(left,top,right,bottom) + img_crop = img[top:bottom,left:right] + url = "http://123.157.241.94:36807/ppyolo_mbv3/prediction" + img2 = {"key": ["image"], "value": [cv2_to_base64_img(img_crop)]} + r = requests.post(url=url,data=json.dumps(img2), timeout=5) + r = r.json() + print(r) + result = eval(r['value'][0])[0] + cv2.putText(img,str(round(float(result["scores"][0]),2)),(left,top+30), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) + cv2.putText(img,str(result["label_names"][0]),(left,top+60), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) + cv2.rectangle(img,(left ,top),(right,bottom), (0, 0, 255), 2) + cv2.imwrite("./output/ppyolo_result" + image_file[image_file.rfind("/"):],img) +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + image_file,output = main(config) + #solve_output(output,image_file) + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det_bak.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det_bak.py new file mode 100644 index 000000000..323d65ab1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_det_bak.py @@ -0,0 +1,167 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from det_preprocess import det_preprocess +from preprocess import create_operators +from utils.draw_bbox import draw_bbox_results + +import os +import argparse +import time +import yaml +import ast +from functools import reduce +import cv2 +import numpy as np +import paddle + + +class DetPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["det_inference_model_dir"]) + + self.preprocess_ops = create_operators(config["DetPreProcess"][ + "transform_ops"]) + self.config = config + + def preprocess(self, img): + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': np.array( + img.shape[:2], dtype=np.float32), + 'input_shape': self.config["Global"]["image_shape"], + "scale_factor": np.array( + [1., 1.], dtype=np.float32) + } + im, im_info = det_preprocess(img, im_info, self.preprocess_ops) + inputs = self.create_inputs(im, im_info) + return inputs + + def create_inputs(self, im, im_info): + """generate input for different model type + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + model_arch (str): model type + Returns: + inputs (dict): input of model + """ + inputs = {} + inputs['image'] = np.array((im, )).astype('float32') + inputs['im_shape'] = np.array( + (im_info['im_shape'], )).astype('float32') + inputs['scale_factor'] = np.array( + (im_info['scale_factor'], )).astype('float32') + print(inputs) + return inputs + + def parse_det_results(self, pred, threshold, label_list): + max_det_results = self.config["Global"]["max_det_results"] + keep_indexes = pred[:, 1].argsort()[::-1][:max_det_results] + results = [] + for idx in keep_indexes: + single_res = pred[idx] + class_id = int(single_res[0]) + score = single_res[1] + bbox = single_res[2:] + if score < threshold: + continue + label_name = label_list[class_id] + ''' + results.append({ + "class_id": class_id, + "score": score, + "bbox": bbox, + "label_name": label_name, + })''' + results.append({ + "bbox": bbox, + "rec_docs": "background", + "rec_scores": score, + }) + return results + + def predict(self, image, threshold=0.5, run_benchmark=False): + ''' + Args: + image (str/np.ndarray): path of image/ np.ndarray read by cv2 + threshold (float): threshold of predicted box' score + Returns: + results (dict): include 'boxes': np.ndarray: shape:[N,6], N: number of box, + matix element:[class, score, x_min, y_min, x_max, y_max] + MaskRCNN's results include 'masks': np.ndarray: + shape: [N, im_h, im_w] + ''' + inputs = self.preprocess(image) + np_boxes = None + input_names = self.paddle_predictor.get_input_names() + + for i in range(len(input_names)): + input_tensor = self.paddle_predictor.get_input_handle(input_names[ + i]) + input_tensor.copy_from_cpu(inputs[input_names[i]]) + + t1 = time.time() + self.paddle_predictor.run() + output_names = self.paddle_predictor.get_output_names() + boxes_tensor = self.paddle_predictor.get_output_handle(output_names[0]) + np_boxes = boxes_tensor.copy_to_cpu() + t2 = time.time() + + print("Inference: {} ms per batch image".format((t2 - t1) * 1000.0)) + + # do not perform postprocess in benchmark mode + results = [] + if reduce(lambda x, y: x * y, np_boxes.shape) < 6: + print('[WARNNING] No object detected.') + results = np.array([]) + else: + results = np_boxes + + results = self.parse_det_results(results, + self.config["Global"]["threshold"], + self.config["Global"]["labe_list"]) + return results + + +def main(config): + det_predictor = DetPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = det_predictor.predict(img) + print(output) + draw_bbox_results(img, output, image_file) + print(output) + + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_rec.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_rec.py new file mode 100644 index 000000000..d41c513f8 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_rec.py @@ -0,0 +1,105 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import numpy as np + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from preprocess import create_operators +from postprocess import build_postprocess + + +class RecPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["rec_inference_model_dir"]) + self.preprocess_ops = create_operators(config["RecPreProcess"][ + "transform_ops"]) + self.postprocess = build_postprocess(config["RecPostProcess"]) + + def predict(self, images, feature_normalize=True): + input_names = self.paddle_predictor.get_input_names() + input_tensor = self.paddle_predictor.get_input_handle(input_names[0]) + + output_names = self.paddle_predictor.get_output_names() + output_tensor = self.paddle_predictor.get_output_handle(output_names[ + 0]) + + if not isinstance(images, (list, )): + images = [images] + for idx in range(len(images)): + for ops in self.preprocess_ops: + images[idx] = ops(images[idx]) + image = np.array(images) + + input_tensor.copy_from_cpu(image) + self.paddle_predictor.run() + batch_output = output_tensor.copy_to_cpu() + + if feature_normalize: + feas_norm = np.sqrt( + np.sum(np.square(batch_output), axis=1, keepdims=True)) + batch_output = np.divide(batch_output, feas_norm) + + if self.postprocess is not None: + batch_output = self.postprocess(batch_output) + return batch_output + + +def main(config): + rec_predictor = RecPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + batch_imgs = [] + batch_names = [] + cnt = 0 + for idx, img_path in enumerate(image_list): + img = cv2.imread(img_path) + if img is None: + logger.warning( + "Image file failed to read and has been skipped. The path: {}". + format(img_path)) + else: + img = img[:, :, ::-1] + batch_imgs.append(img) + img_name = os.path.basename(img_path) + batch_names.append(img_name) + cnt += 1 + + if cnt % config["Global"]["batch_size"] == 0 or (idx + 1) == len(image_list): + if len(batch_imgs) == 0: + continue + + batch_results = rec_predictor.predict(batch_imgs) + for number, result_dict in enumerate(batch_results): + filename = batch_names[number] + print("{}:\t{}".format(filename, result_dict)) + batch_imgs = [] + batch_names = [] + + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_system.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_system.py new file mode 100644 index 000000000..fb2d66a53 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/predict_system.py @@ -0,0 +1,145 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import copy +import cv2 +import numpy as np +import faiss +import pickle + +from python.predict_rec import RecPredictor +from python.predict_det import DetPredictor + +from utils import logger +from utils import config +from utils.get_image_list import get_image_list +from utils.draw_bbox import draw_bbox_results + + +class SystemPredictor(object): + def __init__(self, config): + + self.config = config + self.rec_predictor = RecPredictor(config) + self.det_predictor = DetPredictor(config) + + assert 'IndexProcess' in config.keys(), "Index config not found ... " + self.return_k = self.config['IndexProcess']['return_k'] + + index_dir = self.config["IndexProcess"]["index_dir"] + assert os.path.exists(os.path.join( + index_dir, "vector.index")), "vector.index not found ..." + assert os.path.exists(os.path.join( + index_dir, "id_map.pkl")), "id_map.pkl not found ... " + + if config['IndexProcess'].get("binary_index", False): + self.Searcher = faiss.read_index_binary( + os.path.join(index_dir, "vector.index")) + else: + self.Searcher = faiss.read_index( + os.path.join(index_dir, "vector.index")) + + with open(os.path.join(index_dir, "id_map.pkl"), "rb") as fd: + self.id_map = pickle.load(fd) + + def append_self(self, results, shape): + results.append({ + "class_id": 0, + "score": 1.0, + "bbox": + np.array([0, 0, shape[1], shape[0]]), # xmin, ymin, xmax, ymax + "label_name": "foreground", + }) + return results + + def nms_to_rec_results(self, results, thresh=0.1): + filtered_results = [] + x1 = np.array([r["bbox"][0] for r in results]).astype("float32") + y1 = np.array([r["bbox"][1] for r in results]).astype("float32") + x2 = np.array([r["bbox"][2] for r in results]).astype("float32") + y2 = np.array([r["bbox"][3] for r in results]).astype("float32") + scores = np.array([r["rec_scores"] for r in results]) + + areas = (x2 - x1 + 1) * (y2 - y1 + 1) + order = scores.argsort()[::-1] + while order.size > 0: + i = order[0] + xx1 = np.maximum(x1[i], x1[order[1:]]) + yy1 = np.maximum(y1[i], y1[order[1:]]) + xx2 = np.minimum(x2[i], x2[order[1:]]) + yy2 = np.minimum(y2[i], y2[order[1:]]) + + w = np.maximum(0.0, xx2 - xx1 + 1) + h = np.maximum(0.0, yy2 - yy1 + 1) + inter = w * h + ovr = inter / (areas[i] + areas[order[1:]] - inter) + inds = np.where(ovr <= thresh)[0] + order = order[inds + 1] + filtered_results.append(results[i]) + + return filtered_results + + def predict(self, img): + output = [] + # st1: get all detection results + results = self.det_predictor.predict(img) + + # st2: add the whole image for recognition to improve recall + results = self.append_self(results, img.shape) + + # st3: recognition process, use score_thres to ensure accuracy + for result in results: + preds = {} + xmin, ymin, xmax, ymax = result["bbox"].astype("int") + crop_img = img[ymin:ymax, xmin:xmax, :].copy() + rec_results = self.rec_predictor.predict(crop_img) + preds["bbox"] = [xmin, ymin, xmax, ymax] + scores, docs = self.Searcher.search(rec_results, self.return_k) + + # just top-1 result will be returned for the final + if scores[0][0] >= self.config["IndexProcess"]["score_thres"]: + preds["rec_docs"] = self.id_map[docs[0][0]].split()[1] + preds["rec_scores"] = scores[0][0] + output.append(preds) + + # st5: nms to the final results to avoid fetching duplicate results + output = self.nms_to_rec_results( + output, self.config["Global"]["rec_nms_thresold"]) + + return output + + +def main(config): + system_predictor = SystemPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = system_predictor.predict(img) + print(image_file) + draw_bbox_results(img, output, image_file) + print(output) + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/preprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/preprocess.py new file mode 100644 index 000000000..c4b6bca30 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/build_gallery/preprocess.py @@ -0,0 +1,337 @@ +""" +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +from functools import partial +import six +import math +import random +import cv2 +import numpy as np +import importlib +from PIL import Image + +from det_preprocess import DetNormalizeImage, DetPadStride, DetPermute, DetResize + + +def create_operators(params): + """ + create operators based on the config + + Args: + params(list): a dict list, used to create some operators + """ + assert isinstance(params, list), ('operator config should be a list') + mod = importlib.import_module(__name__) + ops = [] + for operator in params: + assert isinstance(operator, + dict) and len(operator) == 1, "yaml format error" + op_name = list(operator)[0] + param = {} if operator[op_name] is None else operator[op_name] + op = getattr(mod, op_name)(**param) + ops.append(op) + + return ops + + +class UnifiedResize(object): + def __init__(self, interpolation=None, backend="cv2"): + _cv2_interp_from_str = { + 'nearest': cv2.INTER_NEAREST, + 'bilinear': cv2.INTER_LINEAR, + 'area': cv2.INTER_AREA, + 'bicubic': cv2.INTER_CUBIC, + 'lanczos': cv2.INTER_LANCZOS4 + } + _pil_interp_from_str = { + 'nearest': Image.NEAREST, + 'bilinear': Image.BILINEAR, + 'bicubic': Image.BICUBIC, + 'box': Image.BOX, + 'lanczos': Image.LANCZOS, + 'hamming': Image.HAMMING + } + + def _pil_resize(src, size, resample): + pil_img = Image.fromarray(src) + pil_img = pil_img.resize(size, resample) + return np.asarray(pil_img) + + if backend.lower() == "cv2": + if isinstance(interpolation, str): + interpolation = _cv2_interp_from_str[interpolation.lower()] + # compatible with opencv < version 4.4.0 + elif interpolation is None: + interpolation = cv2.INTER_LINEAR + self.resize_func = partial(cv2.resize, interpolation=interpolation) + elif backend.lower() == "pil": + if isinstance(interpolation, str): + interpolation = _pil_interp_from_str[interpolation.lower()] + self.resize_func = partial(_pil_resize, resample=interpolation) + else: + logger.warning( + f"The backend of Resize only support \"cv2\" or \"PIL\". \"f{backend}\" is unavailable. Use \"cv2\" instead." + ) + self.resize_func = cv2.resize + + def __call__(self, src, size): + return self.resize_func(src, size) + + +class OperatorParamError(ValueError): + """ OperatorParamError + """ + pass + + +class DecodeImage(object): + """ decode image """ + + def __init__(self, to_rgb=True, to_np=False, channel_first=False): + self.to_rgb = to_rgb + self.to_np = to_np # to numpy + self.channel_first = channel_first # only enabled when to_np is True + + def __call__(self, img): + if six.PY2: + assert type(img) is str and len( + img) > 0, "invalid input 'img' in DecodeImage" + else: + assert type(img) is bytes and len( + img) > 0, "invalid input 'img' in DecodeImage" + data = np.frombuffer(img, dtype='uint8') + img = cv2.imdecode(data, 1) + if self.to_rgb: + assert img.shape[2] == 3, 'invalid shape of image[%s]' % ( + img.shape) + img = img[:, :, ::-1] + + if self.channel_first: + img = img.transpose((2, 0, 1)) + + return img + + +class ResizeImage(object): + """ resize image """ + + def __init__(self, + size=None, + resize_short=None, + interpolation=None, + backend="cv2"): + if resize_short is not None and resize_short > 0: + self.resize_short = resize_short + self.w = None + self.h = None + elif size is not None: + self.resize_short = None + self.w = size if type(size) is int else size[0] + self.h = size if type(size) is int else size[1] + else: + raise OperatorParamError("invalid params for ReisizeImage for '\ + 'both 'size' and 'resize_short' are None") + + self._resize_func = UnifiedResize( + interpolation=interpolation, backend=backend) + + def __call__(self, img): + img_h, img_w = img.shape[:2] + if self.resize_short is not None: + percent = float(self.resize_short) / min(img_w, img_h) + w = int(round(img_w * percent)) + h = int(round(img_h * percent)) + else: + w = self.w + h = self.h + return self._resize_func(img, (w, h)) + + +class CropImage(object): + """ crop image """ + + def __init__(self, size): + if type(size) is int: + self.size = (size, size) + else: + self.size = size # (h, w) + + def __call__(self, img): + w, h = self.size + img_h, img_w = img.shape[:2] + + if img_h < h or img_w < w: + raise Exception( + f"The size({h}, {w}) of CropImage must be greater than size({img_h}, {img_w}) of image. Please check image original size and size of ResizeImage if used." + ) + + w_start = (img_w - w) // 2 + h_start = (img_h - h) // 2 + + w_end = w_start + w + h_end = h_start + h + return img[h_start:h_end, w_start:w_end, :] + + +class RandCropImage(object): + """ random crop image """ + + def __init__(self, + size, + scale=None, + ratio=None, + interpolation=None, + backend="cv2"): + if type(size) is int: + self.size = (size, size) # (h, w) + else: + self.size = size + + self.scale = [0.08, 1.0] if scale is None else scale + self.ratio = [3. / 4., 4. / 3.] if ratio is None else ratio + + self._resize_func = UnifiedResize( + interpolation=interpolation, backend=backend) + + def __call__(self, img): + size = self.size + scale = self.scale + ratio = self.ratio + + aspect_ratio = math.sqrt(random.uniform(*ratio)) + w = 1. * aspect_ratio + h = 1. / aspect_ratio + + img_h, img_w = img.shape[:2] + + bound = min((float(img_w) / img_h) / (w**2), + (float(img_h) / img_w) / (h**2)) + scale_max = min(scale[1], bound) + scale_min = min(scale[0], bound) + + target_area = img_w * img_h * random.uniform(scale_min, scale_max) + target_size = math.sqrt(target_area) + w = int(target_size * w) + h = int(target_size * h) + + i = random.randint(0, img_w - w) + j = random.randint(0, img_h - h) + + img = img[j:j + h, i:i + w, :] + + return self._resize_func(img, size) + + +class RandFlipImage(object): + """ random flip image + flip_code: + 1: Flipped Horizontally + 0: Flipped Vertically + -1: Flipped Horizontally & Vertically + """ + + def __init__(self, flip_code=1): + assert flip_code in [-1, 0, 1 + ], "flip_code should be a value in [-1, 0, 1]" + self.flip_code = flip_code + + def __call__(self, img): + if random.randint(0, 1) == 1: + return cv2.flip(img, self.flip_code) + else: + return img + + +class AutoAugment(object): + def __init__(self): + self.policy = ImageNetPolicy() + + def __call__(self, img): + from PIL import Image + img = np.ascontiguousarray(img) + img = Image.fromarray(img) + img = self.policy(img) + img = np.asarray(img) + + +class NormalizeImage(object): + """ normalize image such as substract mean, divide std + """ + + def __init__(self, + scale=None, + mean=None, + std=None, + order='chw', + output_fp16=False, + channel_num=3): + if isinstance(scale, str): + scale = eval(scale) + assert channel_num in [ + 3, 4 + ], "channel number of input image should be set to 3 or 4." + self.channel_num = channel_num + self.output_dtype = 'float16' if output_fp16 else 'float32' + self.scale = np.float32(scale if scale is not None else 1.0 / 255.0) + self.order = order + mean = mean if mean is not None else [0.485, 0.456, 0.406] + std = std if std is not None else [0.229, 0.224, 0.225] + + shape = (3, 1, 1) if self.order == 'chw' else (1, 1, 3) + self.mean = np.array(mean).reshape(shape).astype('float32') + self.std = np.array(std).reshape(shape).astype('float32') + + def __call__(self, img): + from PIL import Image + if isinstance(img, Image.Image): + img = np.array(img) + + assert isinstance(img, + np.ndarray), "invalid input 'img' in NormalizeImage" + + img = (img.astype('float32') * self.scale - self.mean) / self.std + + if self.channel_num == 4: + img_h = img.shape[1] if self.order == 'chw' else img.shape[0] + img_w = img.shape[2] if self.order == 'chw' else img.shape[1] + pad_zeros = np.zeros( + (1, img_h, img_w)) if self.order == 'chw' else np.zeros( + (img_h, img_w, 1)) + img = (np.concatenate( + (img, pad_zeros), axis=0) + if self.order == 'chw' else np.concatenate( + (img, pad_zeros), axis=2)) + return img.astype(self.output_dtype) + + +class ToCHWImage(object): + """ convert hwc image to chw image + """ + + def __init__(self): + pass + + def __call__(self, img): + from PIL import Image + if isinstance(img, Image.Image): + img = np.array(img) + + return img.transpose((2, 0, 1)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/.recognition_web_service_onlyrec.py.swp b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/.recognition_web_service_onlyrec.py.swp new file mode 100644 index 000000000..3fbc33b04 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/.recognition_web_service_onlyrec.py.swp differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log new file mode 100644 index 000000000..7cbcaf186 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log @@ -0,0 +1,2252 @@ +WARNING 2022-02-17 11:39:15,975 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:39:15,995 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:39:15,995 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:39:15,995 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:39:15,995 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:39:15,995 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:39:15,995 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:39:15,995 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:39:16,008 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:39:16,009 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:39:16,009 [dag.py:662] rec +INFO 2022-02-17 11:39:16,009 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:39:16,009 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:39:16,009 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:39:16,009 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:39:16,009 [dag.py:686] - rec +INFO 2022-02-17 11:39:16,009 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:39:16,009 [dag.py:684] [rec] +INFO 2022-02-17 11:39:16,009 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:39:16,075 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:39:16,084 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:39:16,085 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:39:16,089 [dag.py:832] [DAG] start +INFO 2022-02-17 11:39:16,090 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:39:16,092 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:39:16,099 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:39:16,099 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:39:16,099 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:39:17,129 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:39:19,011 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:39:25,739 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069165.7397573 +INFO 2022-02-17 11:39:25,740 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069165.7409468 +INFO 2022-02-17 11:39:25,741 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +ERROR 2022-02-17 11:39:28,512 [operator.py:1109] (data_id=0 log_id=0) [rec|0] Failed to postprocess: postprocess() takes 4 positional arguments but 5 were given +Traceback (most recent call last): + File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1105, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() takes 4 positional arguments but 5 were given +ERROR 2022-02-17 11:39:28,515 [dag.py:410] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [rec|0] Failed to postprocess: postprocess() takes 4 positional arguments but 5 were given +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:40:11,401 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:40:11,401 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:40:11,401 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:40:11,401 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:40:11,402 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:40:11,402 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:40:11,402 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:40:11,414 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:40:11,414 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:40:11,415 [dag.py:662] rec +INFO 2022-02-17 11:40:11,415 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:40:11,415 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:40:11,415 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:40:11,415 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:40:11,415 [dag.py:686] - rec +INFO 2022-02-17 11:40:11,415 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:40:11,415 [dag.py:684] [rec] +INFO 2022-02-17 11:40:11,415 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:40:11,430 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:40:11,439 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:40:11,440 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:40:11,468 [dag.py:832] [DAG] start +INFO 2022-02-17 11:40:11,469 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:40:11,471 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:40:11,479 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:40:11,480 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:40:11,480 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:40:12,540 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:40:14,355 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:40:26,642 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069226.6426268 +INFO 2022-02-17 11:40:26,643 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069226.643837 +INFO 2022-02-17 11:40:26,644 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 11:40:29,377 [dag.py:405] (data_id=0 log_id=0) Succ predict +INFO 2022-02-17 11:41:41,337 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069301.3372242 +INFO 2022-02-17 11:41:41,337 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069301.3376215 +INFO 2022-02-17 11:41:41,337 [dag.py:369] (data_id=1 log_id=0) Succ Generate ID +INFO 2022-02-17 11:41:41,351 [dag.py:405] (data_id=1 log_id=0) Succ predict +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:43:51,578 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:43:51,579 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:43:51,579 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:43:51,579 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:43:51,579 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:43:51,579 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:43:51,579 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:43:51,591 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:43:51,592 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:43:51,592 [dag.py:662] rec +INFO 2022-02-17 11:43:51,592 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:43:51,592 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:43:51,592 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:43:51,592 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:43:51,593 [dag.py:686] - rec +INFO 2022-02-17 11:43:51,593 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:43:51,593 [dag.py:684] [rec] +INFO 2022-02-17 11:43:51,593 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:43:51,608 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:43:51,617 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:43:51,618 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:43:51,622 [dag.py:832] [DAG] start +INFO 2022-02-17 11:43:51,623 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:43:51,625 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:43:51,632 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:43:51,632 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:43:51,632 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:43:52,682 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:43:54,499 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:44:01,419 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069441.419735 +INFO 2022-02-17 11:44:01,421 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069441.4209738 +INFO 2022-02-17 11:44:01,421 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 11:44:04,142 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:45:00,983 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:45:00,983 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:45:00,984 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:45:00,984 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:45:00,984 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:45:00,984 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:45:00,984 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:45:00,996 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:45:00,997 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:45:00,997 [dag.py:662] rec +INFO 2022-02-17 11:45:00,997 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:45:00,997 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:45:00,997 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:45:00,997 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:45:00,997 [dag.py:686] - rec +INFO 2022-02-17 11:45:00,997 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:45:00,997 [dag.py:684] [rec] +INFO 2022-02-17 11:45:00,997 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:45:01,012 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:45:01,022 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:45:01,022 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:45:01,026 [dag.py:832] [DAG] start +INFO 2022-02-17 11:45:01,027 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:45:01,029 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:45:01,036 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:45:01,037 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:45:01,037 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:45:02,091 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:45:03,902 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:45:06,447 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069506.4474986 +INFO 2022-02-17 11:45:06,448 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069506.4486632 +INFO 2022-02-17 11:45:06,449 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 11:45:09,237 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:49:47,978 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:49:47,979 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:49:47,979 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:49:47,979 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:49:47,979 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:49:47,979 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:49:47,979 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:49:47,991 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:49:47,992 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:49:47,992 [dag.py:662] rec +INFO 2022-02-17 11:49:47,992 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:49:47,992 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:49:47,992 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:49:47,992 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:49:47,992 [dag.py:686] - rec +INFO 2022-02-17 11:49:47,992 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:49:47,992 [dag.py:684] [rec] +INFO 2022-02-17 11:49:47,992 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:49:48,007 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:49:48,017 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:49:48,017 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:49:48,021 [dag.py:832] [DAG] start +INFO 2022-02-17 11:49:48,022 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:49:48,024 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:49:48,030 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:49:48,030 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:49:48,030 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:49:49,065 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:49:50,923 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:49:51,725 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069791.7250278 +INFO 2022-02-17 11:49:51,726 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069791.7262204 +INFO 2022-02-17 11:49:51,726 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +WARNING 2022-02-17 11:50:16,675 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:50:16,693 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:50:16,694 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:50:16,694 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:50:16,694 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:50:16,694 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:50:16,694 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:50:16,694 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:50:16,706 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:50:16,707 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:50:16,707 [dag.py:662] rec +INFO 2022-02-17 11:50:16,707 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:50:16,707 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:50:16,707 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:50:16,707 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:50:16,707 [dag.py:686] - rec +INFO 2022-02-17 11:50:16,707 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:50:16,707 [dag.py:684] [rec] +INFO 2022-02-17 11:50:16,707 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:50:16,783 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:50:16,792 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:50:16,793 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:50:16,797 [dag.py:832] [DAG] start +INFO 2022-02-17 11:50:16,798 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:50:16,800 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:50:16,807 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:50:16,807 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:50:16,807 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:50:17,851 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:50:19,691 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:50:25,635 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069825.635539 +INFO 2022-02-17 11:50:25,636 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069825.6367557 +INFO 2022-02-17 11:50:25,637 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 11:51:35,002 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 11:51:35,003 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:51:35,003 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 11:51:35,003 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 11:51:35,003 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 11:51:35,003 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 11:51:35,003 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 11:51:35,014 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 11:51:35,015 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 11:51:35,015 [dag.py:662] rec +INFO 2022-02-17 11:51:35,015 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 11:51:35,015 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 11:51:35,015 [dag.py:682] (VIEW 0) +INFO 2022-02-17 11:51:35,015 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 11:51:35,015 [dag.py:686] - rec +INFO 2022-02-17 11:51:35,015 [dag.py:682] (VIEW 1) +INFO 2022-02-17 11:51:35,015 [dag.py:684] [rec] +INFO 2022-02-17 11:51:35,015 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 11:51:35,030 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 11:51:35,040 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 11:51:35,040 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 11:51:35,043 [dag.py:832] [DAG] start +INFO 2022-02-17 11:51:35,044 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 11:51:35,046 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 11:51:35,052 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 11:51:35,053 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 11:51:35,053 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 11:51:36,063 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 11:51:37,907 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 11:51:40,088 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645069900.0879674 +INFO 2022-02-17 11:51:40,089 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645069900.0891907 +INFO 2022-02-17 11:51:40,089 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 11:51:42,801 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:15:19,783 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:15:19,783 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:15:19,783 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:15:19,784 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:15:19,784 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:15:19,784 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:15:19,784 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:15:19,796 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:15:19,797 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:15:19,797 [dag.py:662] rec +INFO 2022-02-17 13:15:19,797 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:15:19,797 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:15:19,797 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:15:19,797 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:15:19,797 [dag.py:686] - rec +INFO 2022-02-17 13:15:19,797 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:15:19,797 [dag.py:684] [rec] +INFO 2022-02-17 13:15:19,797 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:15:19,813 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:15:19,822 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:15:19,823 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:15:19,827 [dag.py:832] [DAG] start +INFO 2022-02-17 13:15:19,828 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:15:19,830 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:15:19,837 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:15:19,837 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:15:19,838 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:15:20,885 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:15:22,721 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:15:27,334 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645074927.3338962 +INFO 2022-02-17 13:15:27,335 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645074927.3350475 +INFO 2022-02-17 13:15:27,335 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:15:30,083 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:24:47,291 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:24:47,303 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:24:47,303 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:24:47,303 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:24:47,303 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:24:47,303 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:24:47,303 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:24:47,303 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:24:47,314 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:24:47,315 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:24:47,315 [dag.py:662] rec +INFO 2022-02-17 13:24:47,315 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:24:47,315 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:24:47,315 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:24:47,315 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:24:47,315 [dag.py:686] - rec +INFO 2022-02-17 13:24:47,315 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:24:47,315 [dag.py:684] [rec] +INFO 2022-02-17 13:24:47,315 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:24:47,330 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:24:47,340 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:24:47,340 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:24:47,344 [dag.py:832] [DAG] start +INFO 2022-02-17 13:24:47,345 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:24:47,347 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:24:47,354 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:24:47,354 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:24:47,354 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:24:48,445 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:24:50,308 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:24:52,415 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645075492.4151294 +INFO 2022-02-17 13:24:52,416 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645075492.416351 +INFO 2022-02-17 13:24:52,416 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:24:55,206 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:25:34,907 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:25:34,980 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:25:34,980 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:25:34,980 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:25:34,981 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:25:34,981 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:25:34,981 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:25:34,981 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:25:34,993 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:25:34,994 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:25:34,994 [dag.py:662] rec +INFO 2022-02-17 13:25:34,994 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:25:34,994 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:25:34,994 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:25:34,994 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:25:34,994 [dag.py:686] - rec +INFO 2022-02-17 13:25:34,994 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:25:34,994 [dag.py:684] [rec] +INFO 2022-02-17 13:25:34,994 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:25:35,010 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:25:35,019 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:25:35,020 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:25:35,024 [dag.py:832] [DAG] start +INFO 2022-02-17 13:25:35,024 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:25:35,026 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:25:35,033 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:25:35,034 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:25:35,034 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:25:36,103 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:25:37,964 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:25:39,558 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645075539.5582185 +INFO 2022-02-17 13:25:39,559 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645075539.5594082 +INFO 2022-02-17 13:25:39,559 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:28:36,002 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:28:36,002 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:28:36,002 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:28:36,002 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:28:36,002 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:28:36,002 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:28:36,002 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:28:36,014 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:28:36,014 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:28:36,015 [dag.py:662] rec +INFO 2022-02-17 13:28:36,015 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:28:36,015 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:28:36,015 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:28:36,015 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:28:36,015 [dag.py:686] - rec +INFO 2022-02-17 13:28:36,015 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:28:36,015 [dag.py:684] [rec] +INFO 2022-02-17 13:28:36,015 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:28:36,030 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:28:36,039 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:28:36,039 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:28:36,043 [dag.py:832] [DAG] start +INFO 2022-02-17 13:28:36,044 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:28:36,046 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:28:36,053 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:28:36,053 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:28:36,053 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:28:37,113 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:28:38,927 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:28:44,185 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645075724.1851497 +INFO 2022-02-17 13:28:44,186 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645075724.186309 +INFO 2022-02-17 13:28:44,186 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:28:46,924 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:33:50,875 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:33:50,875 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:33:50,887 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:33:50,887 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:33:50,887 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:33:50,887 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:33:50,887 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:33:50,887 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:33:50,887 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:33:50,901 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:33:50,902 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:33:50,902 [dag.py:662] rec +INFO 2022-02-17 13:33:50,902 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:33:50,902 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:33:50,902 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:33:50,902 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:33:50,902 [dag.py:686] - rec +INFO 2022-02-17 13:33:50,902 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:33:50,902 [dag.py:684] [rec] +INFO 2022-02-17 13:33:50,902 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:33:50,918 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:33:50,977 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:33:50,977 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:33:50,982 [dag.py:832] [DAG] start +INFO 2022-02-17 13:33:50,983 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:33:50,985 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:33:50,991 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:33:50,992 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:33:50,992 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:33:52,067 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:33:53,924 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:34:04,445 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645076044.4450066 +INFO 2022-02-17 13:34:04,446 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645076044.4461415 +INFO 2022-02-17 13:34:04,446 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:34:07,203 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:38:14,876 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:38:14,971 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:38:14,971 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:38:14,971 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:38:14,971 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:38:14,971 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:38:14,971 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:38:14,972 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:38:14,985 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:38:14,986 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:38:14,986 [dag.py:662] rec +INFO 2022-02-17 13:38:14,986 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:38:14,986 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:38:14,986 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:38:14,986 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:38:14,986 [dag.py:686] - rec +INFO 2022-02-17 13:38:14,986 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:38:14,986 [dag.py:684] [rec] +INFO 2022-02-17 13:38:14,986 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:38:15,002 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:38:15,011 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:38:15,012 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:38:15,016 [dag.py:832] [DAG] start +INFO 2022-02-17 13:38:15,017 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:38:15,019 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:38:15,026 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:38:15,026 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:38:15,026 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:38:16,054 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:38:17,886 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:38:25,102 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645076305.1018336 +INFO 2022-02-17 13:38:25,103 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645076305.1030562 +INFO 2022-02-17 13:38:25,103 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:38:27,823 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:41:58,835 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:41:58,837 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:41:58,878 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:41:58,879 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:41:58,879 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:41:58,879 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:41:58,879 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:41:58,879 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:41:58,879 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:41:58,891 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:41:58,892 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:41:58,892 [dag.py:662] rec +INFO 2022-02-17 13:41:58,892 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:41:58,892 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:41:58,892 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:41:58,892 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:41:58,892 [dag.py:686] - rec +INFO 2022-02-17 13:41:58,892 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:41:58,892 [dag.py:684] [rec] +INFO 2022-02-17 13:41:58,892 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:41:58,908 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:41:58,918 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:41:58,918 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:41:58,922 [dag.py:832] [DAG] start +INFO 2022-02-17 13:41:58,923 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:41:58,925 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:41:58,931 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:41:58,932 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:41:58,932 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:41:59,935 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:42:02,434 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:42:28,077 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645076548.076952 +INFO 2022-02-17 13:42:28,078 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645076548.078083 +INFO 2022-02-17 13:42:28,078 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:42:30,814 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:45:45,888 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:45:45,888 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:45:45,888 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:45:45,888 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:45:45,888 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:45:45,889 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:45:45,889 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:45:45,900 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:45:45,901 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:45:45,901 [dag.py:662] rec +INFO 2022-02-17 13:45:45,901 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:45:45,901 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:45:45,901 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:45:45,901 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:45:45,901 [dag.py:686] - rec +INFO 2022-02-17 13:45:45,901 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:45:45,901 [dag.py:684] [rec] +INFO 2022-02-17 13:45:45,901 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:45:45,916 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:45:45,926 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:45:45,926 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:45:45,930 [dag.py:832] [DAG] start +INFO 2022-02-17 13:45:45,931 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:45:45,933 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:45:45,939 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:45:45,940 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:45:45,940 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:45:46,919 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:45:49,242 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645076749.2424037 +INFO 2022-02-17 13:45:49,243 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645076749.2436109 +INFO 2022-02-17 13:45:49,244 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:45:49,645 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:45:52,450 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 13:48:08,096 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:48:08,170 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:48:08,170 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 13:48:08,181 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 13:48:08,182 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:48:08,182 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 13:48:08,182 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 13:48:08,182 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 13:48:08,182 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 13:48:08,182 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 13:48:08,195 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 13:48:08,196 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 13:48:08,196 [dag.py:662] rec +INFO 2022-02-17 13:48:08,196 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 13:48:08,196 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 13:48:08,196 [dag.py:682] (VIEW 0) +INFO 2022-02-17 13:48:08,196 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 13:48:08,196 [dag.py:686] - rec +INFO 2022-02-17 13:48:08,196 [dag.py:682] (VIEW 1) +INFO 2022-02-17 13:48:08,196 [dag.py:684] [rec] +INFO 2022-02-17 13:48:08,196 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 13:48:08,213 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 13:48:08,222 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 13:48:08,223 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 13:48:08,227 [dag.py:832] [DAG] start +INFO 2022-02-17 13:48:08,228 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 13:48:08,230 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 13:48:08,275 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 13:48:08,276 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 13:48:08,276 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 13:48:09,338 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 13:48:11,286 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645076891.2866788 +INFO 2022-02-17 13:48:11,287 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645076891.2878144 +INFO 2022-02-17 13:48:11,288 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 13:48:12,468 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 13:48:15,399 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:00:34,472 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:00:34,483 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 14:00:34,483 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:00:34,483 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 14:00:34,483 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 14:00:34,483 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 14:00:34,483 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 14:00:34,483 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 14:00:34,495 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 14:00:34,496 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 14:00:34,496 [dag.py:662] rec +INFO 2022-02-17 14:00:34,496 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 14:00:34,496 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 14:00:34,496 [dag.py:682] (VIEW 0) +INFO 2022-02-17 14:00:34,496 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 14:00:34,497 [dag.py:686] - rec +INFO 2022-02-17 14:00:34,497 [dag.py:682] (VIEW 1) +INFO 2022-02-17 14:00:34,497 [dag.py:684] [rec] +INFO 2022-02-17 14:00:34,497 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 14:00:34,512 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 14:00:34,522 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 14:00:34,522 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 14:00:34,526 [dag.py:832] [DAG] start +INFO 2022-02-17 14:00:34,527 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 14:00:34,529 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 14:00:34,536 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:00:34,536 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 14:00:34,536 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 14:00:35,542 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 14:00:38,036 [operator.py:1317] [rec|0] Succ init +WARNING 2022-02-17 14:00:53,138 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:00:56,641 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645077656.6412687 +INFO 2022-02-17 14:00:56,642 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645077656.642419 +INFO 2022-02-17 14:00:56,642 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 14:00:59,372 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 14:09:35,769 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:09:35,783 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 2, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 14:09:35,783 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:09:35,783 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 14:09:35,783 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 14:09:35,783 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":2, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 14:09:35,783 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 14:09:35,783 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 14:09:35,797 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 14:09:35,798 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 14:09:35,798 [dag.py:662] rec +INFO 2022-02-17 14:09:35,798 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 14:09:35,798 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 14:09:35,798 [dag.py:682] (VIEW 0) +INFO 2022-02-17 14:09:35,798 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 14:09:35,798 [dag.py:686] - rec +INFO 2022-02-17 14:09:35,798 [dag.py:682] (VIEW 1) +INFO 2022-02-17 14:09:35,799 [dag.py:684] [rec] +INFO 2022-02-17 14:09:35,799 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 14:09:35,816 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 14:09:35,826 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 14:09:35,827 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 14:09:35,832 [dag.py:832] [DAG] start +INFO 2022-02-17 14:09:35,833 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 14:09:35,870 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 14:09:35,876 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:09:35,877 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 14:09:35,877 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 14:09:36,950 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 14:09:39,525 [operator.py:1317] [rec|0] Succ init +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:10:28,773 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:10:28,784 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 1, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 14:10:28,784 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:10:28,784 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 14:10:28,784 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 14:10:28,784 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":1, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 14:10:28,784 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 14:10:28,784 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 14:10:28,796 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 14:10:28,797 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 14:10:28,797 [dag.py:662] rec +INFO 2022-02-17 14:10:28,797 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 14:10:28,797 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 14:10:28,797 [dag.py:682] (VIEW 0) +INFO 2022-02-17 14:10:28,797 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 14:10:28,797 [dag.py:686] - rec +INFO 2022-02-17 14:10:28,798 [dag.py:682] (VIEW 1) +INFO 2022-02-17 14:10:28,798 [dag.py:684] [rec] +INFO 2022-02-17 14:10:28,798 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 14:10:28,813 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 14:10:28,822 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 14:10:28,823 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 14:10:28,827 [dag.py:832] [DAG] start +INFO 2022-02-17 14:10:28,827 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 14:10:28,870 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 14:10:28,873 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:10:28,873 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 14:10:28,873 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 14:10:29,826 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:False, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 14:10:32,335 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 14:10:33,070 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645078233.070603 +INFO 2022-02-17 14:10:33,071 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645078233.0718045 +INFO 2022-02-17 14:10:33,072 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 14:10:35,794 [dag.py:405] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-17 14:14:29,769 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:14:29,783 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 1, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 14:14:29,783 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:14:29,783 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 14:14:29,783 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 14:14:29,783 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":1, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 14:14:29,783 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 14:14:29,783 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 14:14:29,795 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 14:14:29,796 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 14:14:29,796 [dag.py:662] rec +INFO 2022-02-17 14:14:29,796 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 14:14:29,796 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 14:14:29,796 [dag.py:682] (VIEW 0) +INFO 2022-02-17 14:14:29,796 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 14:14:29,796 [dag.py:686] - rec +INFO 2022-02-17 14:14:29,796 [dag.py:682] (VIEW 1) +INFO 2022-02-17 14:14:29,796 [dag.py:684] [rec] +INFO 2022-02-17 14:14:29,796 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 14:14:29,812 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 14:14:29,822 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 14:14:29,823 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 14:14:29,827 [dag.py:832] [DAG] start +INFO 2022-02-17 14:14:29,827 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 14:14:29,829 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 14:14:29,837 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:14:29,837 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 14:14:29,837 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 14:14:30,832 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:False, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 14:14:33,327 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 14:14:39,181 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645078479.1810365 +INFO 2022-02-17 14:14:39,182 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645078479.1821644 +INFO 2022-02-17 14:14:39,182 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 14:14:41,917 [dag.py:405] (data_id=0 log_id=0) Succ predict +INFO 2022-02-17 14:26:17,673 [loader.py:54] Loading faiss with AVX2 support. +INFO 2022-02-17 14:26:17,673 [loader.py:58] Could not load library with AVX2 support due to: +ModuleNotFoundError("No module named 'faiss.swigfaiss_avx2'") +INFO 2022-02-17 14:26:17,673 [loader.py:64] Loading faiss. +INFO 2022-02-17 14:26:17,693 [loader.py:66] Successfully loaded faiss. +WARNING 2022-02-17 14:26:17,697 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:26:17,697 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-17 14:26:17,711 [operator.py:181] local_service_conf: {'model_config': './general_PPLCNet_x2_5_lite_v1.0_serving', 'device_type': 1, 'devices': '0', 'client_type': 'local_predictor', 'fetch_list': ['feature'], 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-17 14:26:17,711 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['feature'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:26:17,712 [operator.py:285] rec + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['feature'] + client_config: ./general_PPLCNet_x2_5_lite_v1.0_serving/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-17 14:26:17,712 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-17 14:26:17,712 [pipeline_server.py:218] +{ + "worker_num":1, + "http_port":9315, + "rpc_port":9314, + "dag":{ + "is_thread_op":false, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "tracer":{ + "interval_s":-1 + }, + "channel_recv_frist_arrive":false + }, + "op":{ + "rec":{ + "concurrency":1, + "local_service_conf":{ + "model_config":"./general_PPLCNet_x2_5_lite_v1.0_serving", + "device_type":1, + "devices":"0", + "client_type":"local_predictor", + "fetch_list":[ + "feature" + ], + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "build_dag_each_worker":false +} +INFO 2022-02-17 14:26:17,712 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-17 14:26:17,712 [operator.py:308] Op(rec) use local rpc service at port: [] +INFO 2022-02-17 14:26:17,726 [dag.py:496] [DAG] Succ init +INFO 2022-02-17 14:26:17,727 [dag.py:659] ================= USED OP ================= +INFO 2022-02-17 14:26:17,727 [dag.py:662] rec +INFO 2022-02-17 14:26:17,727 [dag.py:663] ------------------------------------------- +INFO 2022-02-17 14:26:17,727 [dag.py:680] ================== DAG ==================== +INFO 2022-02-17 14:26:17,727 [dag.py:682] (VIEW 0) +INFO 2022-02-17 14:26:17,727 [dag.py:684] [@DAGExecutor] +INFO 2022-02-17 14:26:17,727 [dag.py:686] - rec +INFO 2022-02-17 14:26:17,727 [dag.py:682] (VIEW 1) +INFO 2022-02-17 14:26:17,727 [dag.py:684] [rec] +INFO 2022-02-17 14:26:17,727 [dag.py:687] ------------------------------------------- +INFO 2022-02-17 14:26:17,743 [dag.py:730] op:rec add input channel. +INFO 2022-02-17 14:26:17,757 [dag.py:759] last op:rec add output channel +INFO 2022-02-17 14:26:17,758 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-17 14:26:17,764 [dag.py:832] [DAG] start +INFO 2022-02-17 14:26:17,765 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-17 14:26:17,767 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-17 14:26:17,777 [local_service_handler.py:172] Models(./general_PPLCNet_x2_5_lite_v1.0_serving) will be launched by device gpu. use_gpu:True, use_trt:False, use_lite:False, use_xpu:False, device_type:1, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-17 14:26:17,777 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-17 14:26:17,777 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-17 14:26:18,830 [local_predict.py:153] LocalPredictor load_model_config params: model_path:./general_PPLCNet_x2_5_lite_v1.0_serving, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:False, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-17 14:26:20,559 [operator.py:1317] [rec|0] Succ init +INFO 2022-02-17 14:26:37,796 [pipeline_server.py:56] (log_id=0) inference request name:recognition self.name:recognition time:1645079197.7964764 +INFO 2022-02-17 14:26:37,797 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:recognition, method:prediction, time:1645079197.7977042 +INFO 2022-02-17 14:26:37,798 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-17 14:26:39,784 [dag.py:405] (data_id=0 log_id=0) Succ predict diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log.wf b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log.wf new file mode 100644 index 000000000..3c85322cf --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.log.wf @@ -0,0 +1,466 @@ +WARNING 2022-02-17 11:39:15,975 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:39:15,976 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:39:15,977 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2022-02-17 11:39:28,512 [operator.py:1109] (data_id=0 log_id=0) [rec|0] Failed to postprocess: postprocess() takes 4 positional arguments but 5 were given +Traceback (most recent call last): + File "/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1105, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() takes 4 positional arguments but 5 were given +ERROR 2022-02-17 11:39:28,515 [dag.py:410] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [rec|0] Failed to postprocess: postprocess() takes 4 positional arguments but 5 were given +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:40:11,389 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:40:11,390 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:43:51,478 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:43:51,479 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:45:00,971 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:45:00,972 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:49:47,902 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:49:47,903 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 11:50:16,675 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:50:16,676 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:50:16,677 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 11:51:34,991 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 11:51:34,992 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:15:00,101 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:15:00,102 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:15:19,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:15:19,772 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:24:47,291 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:24:47,292 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:25:34,907 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:25:34,968 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:25:34,969 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:28:35,990 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:28:35,991 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:33:50,873 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:33:50,874 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:33:50,875 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:33:50,875 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:38:14,874 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:38:14,875 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:38:14,876 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:41:58,835 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:41:58,836 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:41:58,837 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:45:45,876 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:45:45,877 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 13:48:08,096 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 13:48:08,168 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 13:48:08,169 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 13:48:08,170 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 13:48:08,170 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:00:34,470 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:00:34,471 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:00:34,472 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:00:53,138 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:00:53,139 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:00:53,140 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:09:35,769 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:09:35,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:09:35,771 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:10:28,771 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:10:28,772 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:10:28,773 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:14:29,769 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:14:29,770 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:14:29,771 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-17 14:26:17,697 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-17 14:26:17,697 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] tracer not set, use default: {} +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] interval_s not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-17 14:26:17,698 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.tracer b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/PipelineServingLogs/pipeline.tracer new file mode 100644 index 000000000..e69de29bb diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/ProcessInfo.json b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/ProcessInfo.json new file mode 100644 index 000000000..4e88c39e5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/ProcessInfo.json @@ -0,0 +1 @@ +[{"pid": 827, "port": [9314, 9315], "model": "pipline", "start_time": 1645079177.7113242}] \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/config_onlyrec.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/config_onlyrec.yml new file mode 100644 index 000000000..c04957537 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/config_onlyrec.yml @@ -0,0 +1,33 @@ +#worker_num, 鏈澶у苟鍙戞暟銆傚綋build_dag_each_worker=True鏃, 妗嗘灦浼氬垱寤簑orker_num涓繘绋嬶紝姣忎釜杩涚▼鍐呮瀯寤篻rpcSever鍜孌AG +##褰揵uild_dag_each_worker=False鏃讹紝妗嗘灦浼氳缃富绾跨▼grpc绾跨▼姹犵殑max_workers=worker_num +worker_num: 1 + +#http绔彛, rpc_port鍜宧ttp_port涓嶅厑璁稿悓鏃朵负绌恒傚綋rpc_port鍙敤涓攈ttp_port涓虹┖鏃讹紝涓嶈嚜鍔ㄧ敓鎴恏ttp_port +http_port: 9315 +rpc_port: 9314 + +dag: + #op璧勬簮绫诲瀷, True, 涓虹嚎绋嬫ā鍨嬶紱False锛屼负杩涚▼妯″瀷 + is_thread_op: False +op: + rec: + #骞跺彂鏁帮紝is_thread_op=True鏃讹紝涓虹嚎绋嬪苟鍙戯紱鍚﹀垯涓鸿繘绋嬪苟鍙 + concurrency: 1 + + #褰搊p閰嶇疆娌℃湁server_endpoints鏃讹紝浠巐ocal_service_conf璇诲彇鏈湴鏈嶅姟閰嶇疆 + local_service_conf: + + #uci妯″瀷璺緞 + model_config: ./general_PPLCNet_x2_5_lite_v1.0_serving + + #璁＄畻纭欢绫诲瀷: 绌虹己鏃剁敱devices鍐冲畾(CPU/GPU)锛0=cpu, 1=gpu, 2=tensorRT, 3=arm cpu, 4=kunlun xpu + device_type: 1 + + #璁＄畻纭欢ID锛屽綋devices涓""鎴栦笉鍐欐椂涓篊PU棰勬祴锛涘綋devices涓"0", "0,1,2"鏃朵负GPU棰勬祴锛岃〃绀轰娇鐢ㄧ殑GPU鍗 + devices: "0" # "0,1" + + #client绫诲瀷锛屽寘鎷琤rpc, grpc鍜宭ocal_predictor.local_predictor涓嶅惎鍔⊿erving鏈嶅姟锛岃繘绋嬪唴棰勬祴 + client_type: local_predictor + + #Fetch缁撴灉鍒楄〃锛屼互client_config涓璮etch_var鐨刟lias_name涓哄噯 + fetch_list: ["feature"] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/pipeline_http_client.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/pipeline_http_client.py new file mode 100644 index 000000000..2c1421e29 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/pipeline_http_client.py @@ -0,0 +1,52 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# from paddle_serving_server.pipeline import PipelineClient +import numpy as np +import requests +import json +import cv2 +import base64 +import os +from time import * +import threading + + +def demo(url,data,i): + begin_time = time() + r = requests.post(url=url, data=json.dumps(data)) + end_time = time() + run_time = end_time-begin_time + print ('绾跨▼ %d 鏃堕棿 %f '%(i,run_time)) + print(r.json()) + + +def cv2_to_base64(image): + return base64.b64encode(image).decode('utf8') + +url = "http://127.0.0.1:9315/recognition/prediction" +with open(os.path.join(".", "test.jpg"), 'rb') as file: + image_data1 = file.read() +image = cv2_to_base64(image_data1) + +for i in range(1): + print(i) + data = {"key": ["image"], "value": [image]} + r = requests.post(url=url, data=json.dumps(data)) + print(r.json()) + #t = threading.Thread(target=demo, args=(url,data,i,)) + #t.start() + + + + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/recognition_web_service_onlyrec.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/recognition_web_service_onlyrec.py new file mode 100644 index 000000000..0e5484d23 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/recognition_web_service_onlyrec.py @@ -0,0 +1,220 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from paddle_serving_server.web_service import WebService, Op +import logging +import numpy as np +import sys +import cv2 +from paddle_serving_app.reader import * +import base64 +import os +import faiss +import pickle +import json + + +class DetOp(Op): + def init_op(self): + self.img_preprocess = Sequential([ + #Resize((416, 416)), + Div(255.0), + Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225], False), + Transpose((2, 0, 1)) + #Resize((416, 416)), Transpose((2, 0, 1)) + ]) + + self.img_postprocess = RCNNPostprocess("label_list.txt", "output") + self.threshold = 0.3 + self.max_det_results = 5 + + def Deresize(self, im, im_scale_x, im_scale_y): + #print("99999999999999997777777") + #print(im) + #print(im_scale_x,im_scale_y,cv2.INTER_LINEAR) + + im = cv2.resize( + im, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=2) + #print(im) + return im + + def generate_scale(self, im): + """ + Args: + im (np.ndarray): image (np.ndarray) + Returns: + im_scale_x: the resize ratio of X + im_scale_y: the resize ratio of Y + """ + target_size = [416, 416] + origin_shape = im.shape[:2] + resize_h, resize_w = target_size + im_scale_y = resize_h / float(origin_shape[0]) + im_scale_x = resize_w / float(origin_shape[1]) + return im_scale_y, im_scale_x + + def preprocess(self, input_dicts, data_id, log_id): + (_, input_dict), = input_dicts.items() + imgs = [] + raw_imgs = [] + for key in input_dict.keys(): + data = base64.b64decode(input_dict[key].encode('utf8')) + raw_imgs.append(data) + data = np.fromstring(data, np.uint8) + raw_im = cv2.imdecode(data, cv2.IMREAD_COLOR)[:, :, ::-1] + im_scale_y, im_scale_x = self.generate_scale(raw_im) + raw_im = self.Deresize(raw_im, im_scale_x, im_scale_y) + im = self.img_preprocess(raw_im) + #im = im.transpose((2, 0, 1)).copy() + #print(im) + imgs.append({ + "image": im[np.newaxis, :], + "im_shape": + np.array(list(im.shape[1:])).reshape(-1)[np.newaxis, :], + "scale_factor": + np.array([[im_scale_y, im_scale_x]]).astype('float32'), + }) + self.raw_img = raw_imgs + + feed_dict = { + "image": np.concatenate( + [x["image"] for x in imgs], axis=0), + "im_shape": np.concatenate( + [x["im_shape"] for x in imgs], axis=0), + "scale_factor": np.concatenate( + [x["scale_factor"] for x in imgs], axis=0) + } + #print("feed_dict",feed_dict) + return feed_dict, False, None, "" + + def postprocess(self, input_dicts, fetch_dict, log_id): + boxes = self.img_postprocess(fetch_dict, visualize=False) + #print("boxes",boxes) + boxes.sort(key=lambda x: x["score"], reverse=True) + boxes = filter(lambda x: x["score"] >= self.threshold, + boxes[:self.max_det_results]) + boxes = list(boxes) + for i in range(len(boxes)): + boxes[i]["bbox"][2] += boxes[i]["bbox"][0] - 1 + boxes[i]["bbox"][3] += boxes[i]["bbox"][1] - 1 + result = json.dumps(boxes) + res_dict = {"bbox_result": result, "image": self.raw_img} + return res_dict, None, "" + + +class RecOp(Op): + def init_op(self): + self.seq = Sequential([ + BGR2RGB(), Resize((224, 224)), Div(255), + Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225], + False), Transpose((2, 0, 1)) + ]) + + index_dir = "./index_result/" + assert os.path.exists(os.path.join( + index_dir, "vector.index")), "vector.index not found ..." + assert os.path.exists(os.path.join( + index_dir, "id_map.pkl")), "id_map.pkl not found ... " + + self.searcher = faiss.read_index( + os.path.join(index_dir, "vector.index")) + + with open(os.path.join(index_dir, "id_map.pkl"), "rb") as fd: + self.id_map = pickle.load(fd) + + self.rec_nms_thresold = 0.05 + self.rec_score_thres = 0.5 + self.feature_normalize = True + self.return_k =5 + + def preprocess(self, input_dicts, data_id, log_id): + (_, input_dict), = input_dicts.items() + + raw_img = input_dict["image"] + raw_img = base64.b64decode(raw_img) + data = np.frombuffer(raw_img, np.uint8) + origin_img = cv2.imdecode(data, cv2.IMREAD_COLOR) + #construct batch images for rec + imgs = [] + img = self.seq(origin_img) + imgs.append(img[np.newaxis, :].copy()) + + input_imgs = np.concatenate(imgs, axis=0) + return {"x": input_imgs}, False, None, "" + + def nms_to_rec_results(self, results, thresh=0.1): + #print("results",results) + filtered_results = [] + x1 = np.array([r["bbox"][0] for r in results]).astype("float32") + y1 = np.array([r["bbox"][1] for r in results]).astype("float32") + x2 = np.array([r["bbox"][2] for r in results]).astype("float32") + y2 = np.array([r["bbox"][3] for r in results]).astype("float32") + scores = np.array([r["rec_scores"] for r in results]) + + areas = (x2 - x1 + 1) * (y2 - y1 + 1) + order = scores.argsort()[::-1] + while order.size > 0: + i = order[0] + xx1 = np.maximum(x1[i], x1[order[1:]]) + yy1 = np.maximum(y1[i], y1[order[1:]]) + xx2 = np.minimum(x2[i], x2[order[1:]]) + yy2 = np.minimum(y2[i], y2[order[1:]]) + + w = np.maximum(0.0, xx2 - xx1 + 1) + h = np.maximum(0.0, yy2 - yy1 + 1) + inter = w * h + ovr = inter / (areas[i] + areas[order[1:]] - inter) + inds = np.where(ovr <= thresh)[0] + order = order[inds + 1] + filtered_results.append(results[i]) + return filtered_results + + def postprocess(self, input_dicts, fetch_dict, log_id,data_id = 0): + #print("fetch_dict",fetch_dict) + batch_features = fetch_dict["feature"] + + if self.feature_normalize: + feas_norm = np.sqrt( + np.sum(np.square(batch_features), axis=1, keepdims=True)) + batch_features = np.divide(batch_features, feas_norm) + + scores, docs = self.searcher.search(batch_features, self.return_k) + print(scores) + results = [] + for i in range(scores.shape[0]): + pred = {} + if scores[i][0] >= self.rec_score_thres: + pred["rec_docs"] = self.id_map[docs[i][0]].split()[1] + pred["rec_scores"] = scores[i][0] + results.append(pred) + + #do nms + #results = self.nms_to_rec_results(results, self.rec_nms_thresold) + return {"result": str(results)}, None, "" + + +class RecognitionService(WebService): + def get_pipeline_response(self, read_op): + #det_op = DetOp(name="det", input_ops=[read_op]) + rec_op = RecOp(name="rec", input_ops=[read_op]) + return rec_op + + +product_recog_service = RecognitionService(name="recognition") +product_recog_service.prepare_pipeline_config("config_onlyrec.yml") +product_recog_service.run_service() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/test.jpg b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/test.jpg new file mode 100644 index 000000000..c7c19da9a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/general_PPLCNet_x2_5_lite_v1.0/test.jpg differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/index_label.txt b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/index_label.txt new file mode 100644 index 000000000..336b98473 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/index_label.txt @@ -0,0 +1,144 @@ +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15114.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15010.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15038.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15042.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15111.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15127.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15028.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10036.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15033.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15109.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10033.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15047.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15023.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15068.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15107.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15074.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15035.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15063.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15043.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15015.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15124.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10011.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15100.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15106.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15108.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15050.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15113.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15016.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15037.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15110.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15018.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15030.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15014.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15060.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15069.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15133.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15041.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15141.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15020.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15104.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15022.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10032.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15034.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15053.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15058.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15051.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15139.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15119.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15017.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15009.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15131.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15101.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15054.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15064.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15130.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15011.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15120.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15048.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15032.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15052.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10017.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15123.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15073.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15055.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15062.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10024.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15117.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15125.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15070.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15046.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15049.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15142.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15129.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15029.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10012.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15126.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15045.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15007.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15057.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15105.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15024.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15121.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15036.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15140.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15039.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15102.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15027.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15008.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15025.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15115.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15021.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15132.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15071.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15031.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15026.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15044.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15118.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15116.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15112.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15135.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15061.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15067.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15137.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15013.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10019.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15066.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15072.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10030.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15056.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15059.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15122.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/0-10001.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15012.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/鎽╂墭杞/15103.jpg 鎽╂墭杞 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10528.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10504.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10501.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10195.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10377.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10259.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10285.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10201.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10182.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/浜/0-10522.jpg 浜 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10112.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10015.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10013.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10073.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10020.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10071.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10078.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10055.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10072.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鍏朵粬鏉傜墿/0-10014.jpg 鍏朵粬鏉傜墿 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10549.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10381.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10193.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10290.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10685.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10180.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-11043.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10461.jpg 鑷杞 +/home/aistudio/data/data128448/index_motorcycle/鑷杞/0-10777.jpg 鑷杞 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/make_label.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/make_label.py new file mode 100644 index 000000000..f6a563cd7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/make_label.py @@ -0,0 +1,11 @@ +import os +root_path = "/home/aistudio/data/data128448/index_motorcycle/" +dirs = os.listdir(root_path) +dir_dict = {"person":"浜","motorcycle":"鐢电摱杞/鎽╂墭杞","bicycle":"鑷杞","others":"鍏朵粬"} +with open("index_label.txt","w") as f: + for dir in dirs: + path = root_path + dir + "/" + print(path) + filenames = os.listdir(path) + for filename in filenames: + f.write(path+filename+" "+dir+"\n") \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams new file mode 100644 index 000000000..014e7c221 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams.info b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams.info new file mode 100644 index 000000000..4b645bfe1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdiparams.info differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdmodel b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdmodel new file mode 100644 index 000000000..a444c2daa Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/models/general_PPLCNet_x2_5_lite_v1.0_infer/inference.pdmodel differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log new file mode 100644 index 000000000..3a6db0710 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log @@ -0,0 +1,8074 @@ +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 02:45:16,606 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 02:45:16,607 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 02:45:16,607 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 02:45:16,607 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 02:45:16,607 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18082, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9998, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 02:45:16,607 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 02:45:16,607 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 02:45:16,631 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 02:45:16,632 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 02:45:16,632 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 02:45:16,632 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 02:45:16,677 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 02:45:16,681 [dag.py:816] [DAG] start +INFO 2021-12-29 02:45:16,682 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 02:45:16,688 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 02:45:16,710 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 02:45:16,710 [operator.py:1163] Init cuda env in process 0 +INFO 2021-12-29 02:45:16,710 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 02:45:17,939 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 02:45:19,138 [operator.py:1174] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 02:45:44,185 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 02:45:44,187 [operator.py:1422] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 02:45:44,188 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 02:45:45,873 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 76, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 02:45:45,877 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 03:07:14,512 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 03:07:14,513 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:07:14,513 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 03:07:14,513 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 03:07:14,513 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18082, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9998, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 03:07:14,513 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 03:07:14,513 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 03:07:14,538 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 03:07:14,539 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 03:07:14,539 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 03:07:14,539 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 03:07:14,585 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 03:07:14,589 [dag.py:816] [DAG] start +INFO 2021-12-29 03:07:14,589 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 03:07:14,595 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 03:07:14,617 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:07:14,617 [operator.py:1163] Init cuda env in process 0 +INFO 2021-12-29 03:07:14,618 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 03:07:15,847 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 03:07:17,038 [operator.py:1174] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 03:07:20,880 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 03:07:20,882 [operator.py:1422] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 03:07:20,882 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 03:07:22,696 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:07:22,700 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:10:13,372 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 03:10:13,375 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 03:10:13,375 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:10:13,375 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 03:10:13,375 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 03:10:13,375 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18082, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9998, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 03:10:13,375 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 03:10:13,375 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 03:10:13,393 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 03:10:13,393 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 03:10:13,394 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 03:10:13,394 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 03:10:13,436 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 03:10:13,439 [dag.py:816] [DAG] start +INFO 2021-12-29 03:10:13,440 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 03:10:13,445 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 03:10:13,468 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:10:13,469 [operator.py:1163] Init cuda env in process 0 +INFO 2021-12-29 03:10:13,469 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 03:10:14,628 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 03:10:15,826 [operator.py:1174] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 03:10:19,409 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 03:10:19,411 [operator.py:1422] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 03:10:19,411 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 03:10:21,260 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 78, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:10:21,264 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:11:47,325 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 03:11:47,325 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 03:11:47,325 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:11:47,325 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 03:11:47,325 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 03:11:47,326 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18082, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9998, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 03:11:47,326 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 03:11:47,326 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 03:11:47,348 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 03:11:47,349 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 03:11:47,349 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 03:11:47,349 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 03:11:47,395 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 03:11:47,398 [dag.py:816] [DAG] start +INFO 2021-12-29 03:11:47,399 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 03:11:47,403 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 03:11:47,431 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 03:11:47,432 [operator.py:1163] Init cuda env in process 0 +INFO 2021-12-29 03:11:47,432 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 03:11:48,697 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 03:11:49,910 [operator.py:1174] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 03:11:53,938 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 03:11:53,939 [operator.py:1422] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 03:11:53,940 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 03:11:55,757 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 78, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:11:55,761 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 05:37:04,891 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 05:37:04,892 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:37:04,892 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 05:37:04,892 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 05:37:04,892 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 05:37:04,892 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 05:37:04,892 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 05:37:04,915 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 05:37:04,916 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 05:37:04,916 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 05:37:04,916 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 05:37:04,962 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 05:37:04,967 [dag.py:816] [DAG] start +INFO 2021-12-29 05:37:04,968 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 05:37:04,974 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 05:37:04,992 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:37:04,993 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 05:37:04,993 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 05:37:06,170 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 05:37:07,358 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 05:37:14,620 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 05:37:14,621 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 05:37:14,621 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 05:37:16,537 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:37:16,542 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:40:11,811 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:40:11,811 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 05:40:11,811 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 05:40:11,811 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:40:11,811 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 05:40:11,811 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 05:40:11,812 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 05:40:11,812 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 05:40:11,812 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 05:40:11,837 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 05:40:11,838 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 05:40:11,839 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 05:40:11,839 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 05:40:11,880 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 05:40:11,884 [dag.py:816] [DAG] start +INFO 2021-12-29 05:40:11,885 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 05:40:11,890 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 05:40:11,909 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:40:11,910 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 05:40:11,910 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 05:40:13,297 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 05:40:14,485 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 05:40:16,831 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 05:40:16,832 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 05:40:16,834 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 05:40:18,654 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:40:18,658 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 05:42:11,545 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 05:42:11,546 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:42:11,546 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 05:42:11,546 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 05:42:11,546 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 05:42:11,546 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 05:42:11,546 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 05:42:11,569 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 05:42:11,569 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 05:42:11,570 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 05:42:11,570 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 05:42:11,615 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 05:42:11,618 [dag.py:816] [DAG] start +INFO 2021-12-29 05:42:11,619 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 05:42:11,626 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 05:42:11,650 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 05:42:11,651 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 05:42:11,651 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 05:42:12,841 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 05:42:14,032 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 05:42:17,467 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 05:42:17,469 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 05:42:17,470 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 05:42:19,333 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:42:19,340 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:08:54,357 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:08:54,358 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:08:54,358 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:08:54,358 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:08:54,358 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:08:54,358 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:08:54,358 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:08:54,380 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:08:54,381 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:08:54,381 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:08:54,381 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:08:54,427 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:08:54,431 [dag.py:816] [DAG] start +INFO 2021-12-29 06:08:54,432 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:08:54,438 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:08:54,460 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:08:54,460 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:08:54,461 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:08:55,641 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:08:56,841 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'yaml' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 33, in init_op + yml_conf = yaml.safe_load(f) +NameError: name 'yaml' is not defined +INFO 2021-12-29 06:09:00,289 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:09:00,290 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:09:00,291 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:10:19,805 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:10:19,805 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:10:19,806 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:10:19,806 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:10:19,806 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:10:19,806 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:10:19,806 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:10:19,830 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:10:19,831 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:10:19,831 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:10:19,831 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:10:19,878 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:10:19,882 [dag.py:816] [DAG] start +INFO 2021-12-29 06:10:19,883 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:10:19,890 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:10:19,904 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:10:19,905 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:10:19,905 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:10:21,094 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:10:22,339 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'yaml' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 33, in init_op + yml_conf = yaml.safe_load(f) +NameError: name 'yaml' is not defined +INFO 2021-12-29 06:10:26,440 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:10:26,441 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:10:26,442 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-29 06:11:59,537 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:11:59,538 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:11:59,538 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:12:08,933 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:12:08,934 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:12:08,934 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:12:08,934 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:12:08,934 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:12:08,934 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:12:08,934 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:12:08,958 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:12:08,959 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:12:08,959 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:12:08,959 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:12:09,004 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:12:09,008 [dag.py:816] [DAG] start +INFO 2021-12-29 06:12:09,009 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:12:09,016 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:12:09,032 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:12:09,032 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:12:09,032 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:12:10,227 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:12:11,443 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 40, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +INFO 2021-12-29 06:12:12,867 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:12:12,868 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:12:12,869 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +WARNING 2021-12-29 06:12:47,188 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:12:47,188 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:12:47,191 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:12:47,191 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:12:47,191 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:12:47,191 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:12:47,191 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:12:47,191 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:12:47,191 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:12:47,213 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:12:47,213 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:12:47,213 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:12:47,213 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:12:47,259 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:12:47,263 [dag.py:816] [DAG] start +INFO 2021-12-29 06:12:47,264 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:12:47,269 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:12:47,293 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:12:47,293 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:12:47,293 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:12:48,478 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:12:49,708 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 41, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +INFO 2021-12-29 06:12:50,036 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:12:50,037 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:12:50,038 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +WARNING 2021-12-29 06:15:10,463 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:15:10,466 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:15:10,466 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:15:10,466 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:15:10,466 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:15:10,466 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:15:10,466 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:15:10,466 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:15:10,488 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:15:10,488 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:15:10,489 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:15:10,489 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:15:10,535 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:15:10,540 [dag.py:816] [DAG] start +INFO 2021-12-29 06:15:10,540 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:15:10,546 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:15:10,568 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:15:10,569 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:15:10,569 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:15:11,748 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:15:12,951 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 43, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:17:36,321 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:17:36,321 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:17:36,324 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:17:36,324 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:17:36,324 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:17:36,324 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:17:36,324 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:17:36,324 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:17:36,324 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:17:36,346 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:17:36,347 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:17:36,347 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:17:36,348 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:17:36,391 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:17:36,394 [dag.py:816] [DAG] start +INFO 2021-12-29 06:17:36,395 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:17:36,401 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:17:36,427 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:17:36,428 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:17:36,428 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:17:37,610 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:17:38,816 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 44, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:18:17,411 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:18:17,412 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:18:17,412 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:18:17,412 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:18:17,412 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:18:17,412 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:18:17,412 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:18:17,436 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:18:17,436 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:18:17,436 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:18:17,437 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:18:17,477 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:18:17,482 [dag.py:816] [DAG] start +INFO 2021-12-29 06:18:17,483 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:18:17,491 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:18:17,505 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:18:17,506 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:18:17,506 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:18:18,701 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:18:19,908 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 45, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:19:57,871 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:19:57,874 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:19:57,874 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:19:57,874 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:19:57,874 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:19:57,874 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:19:57,875 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:19:57,875 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:19:57,900 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:19:57,900 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:19:57,901 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:19:57,901 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:19:57,945 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:19:57,950 [dag.py:816] [DAG] start +INFO 2021-12-29 06:19:57,951 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:19:57,957 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:19:57,975 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:19:57,975 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:19:57,976 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:19:59,153 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:20:00,415 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 45, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:21:28,632 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:21:28,632 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:21:28,632 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:21:28,632 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:21:28,632 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:21:28,632 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:21:28,632 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:21:28,657 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:21:28,658 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:21:28,658 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:21:28,658 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:21:28,702 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:21:28,705 [dag.py:816] [DAG] start +INFO 2021-12-29 06:21:28,706 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:21:28,711 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:21:28,728 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:21:28,728 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:21:28,729 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:21:29,937 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:21:31,123 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 46, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:25:12,051 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:25:12,051 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:25:12,054 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:25:12,054 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:25:12,054 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:25:12,054 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:25:12,054 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:25:12,054 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:25:12,054 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:25:12,078 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:25:12,078 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:25:12,079 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:25:12,079 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:25:12,121 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:25:12,125 [dag.py:816] [DAG] start +INFO 2021-12-29 06:25:12,125 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:25:12,131 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:25:12,155 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:25:12,156 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:25:12,156 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:25:13,482 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:25:14,695 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 46, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:25:29,447 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:25:29,447 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:25:29,447 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:25:29,447 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:25:29,448 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:25:29,448 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:25:29,448 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:25:29,471 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:25:29,472 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:25:29,472 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:25:29,472 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:25:29,515 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:25:29,519 [dag.py:816] [DAG] start +INFO 2021-12-29 06:25:29,519 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:25:29,525 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:25:29,544 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:25:29,544 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:25:29,545 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:25:30,728 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:25:31,935 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 47, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:31:43,452 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:31:43,455 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:31:43,455 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:31:43,455 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:31:43,455 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:31:43,455 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:31:43,455 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:31:43,456 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:31:43,479 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:31:43,479 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:31:43,480 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:31:43,480 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:31:43,524 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:31:43,527 [dag.py:816] [DAG] start +INFO 2021-12-29 06:31:43,528 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:31:43,534 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:31:43,559 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:31:43,560 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:31:43,560 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:31:44,753 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2021-12-29 06:31:45,962 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 49, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:32:31,020 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:32:31,023 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:32:31,023 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:32:31,023 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:32:31,023 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:32:31,023 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:32:31,023 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:32:31,024 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:32:31,048 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:32:31,049 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:32:31,049 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:32:31,049 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:32:31,091 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:32:31,095 [dag.py:816] [DAG] start +INFO 2021-12-29 06:32:31,096 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:32:31,102 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:32:31,127 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:32:31,128 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:32:31,128 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:32:32,329 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:32:33,520 [operator.py:1178] [ppyolo_mbv3|0] Succ init +WARNING 2021-12-29 06:33:05,247 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:33:05,252 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:33:05,252 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:33:05,252 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:33:05,252 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:33:05,252 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:33:05,253 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:33:05,253 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:33:05,276 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:33:05,277 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:33:05,277 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:33:05,277 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:33:05,321 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:33:05,324 [dag.py:816] [DAG] start +INFO 2021-12-29 06:33:05,325 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:33:05,331 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:33:05,355 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:33:05,356 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:33:05,356 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:33:06,480 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:33:07,686 [operator.py:1178] [ppyolo_mbv3|0] Succ init +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:40:19,259 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:40:19,259 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:40:19,259 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:40:19,259 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:40:19,259 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:40:19,260 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:40:19,260 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:40:19,260 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:40:19,285 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:40:19,286 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:40:19,286 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:40:19,286 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:40:19,333 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:40:19,337 [dag.py:816] [DAG] start +INFO 2021-12-29 06:40:19,337 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:40:19,343 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:40:19,359 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:40:19,360 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:40:19,360 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:40:20,576 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:40:21,771 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:40:25,060 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:40:25,061 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:40:25,062 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:40:25,105 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 54, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:40:25,111 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:42:03,383 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:42:03,383 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:42:03,383 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:42:03,383 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:42:03,383 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:42:03,383 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:42:03,384 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:42:03,408 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:42:03,409 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:42:03,409 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:42:03,410 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:42:03,452 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:42:03,456 [dag.py:816] [DAG] start +INFO 2021-12-29 06:42:03,456 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:42:03,464 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:42:03,483 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:42:03,484 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:42:03,484 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:42:04,712 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:42:05,914 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:42:13,067 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:42:13,069 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:42:13,070 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:42:13,106 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 54, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:42:13,112 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:42:44,176 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:42:44,176 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:42:44,177 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:42:44,177 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:42:44,177 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:42:44,177 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:42:44,177 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:42:44,201 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:42:44,201 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:42:44,202 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:42:44,202 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:42:44,248 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:42:44,252 [dag.py:816] [DAG] start +INFO 2021-12-29 06:42:44,253 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:42:44,259 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:42:44,278 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:42:44,279 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:42:44,279 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:42:45,483 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:42:46,679 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:42:47,152 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:42:47,153 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:42:47,154 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:42:47,195 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 55, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:42:47,200 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:44:34,235 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:44:34,235 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:44:34,235 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:44:34,235 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:44:34,235 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:44:34,235 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:44:34,236 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:44:34,236 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:44:34,258 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:44:34,259 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:44:34,259 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:44:34,259 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:44:34,301 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:44:34,306 [dag.py:816] [DAG] start +INFO 2021-12-29 06:44:34,307 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:44:34,313 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:44:34,329 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:44:34,330 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:44:34,330 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:44:35,557 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:44:36,750 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:44:43,663 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:44:43,664 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:44:43,665 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:44:43,710 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 55, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:44:43,715 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:46:19,032 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:46:19,032 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:46:19,033 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:46:19,033 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:46:19,033 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:46:19,033 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:46:19,033 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:46:19,057 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:46:19,058 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:46:19,058 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:46:19,058 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:46:19,108 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:46:19,111 [dag.py:816] [DAG] start +INFO 2021-12-29 06:46:19,112 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:46:19,118 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:46:19,146 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:46:19,147 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:46:19,147 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:46:20,373 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:46:21,561 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:46:25,488 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:46:25,490 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:46:25,490 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:46:25,581 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 71, in preprocess + im = self.img_preprocess(im) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 492, in __call__ + img = t(img) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 642, in __call__ + return F.normalize(img, self.mean, self.std, self.channel_first) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/functional.py", line 33, in normalize + img -= img_mean +ValueError: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +ERROR 2021-12-29 06:46:25,587 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:51:01,069 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:51:01,069 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:51:01,069 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:51:01,069 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:51:01,069 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:51:01,069 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:51:01,069 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:51:01,094 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:51:01,095 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:51:01,095 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:51:01,095 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:51:01,140 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:51:01,144 [dag.py:816] [DAG] start +INFO 2021-12-29 06:51:01,145 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:51:01,151 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:51:01,171 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:51:01,172 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:51:01,172 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:51:02,379 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:51:03,568 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:51:06,257 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:51:06,258 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:51:06,259 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:51:07,885 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 89, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:51:07,889 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:51:56,944 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:51:56,947 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:51:56,947 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:51:56,947 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:51:56,947 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:51:56,948 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:51:56,948 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:51:56,948 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:51:56,971 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:51:56,972 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:51:56,972 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:51:56,972 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:51:57,019 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:51:57,023 [dag.py:816] [DAG] start +INFO 2021-12-29 06:51:57,024 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:51:57,029 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:51:57,049 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:51:57,050 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:51:57,050 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:51:58,256 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:51:59,445 [operator.py:1178] [ppyolo_mbv3|0] Succ init +WARNING 2021-12-29 06:52:13,864 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:52:13,867 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:52:13,867 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:52:13,867 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:52:13,867 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:52:13,868 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:52:13,868 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:52:13,868 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:52:13,892 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:52:13,893 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:52:13,893 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:52:13,893 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:52:13,936 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:52:13,939 [dag.py:816] [DAG] start +INFO 2021-12-29 06:52:13,940 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:52:13,945 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:52:13,966 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:52:13,966 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:52:13,966 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:52:15,154 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:52:16,355 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:52:18,677 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:52:18,678 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:52:18,679 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:52:20,309 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 86, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:52:20,314 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 06:54:00,626 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 06:54:00,626 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:54:00,627 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 06:54:00,627 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 06:54:00,627 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 06:54:00,627 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 06:54:00,627 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 06:54:00,652 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 06:54:00,653 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 06:54:00,653 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 06:54:00,653 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 06:54:00,699 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 06:54:00,703 [dag.py:816] [DAG] start +INFO 2021-12-29 06:54:00,703 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 06:54:00,709 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 06:54:00,732 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 06:54:00,733 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 06:54:00,733 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 06:54:01,924 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 06:54:03,172 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 06:54:05,727 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 06:54:05,728 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 06:54:05,729 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 06:54:07,453 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 86, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:54:07,458 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:13:12,175 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:13:12,175 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:13:12,178 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:13:12,178 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:13:12,178 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:13:12,178 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:13:12,178 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:13:12,178 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:13:12,178 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:13:12,201 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:13:12,202 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:13:12,202 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:13:12,202 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:13:12,251 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:13:12,254 [dag.py:816] [DAG] start +INFO 2021-12-29 07:13:12,255 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:13:12,260 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:13:12,284 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:13:12,284 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:13:12,285 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:13:13,478 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:13:14,670 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:13:18,607 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:13:18,609 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:13:18,609 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:13:20,265 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:13:20,269 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:16:27,832 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:16:27,832 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:16:27,835 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:16:27,835 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:16:27,835 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:16:27,835 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:16:27,835 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:16:27,835 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:16:27,835 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:16:27,860 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:16:27,861 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:16:27,861 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:16:27,861 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:16:27,904 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:16:27,907 [dag.py:816] [DAG] start +INFO 2021-12-29 07:16:27,908 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:16:27,915 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:16:27,939 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:16:27,939 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:16:27,939 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:16:29,142 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:16:30,333 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:16:32,190 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:16:32,192 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:16:32,193 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:16:32,263 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: name 'im_shape' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 68, in preprocess + "im_shape": im_info[im_shape],#np.array(list(im.shape[1:])).reshape(-1)[np.newaxis,:], +NameError: name 'im_shape' is not defined +ERROR 2021-12-29 07:16:32,267 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: name 'im_shape' is not defined +WARNING 2021-12-29 07:17:33,514 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:17:33,514 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:17:33,517 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:17:33,517 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:17:33,517 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:17:33,517 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:17:33,517 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:17:33,517 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:17:33,518 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:17:33,542 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:17:33,543 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:17:33,543 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:17:33,543 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:17:33,585 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:17:33,588 [dag.py:816] [DAG] start +INFO 2021-12-29 07:17:33,589 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:17:33,594 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:17:33,615 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:17:33,616 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:17:33,616 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:17:34,803 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:17:35,996 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:17:38,155 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:17:38,156 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:17:38,158 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:17:39,797 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:17:39,802 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:19:06,141 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:19:06,144 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:19:06,144 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:19:06,144 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:19:06,144 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:19:06,144 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:19:06,144 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:19:06,144 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:19:06,167 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:19:06,168 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:19:06,168 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:19:06,168 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:19:06,213 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:19:06,217 [dag.py:816] [DAG] start +INFO 2021-12-29 07:19:06,219 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:19:06,224 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:19:06,242 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:19:06,242 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:19:06,243 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:19:07,427 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:19:08,615 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:19:11,136 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:19:11,137 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:19:11,137 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:19:12,763 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:19:12,767 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:42:11,631 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:42:11,631 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:42:11,632 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:42:11,632 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:42:11,632 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:42:11,632 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:42:11,634 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:42:11,659 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:42:11,660 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:42:11,660 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:42:11,660 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:42:11,707 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:42:11,711 [dag.py:816] [DAG] start +INFO 2021-12-29 07:42:11,713 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:42:11,719 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:42:11,742 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:42:11,742 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:42:11,743 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:42:12,964 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:42:14,178 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:42:24,007 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:42:24,009 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:42:24,009 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:42:25,683 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:42:25,687 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:49:58,432 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-29 07:49:58,435 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-29 07:49:58,435 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:49:58,435 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-29 07:49:58,435 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-29 07:49:58,435 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-29 07:49:58,435 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-29 07:49:58,435 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-29 07:49:58,458 [dag.py:493] [DAG] Succ init +INFO 2021-12-29 07:49:58,459 [dag.py:651] ================= USED OP ================= +INFO 2021-12-29 07:49:58,459 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-29 07:49:58,460 [dag.py:655] ------------------------------------------- +INFO 2021-12-29 07:49:58,505 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-29 07:49:58,509 [dag.py:816] [DAG] start +INFO 2021-12-29 07:49:58,509 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-29 07:49:58,515 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-29 07:49:58,539 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-29 07:49:58,540 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-29 07:49:58,540 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-29 07:49:59,775 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-29 07:50:00,998 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-29 07:50:03,927 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-29 07:50:03,928 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-29 07:50:03,930 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-29 07:50:05,801 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_4.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_4.tmp_1.lod' +ERROR 2021-12-29 07:50:05,806 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_4.tmp_1.lod' +WARNING 2021-12-30 06:53:16,236 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 06:53:16,239 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 06:53:16,239 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 06:53:16,239 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 06:53:16,239 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 06:53:16,239 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 06:53:16,239 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 06:53:16,239 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 06:53:16,265 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 06:53:16,266 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 06:53:16,266 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 06:53:16,266 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 06:53:16,310 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 06:53:16,315 [dag.py:816] [DAG] start +INFO 2021-12-30 06:53:16,316 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 06:53:16,323 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 06:53:16,341 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 06:53:16,341 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 06:53:16,342 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 06:53:17,563 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 06:53:18,787 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 06:53:26,765 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 06:53:26,766 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 06:53:26,766 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 06:53:28,471 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_3.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 81, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_3.tmp_1.lod' +ERROR 2021-12-30 06:53:28,476 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_3.tmp_1.lod' +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 07:57:06,807 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 07:57:06,807 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 07:57:06,807 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 07:57:06,807 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 07:57:06,807 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 07:57:06,807 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 07:57:06,808 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 07:57:06,808 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 07:57:06,808 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 07:57:06,831 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 07:57:06,832 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 07:57:06,832 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 07:57:06,832 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 07:57:06,878 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 07:57:06,882 [dag.py:816] [DAG] start +INFO 2021-12-30 07:57:06,883 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 07:57:06,890 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 07:57:06,909 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 07:57:06,910 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 07:57:06,910 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 07:57:08,117 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 07:57:09,361 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 07:57:13,486 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 07:57:13,487 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 07:57:13,487 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 07:57:15,249 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 81, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 07:57:15,253 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:12:07,135 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:12:07,135 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:12:07,135 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:12:07,135 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:12:07,135 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:12:07,135 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:12:07,136 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:12:07,160 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:12:07,161 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:12:07,161 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:12:07,161 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:12:07,206 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:12:07,210 [dag.py:816] [DAG] start +INFO 2021-12-30 08:12:07,211 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:12:07,218 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:12:07,237 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:12:07,237 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:12:07,237 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:12:08,426 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:12:09,638 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:12:13,041 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:12:13,042 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:12:13,043 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:12:14,700 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 82, in postprocess + np_score_list.append(fetch_dict[out_idx]) +KeyError: 0 +ERROR 2021-12-30 08:12:14,707 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:13:45,672 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:13:45,675 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:13:45,675 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:13:45,675 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:13:45,675 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:13:45,675 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:13:45,675 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:13:45,675 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:13:45,700 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:13:45,701 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:13:45,701 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:13:45,701 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:13:45,748 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:13:45,751 [dag.py:816] [DAG] start +INFO 2021-12-30 08:13:45,752 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:13:45,758 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:13:45,779 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:13:45,780 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:13:45,781 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:13:46,997 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:13:48,252 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:13:50,562 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:13:50,562 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:13:50,563 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:13:52,217 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[out_idx]) +KeyError: 0 +ERROR 2021-12-30 08:13:52,220 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:17:15,483 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:17:15,483 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:17:15,483 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:17:15,483 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:17:15,483 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:17:15,483 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:17:15,484 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:17:15,508 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:17:15,509 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:17:15,509 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:17:15,509 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:17:15,556 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:17:15,560 [dag.py:816] [DAG] start +INFO 2021-12-30 08:17:15,560 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:17:15,567 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:17:15,589 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:17:15,589 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:17:15,590 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:17:16,807 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:17:18,022 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:17:20,506 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:17:20,508 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:17:20,509 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:17:22,212 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:17:22,217 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:17:30,895 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:17:30,900 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:17:30,900 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:17:30,900 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:17:30,900 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:17:30,900 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:17:30,901 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:17:30,901 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:17:30,924 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:17:30,925 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:17:30,925 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:17:30,925 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:17:30,971 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:17:30,976 [dag.py:816] [DAG] start +INFO 2021-12-30 08:17:30,977 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:17:30,983 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:17:31,007 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:17:31,008 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:17:31,008 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:17:32,203 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:17:33,418 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:17:37,462 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:17:37,464 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:17:37,464 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:17:39,176 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:17:39,180 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:20:13,195 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:20:13,198 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:20:13,198 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:20:13,198 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:20:13,198 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:20:13,198 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:20:13,198 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:20:13,198 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:20:13,222 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:20:13,223 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:20:13,223 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:20:13,223 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:20:13,267 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:20:13,270 [dag.py:816] [DAG] start +INFO 2021-12-30 08:20:13,271 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:20:13,276 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:20:13,295 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:20:13,295 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:20:13,296 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:20:14,485 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:20:15,692 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:20:19,232 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:20:19,233 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:20:19,233 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:20:20,892 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 84, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:20:20,896 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:31:54,774 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:31:54,774 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:31:54,774 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:31:54,775 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:31:54,775 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:31:54,775 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:31:54,775 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:31:54,799 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:31:54,800 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:31:54,800 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:31:54,800 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:31:54,846 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:31:54,851 [dag.py:816] [DAG] start +INFO 2021-12-30 08:31:54,852 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:31:54,857 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:31:54,877 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:31:54,877 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:31:54,877 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:31:56,080 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:31:57,306 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:31:59,636 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:31:59,638 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:31:59,638 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:32:01,331 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 91, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:32:01,335 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:34:21,409 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:34:21,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:34:21,411 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:34:21,412 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:34:21,412 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:34:21,412 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:34:21,412 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:34:21,412 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:34:21,412 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:34:21,437 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:34:21,438 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:34:21,438 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:34:21,438 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:34:21,482 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:34:21,486 [dag.py:816] [DAG] start +INFO 2021-12-30 08:34:21,487 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:34:21,493 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:34:21,522 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:34:21,522 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:34:21,523 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:34:22,753 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:34:23,983 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:34:27,034 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:34:27,036 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:34:27,036 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:34:28,756 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 96, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:34:28,761 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:43:32,266 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:43:32,266 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:43:32,270 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:43:32,270 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:43:32,270 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:43:32,271 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:43:32,271 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:43:32,271 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:43:32,271 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:43:32,272 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:43:32,301 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:43:32,302 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:43:32,302 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:43:32,302 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:43:32,343 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:43:32,348 [dag.py:816] [DAG] start +INFO 2021-12-30 08:43:32,349 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:43:32,355 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:43:32,375 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:43:32,376 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:43:32,376 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:43:33,601 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:43:34,822 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:43:44,530 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:43:44,532 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:43:44,532 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:43:46,213 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 111, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:43:46,218 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:49:32,281 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:49:32,281 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:49:32,285 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:49:32,286 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:49:32,286 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:49:32,287 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:49:32,287 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:49:32,287 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:49:32,288 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:49:32,288 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:49:32,317 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:49:32,318 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:49:32,318 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:49:32,318 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:49:32,355 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:49:32,359 [dag.py:816] [DAG] start +INFO 2021-12-30 08:49:32,360 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:49:32,365 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:49:32,387 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:49:32,388 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:49:32,388 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:49:33,652 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:49:34,901 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:49:38,204 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:49:38,206 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:49:38,206 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:49:39,879 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 113, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:49:39,884 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:52:42,378 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:52:42,383 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:52:42,384 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:52:42,384 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:52:42,384 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:52:42,385 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:52:42,385 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:52:42,386 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:52:42,408 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:52:42,408 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:52:42,408 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:52:42,409 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:52:42,444 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:52:42,448 [dag.py:816] [DAG] start +INFO 2021-12-30 08:52:42,449 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:52:42,454 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:52:42,478 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:52:42,479 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:52:42,479 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:52:43,723 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:52:44,962 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 08:52:48,301 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 08:52:48,303 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 08:52:48,304 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 08:52:50,006 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 110, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:52:50,011 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:59:43,426 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 08:59:43,427 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 08:59:43,427 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:59:43,427 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 08:59:43,427 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 08:59:43,428 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 08:59:43,428 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 08:59:43,428 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 08:59:43,457 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 08:59:43,458 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 08:59:43,458 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 08:59:43,458 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 08:59:43,496 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 08:59:43,500 [dag.py:816] [DAG] start +INFO 2021-12-30 08:59:43,501 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 08:59:43,507 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 08:59:43,525 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 08:59:43,525 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 08:59:43,526 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 08:59:44,766 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 08:59:45,988 [operator.py:1178] [ppyolo_mbv3|0] Succ init +WARNING 2021-12-30 09:03:09,956 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:03:09,956 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:03:09,961 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:03:09,961 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:03:09,961 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:03:09,962 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:03:09,962 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:03:09,962 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:03:09,962 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:03:09,989 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:03:09,990 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:03:09,990 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:03:09,990 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:03:10,028 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:03:10,033 [dag.py:816] [DAG] start +INFO 2021-12-30 09:03:10,033 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:03:10,039 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:03:10,058 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:03:10,059 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:03:10,059 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:03:11,906 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:03:11,907 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:03:11,907 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:03:11,907 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:03:11,908 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:03:11,908 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:03:11,908 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:03:11,938 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:03:11,939 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:03:11,939 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:03:11,939 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:03:11,981 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:03:11,985 [dag.py:816] [DAG] start +INFO 2021-12-30 09:03:11,986 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:03:11,992 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:03:12,011 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:03:12,011 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:03:12,011 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:03:13,259 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:03:14,522 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:03:20,138 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:03:20,139 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:03:20,139 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:03:22,048 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 09:03:22,053 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 09:07:27,378 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:07:27,383 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:07:27,384 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:07:27,384 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:07:27,384 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:07:27,385 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":18083, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:07:27,385 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:07:27,385 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:07:27,415 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:07:27,416 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:07:27,416 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:07:27,416 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:07:27,458 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:07:27,462 [dag.py:816] [DAG] start +INFO 2021-12-30 09:07:27,463 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:07:27,468 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:07:27,490 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:07:27,491 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:07:27,491 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:07:28,741 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:07:29,959 [operator.py:1178] [ppyolo_mbv3|0] Succ init +WARNING 2021-12-30 09:07:48,750 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:07:48,754 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:07:48,754 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:07:48,755 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:07:48,755 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:07:48,755 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:07:48,755 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:07:48,756 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:07:48,756 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:07:48,756 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:07:48,785 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:07:48,785 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:07:48,786 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:07:48,786 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:07:48,827 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:07:48,831 [dag.py:816] [DAG] start +INFO 2021-12-30 09:07:48,832 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:07:48,838 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:07:48,859 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:07:48,859 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:07:48,860 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:07:50,077 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:07:50,732 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:07:50,733 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:07:50,734 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:07:51,330 [operator.py:1178] [ppyolo_mbv3|0] Succ init +ERROR 2021-12-30 09:07:53,007 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 09:07:53,011 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 09:08:06,502 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:08:06,507 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:08:06,507 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:08:06,508 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:08:06,508 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:08:06,508 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:08:06,508 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:08:06,508 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:08:06,533 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:08:06,534 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:08:06,534 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:08:06,534 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:08:06,572 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:08:06,576 [dag.py:816] [DAG] start +INFO 2021-12-30 09:08:06,577 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:08:06,583 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:08:06,609 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:08:06,610 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:08:06,610 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:08:07,840 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:08:09,064 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:08:09,866 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:08:09,868 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:08:09,868 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:08:11,524 [operator.py:1000] (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +ERROR 2021-12-30 09:08:11,527 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +WARNING 2021-12-30 09:19:39,408 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:19:39,412 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:19:39,412 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:19:39,413 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:19:39,414 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:19:39,414 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:19:39,414 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:19:39,415 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:19:39,415 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:19:39,415 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:19:39,447 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:19:39,447 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:19:39,447 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:19:39,447 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:19:39,488 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:19:39,492 [dag.py:816] [DAG] start +INFO 2021-12-30 09:19:39,493 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:19:39,499 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:19:39,520 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:19:39,520 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:19:39,520 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:19:40,827 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:19:42,045 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:19:45,115 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:19:45,116 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:19:45,117 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:19:46,754 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'res_dict' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 101, in postprocess + res_dict[b] = {} +NameError: name 'res_dict' is not defined +ERROR 2021-12-30 09:19:46,757 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'res_dict' is not defined +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:20:11,184 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:20:11,184 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:20:11,184 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:20:11,184 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:20:11,184 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:20:11,184 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:20:11,185 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:20:11,204 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:20:11,205 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:20:11,205 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:20:11,205 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:20:11,244 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:20:11,247 [dag.py:816] [DAG] start +INFO 2021-12-30 09:20:11,248 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:20:11,254 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:20:11,279 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:20:11,280 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:20:11,280 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:20:12,540 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:20:13,094 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:20:13,095 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:20:13,095 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:20:13,763 [operator.py:1178] [ppyolo_mbv3|0] Succ init +ERROR 2021-12-30 09:20:15,417 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict[b] = {} +IndexError: list assignment index out of range +ERROR 2021-12-30 09:20:15,421 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +WARNING 2021-12-30 09:21:19,542 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:21:19,547 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:21:19,547 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:21:19,547 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:21:19,547 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:21:19,548 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:21:19,548 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:21:19,548 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:21:19,578 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:21:19,578 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:21:19,579 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:21:19,579 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:21:19,619 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:21:19,624 [dag.py:816] [DAG] start +INFO 2021-12-30 09:21:19,625 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:21:19,632 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:21:19,655 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:21:19,655 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:21:19,656 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:21:20,877 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:21:22,086 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:21:22,662 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:21:22,663 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:21:22,666 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:21:24,326 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 103, in postprocess + res_dict[b] = {} +IndexError: list assignment index out of range +ERROR 2021-12-30 09:21:24,330 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:23:31,828 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:23:31,829 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:23:31,829 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:23:31,829 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:23:31,830 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:23:31,830 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:23:31,830 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:23:31,860 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:23:31,861 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:23:31,861 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:23:31,861 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:23:31,903 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:23:31,907 [dag.py:816] [DAG] start +INFO 2021-12-30 09:23:31,907 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:23:31,913 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:23:31,943 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:23:31,943 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:23:31,943 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:23:33,178 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:23:34,435 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:23:35,932 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:23:35,933 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:23:35,934 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:23:37,722 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'a' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 101, in postprocess + for b in range(a.ndim): +NameError: name 'a' is not defined +ERROR 2021-12-30 09:23:37,726 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'a' is not defined +WARNING 2021-12-30 09:24:02,842 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:24:02,842 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:24:02,846 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:24:02,846 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:24:02,847 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:24:02,847 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:24:02,847 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:24:02,847 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:24:02,847 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:24:02,848 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:24:02,877 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:24:02,878 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:24:02,878 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:24:02,878 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:24:02,920 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:24:02,924 [dag.py:816] [DAG] start +INFO 2021-12-30 09:24:02,925 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:24:02,931 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:24:02,953 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:24:02,954 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:24:02,954 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:24:04,079 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:24:04,080 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:24:04,081 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:24:04,185 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:24:05,401 [operator.py:1178] [ppyolo_mbv3|0] Succ init +ERROR 2021-12-30 09:24:07,046 [operator.py:1000] (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +ERROR 2021-12-30 09:24:07,049 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +WARNING 2021-12-30 09:25:54,030 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:25:54,034 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:25:54,034 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:25:54,035 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:25:54,036 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:25:54,036 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:25:54,036 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:25:54,037 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:25:54,037 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:25:54,037 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:25:54,066 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:25:54,066 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:25:54,066 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:25:54,066 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:25:54,109 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:25:54,113 [dag.py:816] [DAG] start +INFO 2021-12-30 09:25:54,114 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:25:54,120 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:25:54,138 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:25:54,138 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:25:54,139 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:25:55,393 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:25:56,645 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:25:56,651 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:25:56,652 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:25:56,652 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:25:58,333 [dag.py:404] (data_id=0 log_id=0) Succ predict +ERROR 2021-12-30 09:25:58,334 [operator.py:1487] (logid=0) Failed to pack RPC response package: +WARNING 2021-12-30 09:39:07,135 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:39:07,140 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:39:07,140 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:39:07,140 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:39:07,140 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:39:07,141 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:39:07,141 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:39:07,141 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:39:07,169 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:39:07,170 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:39:07,170 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:39:07,170 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:39:07,210 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:39:07,215 [dag.py:816] [DAG] start +INFO 2021-12-30 09:39:07,216 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:39:07,223 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:39:07,243 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:39:07,243 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:39:07,244 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:39:08,460 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:39:09,703 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:39:25,749 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:39:25,750 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:39:25,751 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:39:27,547 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 09:40:00,135 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:40:00,136 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:40:00,136 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:40:00,249 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:00,253 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +INFO 2021-12-30 09:40:48,641 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:40:48,642 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:40:48,642 [dag.py:368] (data_id=2 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:40:48,711 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:48,714 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +INFO 2021-12-30 09:40:53,612 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:40:53,613 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:40:53,614 [dag.py:368] (data_id=3 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:40:53,720 [operator.py:973] (data_id=3 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:53,723 [dag.py:409] (data_id=3 log_id=0) Failed to predict: (data_id=3 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:42:01,778 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:42:01,783 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:42:01,783 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:42:01,784 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:42:01,784 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:42:01,784 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:42:01,784 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:42:01,784 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:42:01,813 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:42:01,814 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:42:01,814 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:42:01,814 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:42:01,853 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:42:01,858 [dag.py:816] [DAG] start +INFO 2021-12-30 09:42:01,859 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:42:01,865 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:42:01,889 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:42:01,889 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:42:01,889 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:42:03,109 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:42:04,298 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:42:04,956 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:42:04,958 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:42:04,958 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:42:06,612 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 09:42:06,616 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +INFO 2021-12-30 09:42:19,843 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:42:19,844 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:42:19,845 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:42:19,953 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 09:42:19,957 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:42:23,967 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:42:23,968 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:42:23,968 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:42:23,968 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:42:23,969 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:42:23,969 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:42:23,970 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:42:23,970 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:42:24,000 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:42:24,001 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:42:24,001 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:42:24,001 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:42:24,042 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:42:24,047 [dag.py:816] [DAG] start +INFO 2021-12-30 09:42:24,048 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:42:24,054 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:42:24,072 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:42:24,073 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:42:24,073 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:42:25,280 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:42:26,470 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:42:26,701 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:42:26,702 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:42:26,702 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:42:28,405 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 09:43:10,878 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:43:10,879 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:43:10,879 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:43:10,985 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:43:10,988 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +INFO 2021-12-30 09:43:15,839 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:43:15,840 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:43:15,840 [dag.py:368] (data_id=2 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:43:15,944 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:43:15,950 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:44:54,821 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:44:54,826 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:44:54,826 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:44:54,826 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:44:54,826 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:44:54,827 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:44:54,827 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:44:54,827 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:44:54,856 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:44:54,856 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:44:54,857 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:44:54,857 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:44:54,898 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:44:54,901 [dag.py:816] [DAG] start +INFO 2021-12-30 09:44:54,902 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:44:54,908 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:44:54,927 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:44:54,927 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:44:54,927 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:44:56,140 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:44:57,521 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:44:59,119 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:44:59,121 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:44:59,122 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:45:00,979 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 09:45:05,363 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:45:05,364 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:45:05,364 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:45:05,468 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:45:05,472 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:46:17,679 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:46:17,683 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:46:17,683 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 09:46:17,684 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 09:46:17,684 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:46:17,684 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 09:46:17,684 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 09:46:17,685 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 09:46:17,685 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 09:46:17,685 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 09:46:17,714 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 09:46:17,715 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 09:46:17,715 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 09:46:17,715 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 09:46:17,756 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 09:46:17,760 [dag.py:816] [DAG] start +INFO 2021-12-30 09:46:17,761 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 09:46:17,766 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 09:46:17,795 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 09:46:17,796 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 09:46:17,796 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 09:46:19,035 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 09:46:20,237 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 09:46:21,068 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:46:21,070 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:46:21,071 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 09:46:22,764 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 09:46:25,462 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 09:46:25,463 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 09:46:25,463 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 09:46:25,584 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:46:25,588 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +INFO 2021-12-30 11:05:40,431 [pipeline_server.py:51] (log_id=0) inference request name:recognition self.name:ppyolo_mbv3 +ERROR 2021-12-30 11:05:40,432 [pipeline_server.py:55] (log_id=0) name dismatch error. request.name:recognition,server.name=ppyolo_mbv3 +INFO 2021-12-30 11:05:49,163 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:05:49,164 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:05:49,164 [dag.py:368] (data_id=2 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:05:49,271 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:05:49,274 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:05:54,559 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:05:54,564 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:05:54,564 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:05:54,565 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:05:54,565 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:05:54,566 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:05:54,566 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:05:54,566 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:05:54,596 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:05:54,597 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:05:54,597 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:05:54,597 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:05:54,636 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:05:54,641 [dag.py:816] [DAG] start +INFO 2021-12-30 11:05:54,642 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:05:54,648 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:05:54,671 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:05:54,671 [operator.py:1167] Init cuda env in process 0 +INFO 2021-12-30 11:05:54,672 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:05:55,892 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:05:56,096 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:05:56,097 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:05:56,098 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 11:05:57,079 [operator.py:1178] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 11:05:58,729 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 11:15:50,113 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:15:50,115 [operator.py:1426] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:15:50,115 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:15:50,225 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:15:50,229 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:24:24,432 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:24:24,432 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:24:24,432 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:24:24,433 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:24:24,433 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:24:24,433 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:24:24,433 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:24:24,462 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:24:24,463 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:24:24,463 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:24:24,463 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:24:24,503 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:24:24,508 [dag.py:816] [DAG] start +INFO 2021-12-30 11:24:24,508 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:24:24,513 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:24:24,533 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:24:24,534 [operator.py:1170] Init cuda env in process 0 +INFO 2021-12-30 11:24:24,534 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:24:25,753 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:24:26,946 [operator.py:1181] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 11:24:28,169 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:24:28,171 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:24:28,171 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 11:24:30,117 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 11:24:39,879 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:24:39,880 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:24:39,880 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:24:40,005 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:24:40,009 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:37:49,610 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:37:49,615 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:37:49,615 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:37:49,616 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:37:49,616 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:37:49,617 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:37:49,617 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:37:49,617 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:37:49,647 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:37:49,648 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:37:49,648 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:37:49,648 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:37:49,689 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:37:49,693 [dag.py:816] [DAG] start +INFO 2021-12-30 11:37:49,693 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:37:49,699 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:37:49,728 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:37:49,728 [operator.py:1170] Init cuda env in process 0 +INFO 2021-12-30 11:37:49,729 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:37:50,942 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:37:52,133 [operator.py:1181] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 11:37:52,375 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:37:52,377 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:37:52,377 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:37:54,033 [operator.py:976] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:37:54,039 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +INFO 2021-12-30 11:37:55,868 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:37:55,869 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:37:55,869 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:37:55,960 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:37:55,963 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 11:38:09,257 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:38:09,262 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:38:09,262 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:38:09,262 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:38:09,262 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:38:09,263 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:38:09,263 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:38:09,263 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:38:09,291 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:38:09,292 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:38:09,292 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:38:09,292 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:38:09,331 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:38:09,335 [dag.py:816] [DAG] start +INFO 2021-12-30 11:38:09,336 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:38:09,342 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:38:09,360 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:38:09,361 [operator.py:1170] Init cuda env in process 0 +INFO 2021-12-30 11:38:09,361 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:38:10,578 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:38:10,613 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:38:10,615 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:38:10,615 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 11:38:11,767 [operator.py:1181] [ppyolo_mbv3|0] Succ init +ERROR 2021-12-30 11:38:13,407 [operator.py:976] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:38:13,411 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +INFO 2021-12-30 11:40:01,272 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:40:01,273 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:40:01,274 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +ERROR 2021-12-30 11:40:01,379 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + self.post_process = PicoDetPostProcess( +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:40:01,383 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 11:40:05,557 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:40:05,557 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:40:05,561 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:40:05,561 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:40:05,562 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:40:05,562 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:40:05,562 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:40:05,562 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:40:05,562 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:40:05,592 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:40:05,593 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:40:05,593 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:40:05,593 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:40:05,633 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:40:05,638 [dag.py:816] [DAG] start +INFO 2021-12-30 11:40:05,639 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:40:05,646 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:40:05,665 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:40:05,665 [operator.py:1170] Init cuda env in process 0 +INFO 2021-12-30 11:40:05,665 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:40:06,892 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:40:08,120 [operator.py:1181] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 11:40:08,937 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:40:08,939 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:40:08,939 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 11:40:10,596 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 11:40:13,169 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:40:13,169 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:40:13,170 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +INFO 2021-12-30 11:40:13,272 [dag.py:404] (data_id=1 log_id=0) Succ predict +INFO 2021-12-30 11:40:59,103 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:40:59,104 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:40:59,105 [dag.py:368] (data_id=2 log_id=0) Succ Generate ID +INFO 2021-12-30 11:40:59,221 [dag.py:404] (data_id=2 log_id=0) Succ predict +WARNING 2021-12-30 11:41:26,917 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:41:26,921 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2021-12-30 11:41:26,922 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2021-12-30 11:41:26,922 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:41:26,922 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2021-12-30 11:41:26,923 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2021-12-30 11:41:26,923 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2021-12-30 11:41:26,924 [pipeline_server.py:212] ------------------------------------------- +INFO 2021-12-30 11:41:26,924 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2021-12-30 11:41:26,953 [dag.py:493] [DAG] Succ init +INFO 2021-12-30 11:41:26,954 [dag.py:651] ================= USED OP ================= +INFO 2021-12-30 11:41:26,954 [dag.py:654] ppyolo_mbv3 +INFO 2021-12-30 11:41:26,954 [dag.py:655] ------------------------------------------- +INFO 2021-12-30 11:41:26,993 [dag.py:784] [DAG] Succ build DAG +INFO 2021-12-30 11:41:26,997 [dag.py:816] [DAG] start +INFO 2021-12-30 11:41:26,997 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2021-12-30 11:41:27,003 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2021-12-30 11:41:27,026 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2021-12-30 11:41:27,027 [operator.py:1170] Init cuda env in process 0 +INFO 2021-12-30 11:41:27,027 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2021-12-30 11:41:28,253 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2021-12-30 11:41:28,845 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:41:28,846 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:41:28,847 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2021-12-30 11:41:29,451 [operator.py:1181] [ppyolo_mbv3|0] Succ init +INFO 2021-12-30 11:41:31,104 [dag.py:404] (data_id=0 log_id=0) Succ predict +INFO 2021-12-30 11:41:33,109 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2021-12-30 11:41:33,110 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2021-12-30 11:41:33,110 [dag.py:368] (data_id=1 log_id=0) Succ Generate ID +INFO 2021-12-30 11:41:33,209 [dag.py:404] (data_id=1 log_id=0) Succ predict +WARNING 2022-02-14 09:24:33,065 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-14 09:24:33,074 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-14 09:24:33,074 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:24:33,075 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-14 09:24:33,075 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2022-02-14 09:24:33,075 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-14 09:24:33,075 [pipeline_server.py:212] ------------------------------------------- +INFO 2022-02-14 09:24:33,075 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-14 09:24:33,096 [dag.py:493] [DAG] Succ init +INFO 2022-02-14 09:24:33,096 [dag.py:651] ================= USED OP ================= +INFO 2022-02-14 09:24:33,097 [dag.py:654] ppyolo_mbv3 +INFO 2022-02-14 09:24:33,097 [dag.py:655] ------------------------------------------- +INFO 2022-02-14 09:24:33,137 [dag.py:784] [DAG] Succ build DAG +INFO 2022-02-14 09:24:33,141 [dag.py:816] [DAG] start +INFO 2022-02-14 09:24:33,142 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-14 09:24:33,147 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2022-02-14 09:24:33,174 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:24:33,175 [operator.py:1170] Init cuda env in process 0 +INFO 2022-02-14 09:24:33,175 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-14 09:24:34,409 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2022-02-14 09:24:35,602 [operator.py:1179] [ppyolo_mbv3|0] failed to init op: [Errno 2] No such file or directory: 'label_list.txt' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1174, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1363, in _initialize + self.init_op() + File "web_service.py", line 30, in init_op + self.img_postprocess = RCNNPostprocess("label_list.txt", "output") + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 288, in __init__ + with open(label_file) as fin: +FileNotFoundError: [Errno 2] No such file or directory: 'label_list.txt' +WARNING 2022-02-14 09:24:42,704 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-14 09:24:42,709 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-14 09:24:42,709 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:24:42,710 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-14 09:24:42,710 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2022-02-14 09:24:42,710 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-14 09:24:42,710 [pipeline_server.py:212] ------------------------------------------- +INFO 2022-02-14 09:24:42,710 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-14 09:24:42,741 [dag.py:493] [DAG] Succ init +INFO 2022-02-14 09:24:42,741 [dag.py:651] ================= USED OP ================= +INFO 2022-02-14 09:24:42,741 [dag.py:654] ppyolo_mbv3 +INFO 2022-02-14 09:24:42,742 [dag.py:655] ------------------------------------------- +INFO 2022-02-14 09:24:42,783 [dag.py:784] [DAG] Succ build DAG +INFO 2022-02-14 09:24:42,789 [dag.py:816] [DAG] start +INFO 2022-02-14 09:24:42,790 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-14 09:24:42,796 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2022-02-14 09:24:42,815 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:24:42,815 [operator.py:1170] Init cuda env in process 0 +INFO 2022-02-14 09:24:42,815 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-14 09:24:44,053 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +CRITICAL 2022-02-14 09:24:45,275 [operator.py:1179] [ppyolo_mbv3|0] failed to init op: [Errno 2] No such file or directory: 'label_list.txt' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1174, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1363, in _initialize + self.init_op() + File "web_service.py", line 30, in init_op + self.img_postprocess = RCNNPostprocess("label_list.txt", "output") + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 288, in __init__ + with open(label_file) as fin: +FileNotFoundError: [Errno 2] No such file or directory: 'label_list.txt' +INFO 2022-02-14 09:24:57,166 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2022-02-14 09:24:57,168 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2022-02-14 09:24:57,169 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:26:03,675 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-14 09:26:03,676 [operator.py:163] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-14 09:26:03,676 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:26:03,676 [operator.py:267] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-14 09:26:03,676 [pipeline_server.py:204] ============= PIPELINE SERVER ============= +INFO 2022-02-14 09:26:03,677 [pipeline_server.py:207] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0 + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-14 09:26:03,677 [pipeline_server.py:212] ------------------------------------------- +INFO 2022-02-14 09:26:03,677 [operator.py:290] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-14 09:26:03,706 [dag.py:493] [DAG] Succ init +INFO 2022-02-14 09:26:03,707 [dag.py:651] ================= USED OP ================= +INFO 2022-02-14 09:26:03,707 [dag.py:654] ppyolo_mbv3 +INFO 2022-02-14 09:26:03,707 [dag.py:655] ------------------------------------------- +INFO 2022-02-14 09:26:03,747 [dag.py:784] [DAG] Succ build DAG +INFO 2022-02-14 09:26:03,752 [dag.py:816] [DAG] start +INFO 2022-02-14 09:26:03,753 [dag.py:181] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-14 09:26:03,761 [pipeline_server.py:47] [PipelineServicer] succ init +INFO 2022-02-14 09:26:03,776 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-14 09:26:03,777 [operator.py:1170] Init cuda env in process 0 +INFO 2022-02-14 09:26:03,777 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-14 09:26:04,993 [local_predict.py:115] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-14 09:26:05,574 [pipeline_server.py:51] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 +INFO 2022-02-14 09:26:05,576 [operator.py:1429] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction +INFO 2022-02-14 09:26:05,577 [dag.py:368] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-14 09:26:06,239 [operator.py:1181] [ppyolo_mbv3|0] Succ init +INFO 2022-02-14 09:26:07,900 [dag.py:404] (data_id=0 log_id=0) Succ predict +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-16 16:56:51,846 [operator.py:181] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-16 16:56:51,847 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 16:56:51,847 [operator.py:285] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-16 16:56:51,847 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-16 16:56:51,847 [pipeline_server.py:218] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "channel_recv_frist_arrive":false + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-16 16:56:51,847 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-16 16:56:51,847 [operator.py:308] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-16 16:56:51,869 [dag.py:496] [DAG] Succ init +INFO 2022-02-16 16:56:51,869 [dag.py:659] ================= USED OP ================= +INFO 2022-02-16 16:56:51,869 [dag.py:662] ppyolo_mbv3 +INFO 2022-02-16 16:56:51,870 [dag.py:663] ------------------------------------------- +INFO 2022-02-16 16:56:51,870 [dag.py:680] ================== DAG ==================== +INFO 2022-02-16 16:56:51,870 [dag.py:682] (VIEW 0) +INFO 2022-02-16 16:56:51,870 [dag.py:684] [@DAGExecutor] +INFO 2022-02-16 16:56:51,870 [dag.py:686] - ppyolo_mbv3 +INFO 2022-02-16 16:56:51,870 [dag.py:682] (VIEW 1) +INFO 2022-02-16 16:56:51,870 [dag.py:684] [ppyolo_mbv3] +INFO 2022-02-16 16:56:51,870 [dag.py:687] ------------------------------------------- +INFO 2022-02-16 16:56:51,885 [dag.py:730] op:ppyolo_mbv3 add input channel. +INFO 2022-02-16 16:56:51,895 [dag.py:759] last op:ppyolo_mbv3 add output channel +INFO 2022-02-16 16:56:51,895 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-16 16:56:51,899 [dag.py:832] [DAG] start +INFO 2022-02-16 16:56:51,899 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-16 16:56:51,905 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-16 16:56:51,911 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 16:56:51,912 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-16 16:56:51,912 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-16 16:56:52,885 [local_predict.py:153] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-16 16:56:55,000 [operator.py:1317] [ppyolo_mbv3|0] Succ init +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-16 17:05:23,154 [operator.py:181] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-16 17:05:23,154 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 17:05:23,154 [operator.py:285] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-16 17:05:23,154 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-16 17:05:23,155 [pipeline_server.py:218] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "channel_recv_frist_arrive":false + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-16 17:05:23,155 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-16 17:05:23,155 [operator.py:308] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-16 17:05:23,176 [dag.py:496] [DAG] Succ init +INFO 2022-02-16 17:05:23,177 [dag.py:659] ================= USED OP ================= +INFO 2022-02-16 17:05:23,177 [dag.py:662] ppyolo_mbv3 +INFO 2022-02-16 17:05:23,177 [dag.py:663] ------------------------------------------- +INFO 2022-02-16 17:05:23,177 [dag.py:680] ================== DAG ==================== +INFO 2022-02-16 17:05:23,177 [dag.py:682] (VIEW 0) +INFO 2022-02-16 17:05:23,177 [dag.py:684] [@DAGExecutor] +INFO 2022-02-16 17:05:23,177 [dag.py:686] - ppyolo_mbv3 +INFO 2022-02-16 17:05:23,177 [dag.py:682] (VIEW 1) +INFO 2022-02-16 17:05:23,177 [dag.py:684] [ppyolo_mbv3] +INFO 2022-02-16 17:05:23,177 [dag.py:687] ------------------------------------------- +INFO 2022-02-16 17:05:23,192 [dag.py:730] op:ppyolo_mbv3 add input channel. +INFO 2022-02-16 17:05:23,202 [dag.py:759] last op:ppyolo_mbv3 add output channel +INFO 2022-02-16 17:05:23,202 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-16 17:05:23,205 [dag.py:832] [DAG] start +INFO 2022-02-16 17:05:23,206 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-16 17:05:23,211 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-16 17:05:23,229 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 17:05:23,229 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-16 17:05:23,229 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-16 17:05:24,167 [local_predict.py:153] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-16 17:05:26,236 [operator.py:1317] [ppyolo_mbv3|0] Succ init +INFO 2022-02-16 17:05:47,771 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002347.7711458 +INFO 2022-02-16 17:05:47,772 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002347.7724555 +INFO 2022-02-16 17:05:47,772 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-16 17:05:50,421 [dag.py:405] (data_id=0 log_id=0) Succ predict +INFO 2022-02-16 17:05:50,447 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002350.4476814 +INFO 2022-02-16 17:05:50,448 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002350.448209 +INFO 2022-02-16 17:05:50,448 [dag.py:369] (data_id=1 log_id=0) Succ Generate ID +INFO 2022-02-16 17:05:50,527 [dag.py:405] (data_id=1 log_id=0) Succ predict +INFO 2022-02-16 17:07:00,293 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002420.2930179 +INFO 2022-02-16 17:07:00,293 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002420.2936485 +INFO 2022-02-16 17:07:00,293 [dag.py:369] (data_id=2 log_id=0) Succ Generate ID +INFO 2022-02-16 17:07:00,362 [dag.py:405] (data_id=2 log_id=0) Succ predict +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +INFO 2022-02-16 17:09:22,057 [operator.py:181] local_service_conf: {'client_type': 'local_predictor', 'device_type': 2, 'devices': '0', 'fetch_list': ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], 'model_config': 'serving_server/', 'workdir': '', 'thread_num': 2, 'mem_optim': True, 'ir_optim': False, 'precision': 'fp32', 'use_calib': False, 'use_mkldnn': False, 'mkldnn_cache_capacity': 0} +INFO 2022-02-16 17:09:22,058 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'], precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 17:09:22,058 [operator.py:285] ppyolo_mbv3 + input_ops: @DAGExecutor, + server_endpoints: None + fetch_list: ['save_infer_model/scale_0.tmp_1', 'save_infer_model/scale_1.tmp_1', 'save_infer_model/scale_2.tmp_1', 'save_infer_model/scale_3.tmp_1', 'save_infer_model/scale_4.tmp_1', 'save_infer_model/scale_5.tmp_1', 'save_infer_model/scale_6.tmp_1', 'save_infer_model/scale_7.tmp_1'] + client_config: serving_server/serving_server_conf.prototxt + concurrency: 1, + timeout(s): -1, + retry: 1, + batch_size: 1, + auto_batching_timeout(s): None +INFO 2022-02-16 17:09:22,058 [pipeline_server.py:215] ============= PIPELINE SERVER ============= +INFO 2022-02-16 17:09:22,058 [pipeline_server.py:218] +{ + "dag":{ + "is_thread_op":false, + "tracer":{ + "interval_s":30 + }, + "retry":1, + "client_type":"brpc", + "use_profile":false, + "channel_size":0, + "channel_recv_frist_arrive":false + }, + "http_port":2009, + "op":{ + "ppyolo_mbv3":{ + "concurrency":1, + "local_service_conf":{ + "client_type":"local_predictor", + "device_type":2, + "devices":"0", + "fetch_list":[ + "save_infer_model/scale_0.tmp_1", + "save_infer_model/scale_1.tmp_1", + "save_infer_model/scale_2.tmp_1", + "save_infer_model/scale_3.tmp_1", + "save_infer_model/scale_4.tmp_1", + "save_infer_model/scale_5.tmp_1", + "save_infer_model/scale_6.tmp_1", + "save_infer_model/scale_7.tmp_1" + ], + "model_config":"serving_server/", + "workdir":"", + "thread_num":2, + "mem_optim":true, + "ir_optim":false, + "precision":"fp32", + "use_calib":false, + "use_mkldnn":false, + "mkldnn_cache_capacity":0 + }, + "timeout":-1, + "retry":1, + "batch_size":1, + "auto_batching_timeout":-1 + } + }, + "rpc_port":9999, + "worker_num":20, + "build_dag_each_worker":false +} +INFO 2022-02-16 17:09:22,058 [pipeline_server.py:223] ------------------------------------------- +INFO 2022-02-16 17:09:22,058 [operator.py:308] Op(ppyolo_mbv3) use local rpc service at port: [] +INFO 2022-02-16 17:09:22,079 [dag.py:496] [DAG] Succ init +INFO 2022-02-16 17:09:22,080 [dag.py:659] ================= USED OP ================= +INFO 2022-02-16 17:09:22,080 [dag.py:662] ppyolo_mbv3 +INFO 2022-02-16 17:09:22,080 [dag.py:663] ------------------------------------------- +INFO 2022-02-16 17:09:22,080 [dag.py:680] ================== DAG ==================== +INFO 2022-02-16 17:09:22,080 [dag.py:682] (VIEW 0) +INFO 2022-02-16 17:09:22,080 [dag.py:684] [@DAGExecutor] +INFO 2022-02-16 17:09:22,080 [dag.py:686] - ppyolo_mbv3 +INFO 2022-02-16 17:09:22,081 [dag.py:682] (VIEW 1) +INFO 2022-02-16 17:09:22,081 [dag.py:684] [ppyolo_mbv3] +INFO 2022-02-16 17:09:22,081 [dag.py:687] ------------------------------------------- +INFO 2022-02-16 17:09:22,095 [dag.py:730] op:ppyolo_mbv3 add input channel. +INFO 2022-02-16 17:09:22,105 [dag.py:759] last op:ppyolo_mbv3 add output channel +INFO 2022-02-16 17:09:22,105 [dag.py:800] [DAG] Succ build DAG +INFO 2022-02-16 17:09:22,108 [dag.py:832] [DAG] start +INFO 2022-02-16 17:09:22,109 [dag.py:182] [DAG] set in channel succ, name [@DAGExecutor] +INFO 2022-02-16 17:09:22,114 [pipeline_server.py:51] [PipelineServicer] succ init +INFO 2022-02-16 17:09:22,119 [local_service_handler.py:172] Models(serving_server/) will be launched by device gpu. use_gpu:True, use_trt:True, use_lite:False, use_xpu:False, device_type:2, devices:[0], mem_optim:True, ir_optim:False, use_profile:False, thread_num:2, client_type:local_predictor, fetch_names:None, precision:fp32, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None +INFO 2022-02-16 17:09:22,120 [operator.py:1306] Init cuda env in process 0 +INFO 2022-02-16 17:09:22,120 [local_service_handler.py:208] GET_CLIENT : concurrency_idx=0, device_num=1 +INFO 2022-02-16 17:09:23,058 [local_predict.py:153] LocalPredictor load_model_config params: model_path:serving_server/, use_gpu:True, gpu_id:0, use_profile:False, thread_num:2, mem_optim:True, ir_optim:False, use_trt:True, use_lite:False, use_xpu:False, precision:fp32, use_calib:False, use_mkldnn:False, mkldnn_cache_capacity:0, mkldnn_op_list:None, mkldnn_bf16_op_list:None, use_feed_fetch_ops:False, +INFO 2022-02-16 17:09:24,980 [operator.py:1317] [ppyolo_mbv3|0] Succ init +INFO 2022-02-16 17:10:00,414 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002600.4143305 +INFO 2022-02-16 17:10:00,415 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002600.4155881 +INFO 2022-02-16 17:10:00,416 [dag.py:369] (data_id=0 log_id=0) Succ Generate ID +INFO 2022-02-16 17:10:02,981 [dag.py:405] (data_id=0 log_id=0) Succ predict +INFO 2022-02-16 17:14:52,096 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002892.0961268 +INFO 2022-02-16 17:14:52,096 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002892.096749 +INFO 2022-02-16 17:14:52,097 [dag.py:369] (data_id=1 log_id=0) Succ Generate ID +INFO 2022-02-16 17:14:52,172 [dag.py:405] (data_id=1 log_id=0) Succ predict +INFO 2022-02-16 17:15:06,391 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645002906.390965 +INFO 2022-02-16 17:15:06,391 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645002906.3916032 +INFO 2022-02-16 17:15:06,391 [dag.py:369] (data_id=2 log_id=0) Succ Generate ID +INFO 2022-02-16 17:15:06,459 [dag.py:405] (data_id=2 log_id=0) Succ predict +INFO 2022-02-16 17:21:25,074 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003285.0745604 +INFO 2022-02-16 17:21:25,075 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003285.075034 +INFO 2022-02-16 17:21:25,075 [dag.py:369] (data_id=3 log_id=0) Succ Generate ID +INFO 2022-02-16 17:21:25,141 [dag.py:405] (data_id=3 log_id=0) Succ predict +INFO 2022-02-16 17:21:53,963 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003313.9639173 +INFO 2022-02-16 17:21:53,964 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003313.9646337 +INFO 2022-02-16 17:21:53,964 [dag.py:369] (data_id=4 log_id=0) Succ Generate ID +INFO 2022-02-16 17:21:54,033 [dag.py:405] (data_id=4 log_id=0) Succ predict +INFO 2022-02-16 17:22:47,816 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003367.8165774 +INFO 2022-02-16 17:22:47,817 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003367.8172152 +INFO 2022-02-16 17:22:47,817 [dag.py:369] (data_id=5 log_id=0) Succ Generate ID +INFO 2022-02-16 17:22:47,886 [dag.py:405] (data_id=5 log_id=0) Succ predict +INFO 2022-02-16 17:23:03,882 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003383.882905 +INFO 2022-02-16 17:23:03,883 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003383.8835778 +INFO 2022-02-16 17:23:03,883 [dag.py:369] (data_id=6 log_id=0) Succ Generate ID +INFO 2022-02-16 17:23:03,951 [dag.py:405] (data_id=6 log_id=0) Succ predict +INFO 2022-02-16 17:23:30,593 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003410.5935636 +INFO 2022-02-16 17:23:30,594 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003410.5942206 +INFO 2022-02-16 17:23:30,594 [dag.py:369] (data_id=7 log_id=0) Succ Generate ID +INFO 2022-02-16 17:23:30,662 [dag.py:405] (data_id=7 log_id=0) Succ predict +INFO 2022-02-16 17:24:12,780 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003452.7805502 +INFO 2022-02-16 17:24:12,781 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003452.7814457 +INFO 2022-02-16 17:24:12,781 [dag.py:369] (data_id=8 log_id=0) Succ Generate ID +INFO 2022-02-16 17:24:12,853 [dag.py:405] (data_id=8 log_id=0) Succ predict +INFO 2022-02-16 17:24:23,797 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003463.7971623 +INFO 2022-02-16 17:24:23,797 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003463.7978337 +INFO 2022-02-16 17:24:23,798 [dag.py:369] (data_id=9 log_id=0) Succ Generate ID +INFO 2022-02-16 17:24:23,867 [dag.py:405] (data_id=9 log_id=0) Succ predict +INFO 2022-02-16 17:24:43,980 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003483.9801416 +INFO 2022-02-16 17:24:43,980 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003483.9806335 +INFO 2022-02-16 17:24:43,980 [dag.py:369] (data_id=10 log_id=0) Succ Generate ID +INFO 2022-02-16 17:24:44,049 [dag.py:405] (data_id=10 log_id=0) Succ predict +INFO 2022-02-16 17:24:54,159 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003494.15903 +INFO 2022-02-16 17:24:54,159 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003494.1595104 +INFO 2022-02-16 17:24:54,159 [dag.py:369] (data_id=11 log_id=0) Succ Generate ID +INFO 2022-02-16 17:24:54,226 [dag.py:405] (data_id=11 log_id=0) Succ predict +INFO 2022-02-16 17:25:12,386 [pipeline_server.py:56] (log_id=0) inference request name:ppyolo_mbv3 self.name:ppyolo_mbv3 time:1645003512.3861694 +INFO 2022-02-16 17:25:12,386 [operator.py:1723] RequestOp unpack one request. log_id:0, clientip: name:ppyolo_mbv3, method:prediction, time:1645003512.3868122 +INFO 2022-02-16 17:25:12,387 [dag.py:369] (data_id=12 log_id=0) Succ Generate ID +INFO 2022-02-16 17:25:12,455 [dag.py:405] (data_id=12 log_id=0) Succ predict diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log.wf b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log.wf new file mode 100644 index 000000000..377d4cf33 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.log.wf @@ -0,0 +1,2000 @@ +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 02:45:16,604 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 02:45:16,605 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 02:45:16,606 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 02:45:45,873 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 76, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 02:45:45,877 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:07:14,510 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:07:14,511 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:07:14,512 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 03:07:22,696 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:07:22,700 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:10:13,372 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:10:13,373 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:10:13,374 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 03:10:21,260 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 78, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:10:21,264 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 03:11:47,323 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 03:11:47,324 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 03:11:47,325 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 03:11:55,757 [operator.py:969] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 965, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 78, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 429, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 03:11:55,761 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:35:58,321 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:35:58,322 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:35:58,323 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:37:04,889 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:37:04,890 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:37:04,891 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 05:37:16,537 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:37:16,542 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:40:11,809 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:40:11,810 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:40:11,811 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:40:11,811 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 05:40:18,654 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:40:18,658 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:42:11,543 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 05:42:11,544 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 05:42:11,545 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 05:42:19,333 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 77, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 05:42:19,340 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:08:54,355 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:08:54,356 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:08:54,357 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:08:56,841 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'yaml' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 33, in init_op + yml_conf = yaml.safe_load(f) +NameError: name 'yaml' is not defined +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:10:19,803 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:10:19,804 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:10:19,805 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:10:22,339 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'yaml' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 33, in init_op + yml_conf = yaml.safe_load(f) +NameError: name 'yaml' is not defined +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:08,931 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:12:08,932 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:12:08,933 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:12:11,443 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 40, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:12:47,188 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:12:47,188 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:12:47,189 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:12:47,190 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:12:49,708 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 41, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:15:10,463 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:15:10,464 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:15:10,465 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:15:12,951 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 43, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:17:36,321 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:17:36,321 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:17:36,322 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:17:36,323 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:17:38,816 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 44, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:18:17,409 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:18:17,410 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:18:17,411 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:18:19,908 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: __init__() got an unexpected keyword argument 'interp' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 45, in init_op + self.preprocess_ops.append(eval(op_type)(**new_op_info)) +TypeError: __init__() got an unexpected keyword argument 'interp' +WARNING 2021-12-29 06:19:57,871 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:19:57,872 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:19:57,873 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:20:00,415 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 45, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:21:28,629 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:21:28,630 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:21:28,631 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:21:31,123 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 46, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:25:12,051 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:25:12,051 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:25:12,052 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:25:12,053 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:25:14,695 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 46, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:25:29,445 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:25:29,446 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:25:29,447 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:25:31,935 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 47, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:31:43,452 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:31:43,453 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:31:43,454 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2021-12-29 06:31:45,962 [operator.py:1176] [ppyolo_mbv3|0] failed to init op: name 'preprocess_ops' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1171, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1360, in _initialize + self.init_op() + File "web_service.py", line 49, in init_op + preprocess_ops.append(eval(op_type)(**new_op_info)) +NameError: name 'preprocess_ops' is not defined +WARNING 2021-12-29 06:32:31,020 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:32:31,021 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:32:31,022 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 06:33:05,247 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:33:05,250 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:33:05,251 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:40:19,257 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:40:19,258 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:40:19,259 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:40:25,105 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 54, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:40:25,111 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:03,381 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:42:03,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:42:13,106 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 54, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:42:13,112 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:42:44,174 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:42:44,175 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:42:44,176 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:42:47,195 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 55, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:42:47,200 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:44:34,233 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:44:34,234 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:44:34,235 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:44:43,710 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 55, in preprocess + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) +UnboundLocalError: local variable 'im_info' referenced before assignment +ERROR 2021-12-29 06:44:43,715 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: local variable 'im_info' referenced before assignment +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:46:19,030 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:46:19,031 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:46:19,032 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:46:25,581 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 71, in preprocess + im = self.img_preprocess(im) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 492, in __call__ + img = t(img) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 642, in __call__ + return F.normalize(img, self.mean, self.std, self.channel_first) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/functional.py", line 33, in normalize + img -= img_mean +ValueError: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +ERROR 2021-12-29 06:46:25,587 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: operands could not be broadcast together with shapes (3,640,640) (1,1,3) (3,640,640) +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:01,066 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:51:01,067 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:51:01,068 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:51:07,885 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 89, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:51:07,889 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:51:56,944 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:51:56,945 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:51:56,946 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 06:52:13,864 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:52:13,865 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:52:13,866 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:52:20,309 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 86, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:52:20,314 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:54:00,624 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 06:54:00,625 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 06:54:00,626 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 06:54:07,453 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 86, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 06:54:07,458 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:13:12,175 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:13:12,175 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:13:12,176 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:13:12,177 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:13:20,265 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 430, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:13:20,269 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:16:27,832 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:16:27,832 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:16:27,833 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:16:27,834 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:16:32,263 [operator.py:695] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: name 'im_shape' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 678, in _run_preprocess + parsed_data, data_id, logid_dict.get(data_id)) + File "web_service.py", line 68, in preprocess + "im_shape": im_info[im_shape],#np.array(list(im.shape[1:])).reshape(-1)[np.newaxis,:], +NameError: name 'im_shape' is not defined +ERROR 2021-12-29 07:16:32,267 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to preprocess: name 'im_shape' is not defined +WARNING 2021-12-29 07:17:33,514 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:17:33,514 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:17:33,515 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:17:33,516 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:17:39,797 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:17:39,802 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:19:06,141 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:19:06,142 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:19:06,143 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:19:12,763 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:19:12,767 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:40:25,237 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:40:25,238 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:40:25,239 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:42:11,629 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:42:11,630 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:42:11,631 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:42:25,683 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_0.tmp_1.lod' +ERROR 2021-12-29 07:42:25,687 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_0.tmp_1.lod' +WARNING 2021-12-29 07:49:58,432 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-29 07:49:58,433 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-29 07:49:58,434 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-29 07:50:05,801 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_4.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 87, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_4.tmp_1.lod' +ERROR 2021-12-29 07:50:05,806 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_4.tmp_1.lod' +WARNING 2021-12-30 06:53:16,236 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 06:53:16,237 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 06:53:16,238 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 06:53:28,471 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_3.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 81, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_3.tmp_1.lod' +ERROR 2021-12-30 06:53:28,476 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_3.tmp_1.lod' +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 07:57:06,805 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 07:57:06,806 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 07:57:06,807 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 07:57:06,807 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 07:57:15,249 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 81, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 07:57:15,253 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:12:07,133 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:12:07,134 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:12:14,700 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 82, in postprocess + np_score_list.append(fetch_dict[out_idx]) +KeyError: 0 +ERROR 2021-12-30 08:12:14,707 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:13:45,672 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:13:45,673 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:13:45,674 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:13:52,217 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[out_idx]) +KeyError: 0 +ERROR 2021-12-30 08:13:52,220 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:17:15,481 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:17:15,482 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:17:22,212 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:17:22,217 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:17:30,895 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:17:30,898 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:17:30,899 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:17:39,176 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 83, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:17:39,180 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:20:13,195 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:20:13,196 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:20:13,197 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:20:20,892 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 84, in postprocess + np_score_list.append(fetch_dict[i]) +KeyError: 0 +ERROR 2021-12-30 08:20:20,896 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 0 +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:31:54,772 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:31:54,773 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:31:54,774 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:32:01,331 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 91, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:32:01,335 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:34:21,409 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:34:21,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:34:21,410 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:34:21,411 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:34:28,756 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 96, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:34:28,761 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:43:32,266 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:43:32,266 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:43:32,267 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:43:32,268 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:43:32,269 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:43:32,270 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:43:46,213 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 111, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:43:46,218 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:49:32,281 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:49:32,281 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:49:32,282 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:49:32,283 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:49:32,284 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:49:32,285 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:49:39,879 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 113, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:49:39,884 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:52:42,378 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:52:42,379 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:52:42,380 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:52:42,381 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:52:42,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 08:52:50,006 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 110, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 08:52:50,011 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:59:43,422 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 08:59:43,423 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 08:59:43,424 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 08:59:43,425 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 08:59:43,426 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-30 09:03:09,956 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:03:09,956 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:03:09,957 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:03:09,958 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:03:09,959 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:11,902 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:03:11,903 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:03:11,904 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:03:11,905 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:03:22,048 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 09:03:22,053 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 09:07:27,378 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:07:27,379 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:27,380 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:07:27,381 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:07:27,382 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-30 09:07:48,750 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:07:48,751 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:07:48,752 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:07:48,753 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:07:48,754 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:07:48,754 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:07:53,007 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 431, in __call__ + self.clsid2catid) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 346, in _get_bbox_result + lod = [fetch_map[fetch_name + '.lod']] +KeyError: 'save_infer_model/scale_7.tmp_1.lod' +ERROR 2021-12-30 09:07:53,011 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: 'save_infer_model/scale_7.tmp_1.lod' +WARNING 2021-12-30 09:08:06,502 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:08:06,503 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:08:06,504 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:08:06,505 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:08:06,506 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:08:11,524 [operator.py:1000] (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +ERROR 2021-12-30 09:08:11,527 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +WARNING 2021-12-30 09:19:39,408 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:19:39,409 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:19:39,410 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:19:39,411 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:19:39,412 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:19:39,412 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:19:46,754 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'res_dict' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 101, in postprocess + res_dict[b] = {} +NameError: name 'res_dict' is not defined +ERROR 2021-12-30 09:19:46,757 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'res_dict' is not defined +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:20:11,181 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:20:11,182 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:20:11,183 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:20:15,417 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 102, in postprocess + res_dict[b] = {} +IndexError: list assignment index out of range +ERROR 2021-12-30 09:20:15,421 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +WARNING 2021-12-30 09:21:19,542 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:21:19,543 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:21:19,544 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:21:19,545 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:21:19,546 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:21:24,326 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 103, in postprocess + res_dict[b] = {} +IndexError: list assignment index out of range +ERROR 2021-12-30 09:21:24,330 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: list assignment index out of range +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:23:31,824 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:23:31,825 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:23:31,826 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:23:31,827 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:23:37,722 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'a' is not defined +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 101, in postprocess + for b in range(a.ndim): +NameError: name 'a' is not defined +ERROR 2021-12-30 09:23:37,726 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: name 'a' is not defined +WARNING 2021-12-30 09:24:02,842 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:24:02,842 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:24:02,843 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:24:02,844 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:24:02,845 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:24:02,846 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:24:07,046 [operator.py:1000] (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +ERROR 2021-12-30 09:24:07,049 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (log_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: output of postprocess funticon must be dict type, but get +WARNING 2021-12-30 09:25:54,030 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:25:54,031 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:25:54,032 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:25:54,033 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:25:54,034 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:25:54,034 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:25:58,334 [operator.py:1487] (logid=0) Failed to pack RPC response package: +WARNING 2021-12-30 09:39:07,135 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:39:07,136 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:39:07,137 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:39:07,138 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:39:07,139 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:40:00,249 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:00,253 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:48,711 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:48,714 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:53,720 [operator.py:973] (data_id=3 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:40:53,723 [dag.py:409] (data_id=3 log_id=0) Failed to predict: (data_id=3 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:42:01,778 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:42:01,779 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:42:01,780 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:42:01,781 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:42:01,782 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:42:06,612 [operator.py:973] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 09:42:06,616 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 09:42:19,953 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 09:42:19,957 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:42:23,963 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:42:23,964 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:42:23,965 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:42:23,966 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:42:23,967 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:43:10,985 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:43:10,988 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:43:15,944 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:43:15,950 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:44:54,821 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:44:54,822 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:44:54,823 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:44:54,824 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:44:54,825 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:45:05,468 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:45:05,472 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 09:46:17,679 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 09:46:17,680 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 09:46:17,681 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 09:46:17,682 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 09:46:17,683 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 09:46:17,683 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 09:46:25,584 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 09:46:25,588 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:05:40,432 [pipeline_server.py:55] (log_id=0) name dismatch error. request.name:recognition,server.name=ppyolo_mbv3 +ERROR 2021-12-30 11:05:49,271 [operator.py:973] (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:05:49,274 [dag.py:409] (data_id=2 log_id=0) Failed to predict: (data_id=2 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:05:54,559 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:05:54,560 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:05:54,561 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:05:54,562 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:05:54,563 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 11:15:50,225 [operator.py:973] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 969, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:15:50,229 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:24:24,428 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:24:24,429 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:24:24,430 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:24:24,431 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 11:24:40,005 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) +TypeError: __call__() takes 3 positional arguments but 4 were given +ERROR 2021-12-30 11:24:40,009 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: __call__() takes 3 positional arguments but 4 were given +WARNING 2021-12-30 11:37:49,610 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:37:49,611 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:37:49,612 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:37:49,613 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:37:49,614 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 11:37:54,033 [operator.py:976] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:37:54,039 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:37:55,960 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:37:55,963 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 11:38:09,257 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:38:09,258 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:38:09,259 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:38:09,260 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:38:09,261 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +ERROR 2021-12-30 11:38:13,407 [operator.py:976] (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + np_boxes, np_boxes_num = self.postprocess(np_score_list, np_boxes_list) +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:38:13,411 [dag.py:409] (data_id=0 log_id=0) Failed to predict: (data_id=0 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:40:01,379 [operator.py:976] (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 972, in _run_postprocess + logid_dict.get(data_id)) + File "web_service.py", line 92, in postprocess + self.post_process = PicoDetPostProcess( +TypeError: postprocess() missing 1 required positional argument: 'log_id' +ERROR 2021-12-30 11:40:01,383 [dag.py:409] (data_id=1 log_id=0) Failed to predict: (data_id=1 log_id=0) [ppyolo_mbv3|0] Failed to postprocess: postprocess() missing 1 required positional argument: 'log_id' +WARNING 2021-12-30 11:40:05,557 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:40:05,557 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:40:05,558 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:40:05,559 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:40:05,560 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2021-12-30 11:41:26,917 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2021-12-30 11:41:26,918 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2021-12-30 11:41:26,919 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2021-12-30 11:41:26,920 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2021-12-30 11:41:26,921 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-14 09:24:33,065 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:24:33,072 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:24:33,073 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:24:33,074 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2022-02-14 09:24:35,602 [operator.py:1179] [ppyolo_mbv3|0] failed to init op: [Errno 2] No such file or directory: 'label_list.txt' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1174, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1363, in _initialize + self.init_op() + File "web_service.py", line 30, in init_op + self.img_postprocess = RCNNPostprocess("label_list.txt", "output") + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 288, in __init__ + with open(label_file) as fin: +FileNotFoundError: [Errno 2] No such file or directory: 'label_list.txt' +WARNING 2022-02-14 09:24:42,704 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:24:42,705 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:24:42,706 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:24:42,707 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:24:42,708 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +CRITICAL 2022-02-14 09:24:45,275 [operator.py:1179] [ppyolo_mbv3|0] failed to init op: [Errno 2] No such file or directory: 'label_list.txt' +Traceback (most recent call last): + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1174, in _run + profiler = self._initialize(is_thread_op, concurrency_idx) + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_server/pipeline/operator.py", line 1363, in _initialize + self.init_op() + File "web_service.py", line 30, in init_op + self.img_postprocess = RCNNPostprocess("label_list.txt", "output") + File "/usr/local/python3.7.0/lib/python3.7/site-packages/paddle_serving_app/reader/image_reader.py", line 288, in __init__ + with open(label_file) as fin: +FileNotFoundError: [Errno 2] No such file or directory: 'label_list.txt' +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:26:03,671 [pipeline_server.py:496] [CONF] client_type not set, use default: brpc +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] use_profile not set, use default: False +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] timeout not set, use default: -1 +WARNING 2022-02-14 09:26:03,672 [pipeline_server.py:496] [CONF] retry not set, use default: 1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] workdir not set, use default: +WARNING 2022-02-14 09:26:03,673 [pipeline_server.py:496] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] mem_optim not set, use default: True +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] ir_optim not set, use default: False +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] precision not set, use default: fp32 +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] use_calib not set, use default: False +WARNING 2022-02-14 09:26:03,674 [pipeline_server.py:496] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-14 09:26:03,675 [pipeline_server.py:496] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 16:56:51,836 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 16:56:51,837 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 17:05:23,144 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 17:05:23,145 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] build_dag_each_worker not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] client_type not set, use default: brpc +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] use_profile not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] channel_size not set, use default: 0 +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] channel_recv_frist_arrive not set, use default: False +WARNING 2022-02-16 17:09:22,047 [pipeline_server.py:509] [CONF] timeout not set, use default: -1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] retry not set, use default: 1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] batch_size not set, use default: 1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] auto_batching_timeout not set, use default: -1 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] workdir not set, use default: +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] thread_num not set, use default: 2 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] mem_optim not set, use default: True +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] ir_optim not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] precision not set, use default: fp32 +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] use_calib not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] use_mkldnn not set, use default: False +WARNING 2022-02-16 17:09:22,048 [pipeline_server.py:509] [CONF] mkldnn_cache_capacity not set, use default: 0 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.tracer b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.tracer new file mode 100644 index 000000000..e32c459a2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/PipelineServingLogs/pipeline.tracer @@ -0,0 +1,6853 @@ +2021-12-29 02:45:16,713 ==================== TRACER ====================== +2021-12-29 02:45:16,715 Channel (server worker num[20]): +2021-12-29 02:45:16,717 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:45:16,718 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:45:46,748 ==================== TRACER ====================== +2021-12-29 02:45:46,749 DAGExecutor: +2021-12-29 02:45:46,749 Query count[1] +2021-12-29 02:45:46,749 QPS[0.03333333333333333 q/s] +2021-12-29 02:45:46,750 Succ[0.0] +2021-12-29 02:45:46,750 Error req[0] +2021-12-29 02:45:46,750 Latency: +2021-12-29 02:45:46,750 ave[1691.297 ms] +2021-12-29 02:45:46,750 .50[1691.297 ms] +2021-12-29 02:45:46,751 .60[1691.297 ms] +2021-12-29 02:45:46,751 .70[1691.297 ms] +2021-12-29 02:45:46,751 .80[1691.297 ms] +2021-12-29 02:45:46,751 .90[1691.297 ms] +2021-12-29 02:45:46,751 .95[1691.297 ms] +2021-12-29 02:45:46,752 .99[1691.297 ms] +2021-12-29 02:45:46,752 Channel (server worker num[20]): +2021-12-29 02:45:46,753 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:45:46,753 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:46:16,784 ==================== TRACER ====================== +2021-12-29 02:46:16,785 Channel (server worker num[20]): +2021-12-29 02:46:16,786 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:46:16,786 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:46:46,817 ==================== TRACER ====================== +2021-12-29 02:46:46,818 Channel (server worker num[20]): +2021-12-29 02:46:46,818 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:46:46,819 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:47:16,849 ==================== TRACER ====================== +2021-12-29 02:47:16,850 Channel (server worker num[20]): +2021-12-29 02:47:16,851 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:47:16,852 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:47:46,882 ==================== TRACER ====================== +2021-12-29 02:47:46,883 Channel (server worker num[20]): +2021-12-29 02:47:46,884 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:47:46,884 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:48:16,888 ==================== TRACER ====================== +2021-12-29 02:48:16,889 Channel (server worker num[20]): +2021-12-29 02:48:16,890 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:48:16,891 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:48:46,916 ==================== TRACER ====================== +2021-12-29 02:48:46,917 Channel (server worker num[20]): +2021-12-29 02:48:46,918 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:48:46,919 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:49:16,949 ==================== TRACER ====================== +2021-12-29 02:49:16,950 Channel (server worker num[20]): +2021-12-29 02:49:16,951 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:49:16,951 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:49:46,982 ==================== TRACER ====================== +2021-12-29 02:49:46,982 Channel (server worker num[20]): +2021-12-29 02:49:46,983 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:49:46,984 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:50:17,014 ==================== TRACER ====================== +2021-12-29 02:50:17,015 Channel (server worker num[20]): +2021-12-29 02:50:17,016 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:50:17,017 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:50:47,047 ==================== TRACER ====================== +2021-12-29 02:50:47,048 Channel (server worker num[20]): +2021-12-29 02:50:47,049 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:50:47,049 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:51:17,080 ==================== TRACER ====================== +2021-12-29 02:51:17,080 Channel (server worker num[20]): +2021-12-29 02:51:17,081 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:51:17,082 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:51:47,112 ==================== TRACER ====================== +2021-12-29 02:51:47,113 Channel (server worker num[20]): +2021-12-29 02:51:47,114 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:51:47,114 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:52:17,145 ==================== TRACER ====================== +2021-12-29 02:52:17,146 Channel (server worker num[20]): +2021-12-29 02:52:17,146 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:52:17,147 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:52:47,177 ==================== TRACER ====================== +2021-12-29 02:52:47,178 Channel (server worker num[20]): +2021-12-29 02:52:47,179 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:52:47,179 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:53:17,210 ==================== TRACER ====================== +2021-12-29 02:53:17,211 Channel (server worker num[20]): +2021-12-29 02:53:17,211 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:53:17,212 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:53:47,243 ==================== TRACER ====================== +2021-12-29 02:53:47,243 Channel (server worker num[20]): +2021-12-29 02:53:47,244 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:53:47,245 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:54:17,275 ==================== TRACER ====================== +2021-12-29 02:54:17,276 Channel (server worker num[20]): +2021-12-29 02:54:17,277 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:54:17,278 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:54:47,308 ==================== TRACER ====================== +2021-12-29 02:54:47,309 Channel (server worker num[20]): +2021-12-29 02:54:47,310 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:54:47,310 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:55:17,341 ==================== TRACER ====================== +2021-12-29 02:55:17,342 Channel (server worker num[20]): +2021-12-29 02:55:17,342 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:55:17,343 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:55:47,373 ==================== TRACER ====================== +2021-12-29 02:55:47,374 Channel (server worker num[20]): +2021-12-29 02:55:47,375 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:55:47,376 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:56:17,406 ==================== TRACER ====================== +2021-12-29 02:56:17,407 Channel (server worker num[20]): +2021-12-29 02:56:17,408 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:56:17,409 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:56:47,439 ==================== TRACER ====================== +2021-12-29 02:56:47,440 Channel (server worker num[20]): +2021-12-29 02:56:47,441 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:56:47,441 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:57:17,453 ==================== TRACER ====================== +2021-12-29 02:57:17,454 Channel (server worker num[20]): +2021-12-29 02:57:17,455 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:57:17,456 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:57:47,486 ==================== TRACER ====================== +2021-12-29 02:57:47,487 Channel (server worker num[20]): +2021-12-29 02:57:47,488 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:57:47,489 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:58:17,519 ==================== TRACER ====================== +2021-12-29 02:58:17,520 Channel (server worker num[20]): +2021-12-29 02:58:17,521 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:58:17,521 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:58:47,552 ==================== TRACER ====================== +2021-12-29 02:58:47,553 Channel (server worker num[20]): +2021-12-29 02:58:47,553 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:58:47,554 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:59:17,584 ==================== TRACER ====================== +2021-12-29 02:59:17,585 Channel (server worker num[20]): +2021-12-29 02:59:17,586 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:59:17,587 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 02:59:47,617 ==================== TRACER ====================== +2021-12-29 02:59:47,618 Channel (server worker num[20]): +2021-12-29 02:59:47,619 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 02:59:47,620 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:00:17,650 ==================== TRACER ====================== +2021-12-29 03:00:17,651 Channel (server worker num[20]): +2021-12-29 03:00:17,652 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:00:17,652 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:00:47,683 ==================== TRACER ====================== +2021-12-29 03:00:47,684 Channel (server worker num[20]): +2021-12-29 03:00:47,684 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:00:47,685 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:01:17,715 ==================== TRACER ====================== +2021-12-29 03:01:17,716 Channel (server worker num[20]): +2021-12-29 03:01:17,717 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:01:17,718 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:01:47,748 ==================== TRACER ====================== +2021-12-29 03:01:47,749 Channel (server worker num[20]): +2021-12-29 03:01:47,750 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:01:47,750 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:02:17,781 ==================== TRACER ====================== +2021-12-29 03:02:17,782 Channel (server worker num[20]): +2021-12-29 03:02:17,782 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:02:17,783 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:02:47,813 ==================== TRACER ====================== +2021-12-29 03:02:47,814 Channel (server worker num[20]): +2021-12-29 03:02:47,815 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:02:47,816 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:03:17,846 ==================== TRACER ====================== +2021-12-29 03:03:17,847 Channel (server worker num[20]): +2021-12-29 03:03:17,848 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:03:17,848 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:03:47,879 ==================== TRACER ====================== +2021-12-29 03:03:47,879 Channel (server worker num[20]): +2021-12-29 03:03:47,880 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:03:47,881 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:04:17,911 ==================== TRACER ====================== +2021-12-29 03:04:17,912 Channel (server worker num[20]): +2021-12-29 03:04:17,913 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:04:17,914 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:04:47,944 ==================== TRACER ====================== +2021-12-29 03:04:47,945 Channel (server worker num[20]): +2021-12-29 03:04:47,946 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:04:47,947 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:05:17,965 ==================== TRACER ====================== +2021-12-29 03:05:17,966 Channel (server worker num[20]): +2021-12-29 03:05:17,967 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:05:17,968 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:05:47,998 ==================== TRACER ====================== +2021-12-29 03:05:47,999 Channel (server worker num[20]): +2021-12-29 03:05:48,000 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:05:48,001 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:06:18,031 ==================== TRACER ====================== +2021-12-29 03:06:18,032 Channel (server worker num[20]): +2021-12-29 03:06:18,033 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:06:18,033 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:06:48,064 ==================== TRACER ====================== +2021-12-29 03:06:48,065 Channel (server worker num[20]): +2021-12-29 03:06:48,065 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:06:48,066 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:07:14,621 ==================== TRACER ====================== +2021-12-29 03:07:14,623 Channel (server worker num[20]): +2021-12-29 03:07:14,625 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:07:14,625 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:07:44,656 ==================== TRACER ====================== +2021-12-29 03:07:44,657 DAGExecutor: +2021-12-29 03:07:44,657 Query count[1] +2021-12-29 03:07:44,657 QPS[0.03333333333333333 q/s] +2021-12-29 03:07:44,657 Succ[0.0] +2021-12-29 03:07:44,658 Error req[0] +2021-12-29 03:07:44,658 Latency: +2021-12-29 03:07:44,658 ave[1819.424 ms] +2021-12-29 03:07:44,658 .50[1819.424 ms] +2021-12-29 03:07:44,658 .60[1819.424 ms] +2021-12-29 03:07:44,659 .70[1819.424 ms] +2021-12-29 03:07:44,659 .80[1819.424 ms] +2021-12-29 03:07:44,659 .90[1819.424 ms] +2021-12-29 03:07:44,659 .95[1819.424 ms] +2021-12-29 03:07:44,659 .99[1819.424 ms] +2021-12-29 03:07:44,659 Channel (server worker num[20]): +2021-12-29 03:07:44,660 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:07:44,661 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:08:14,691 ==================== TRACER ====================== +2021-12-29 03:08:14,692 Channel (server worker num[20]): +2021-12-29 03:08:14,693 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:08:14,694 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:08:44,724 ==================== TRACER ====================== +2021-12-29 03:08:44,725 Channel (server worker num[20]): +2021-12-29 03:08:44,726 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:08:44,727 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:09:14,757 ==================== TRACER ====================== +2021-12-29 03:09:14,758 Channel (server worker num[20]): +2021-12-29 03:09:14,759 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:09:14,760 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:09:44,776 ==================== TRACER ====================== +2021-12-29 03:09:44,777 Channel (server worker num[20]): +2021-12-29 03:09:44,778 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:09:44,779 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:10:13,471 ==================== TRACER ====================== +2021-12-29 03:10:13,473 Channel (server worker num[20]): +2021-12-29 03:10:13,475 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:10:13,476 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:10:43,506 ==================== TRACER ====================== +2021-12-29 03:10:43,508 DAGExecutor: +2021-12-29 03:10:43,508 Query count[1] +2021-12-29 03:10:43,508 QPS[0.03333333333333333 q/s] +2021-12-29 03:10:43,508 Succ[0.0] +2021-12-29 03:10:43,508 Error req[0] +2021-12-29 03:10:43,509 Latency: +2021-12-29 03:10:43,509 ave[1855.084 ms] +2021-12-29 03:10:43,509 .50[1855.084 ms] +2021-12-29 03:10:43,509 .60[1855.084 ms] +2021-12-29 03:10:43,509 .70[1855.084 ms] +2021-12-29 03:10:43,509 .80[1855.084 ms] +2021-12-29 03:10:43,510 .90[1855.084 ms] +2021-12-29 03:10:43,510 .95[1855.084 ms] +2021-12-29 03:10:43,510 .99[1855.084 ms] +2021-12-29 03:10:43,510 Channel (server worker num[20]): +2021-12-29 03:10:43,511 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:10:43,512 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:11:13,542 ==================== TRACER ====================== +2021-12-29 03:11:13,543 Channel (server worker num[20]): +2021-12-29 03:11:13,544 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:11:13,544 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:11:43,560 ==================== TRACER ====================== +2021-12-29 03:11:43,561 Channel (server worker num[20]): +2021-12-29 03:11:43,562 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:11:43,562 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:11:47,437 ==================== TRACER ====================== +2021-12-29 03:11:47,438 Channel (server worker num[20]): +2021-12-29 03:11:47,440 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:11:47,441 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:12:17,457 ==================== TRACER ====================== +2021-12-29 03:12:17,458 DAGExecutor: +2021-12-29 03:12:17,459 Query count[1] +2021-12-29 03:12:17,459 QPS[0.03333333333333333 q/s] +2021-12-29 03:12:17,459 Succ[0.0] +2021-12-29 03:12:17,459 Error req[0] +2021-12-29 03:12:17,459 Latency: +2021-12-29 03:12:17,460 ave[1822.881 ms] +2021-12-29 03:12:17,460 .50[1822.881 ms] +2021-12-29 03:12:17,460 .60[1822.881 ms] +2021-12-29 03:12:17,460 .70[1822.881 ms] +2021-12-29 03:12:17,460 .80[1822.881 ms] +2021-12-29 03:12:17,461 .90[1822.881 ms] +2021-12-29 03:12:17,461 .95[1822.881 ms] +2021-12-29 03:12:17,461 .99[1822.881 ms] +2021-12-29 03:12:17,461 Channel (server worker num[20]): +2021-12-29 03:12:17,462 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:12:17,463 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:12:47,493 ==================== TRACER ====================== +2021-12-29 03:12:47,494 Channel (server worker num[20]): +2021-12-29 03:12:47,495 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:12:47,495 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:13:17,502 ==================== TRACER ====================== +2021-12-29 03:13:17,502 Channel (server worker num[20]): +2021-12-29 03:13:17,503 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:13:17,504 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:13:47,534 ==================== TRACER ====================== +2021-12-29 03:13:47,535 Channel (server worker num[20]): +2021-12-29 03:13:47,536 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:13:47,537 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:14:17,567 ==================== TRACER ====================== +2021-12-29 03:14:17,568 Channel (server worker num[20]): +2021-12-29 03:14:17,569 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:14:17,570 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:14:47,595 ==================== TRACER ====================== +2021-12-29 03:14:47,596 Channel (server worker num[20]): +2021-12-29 03:14:47,597 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:14:47,598 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:15:17,628 ==================== TRACER ====================== +2021-12-29 03:15:17,629 Channel (server worker num[20]): +2021-12-29 03:15:17,630 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:15:17,631 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:15:47,661 ==================== TRACER ====================== +2021-12-29 03:15:47,662 Channel (server worker num[20]): +2021-12-29 03:15:47,663 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:15:47,663 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:16:17,694 ==================== TRACER ====================== +2021-12-29 03:16:17,695 Channel (server worker num[20]): +2021-12-29 03:16:17,695 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:16:17,696 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:16:47,707 ==================== TRACER ====================== +2021-12-29 03:16:47,707 Channel (server worker num[20]): +2021-12-29 03:16:47,708 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:16:47,709 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:17:17,739 ==================== TRACER ====================== +2021-12-29 03:17:17,740 Channel (server worker num[20]): +2021-12-29 03:17:17,741 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:17:17,742 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:17:47,769 ==================== TRACER ====================== +2021-12-29 03:17:47,770 Channel (server worker num[20]): +2021-12-29 03:17:47,770 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:17:47,771 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:18:17,801 ==================== TRACER ====================== +2021-12-29 03:18:17,802 Channel (server worker num[20]): +2021-12-29 03:18:17,803 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:18:17,804 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:18:47,834 ==================== TRACER ====================== +2021-12-29 03:18:47,835 Channel (server worker num[20]): +2021-12-29 03:18:47,836 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:18:47,837 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:19:17,846 ==================== TRACER ====================== +2021-12-29 03:19:17,847 Channel (server worker num[20]): +2021-12-29 03:19:17,848 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:19:17,848 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:19:47,879 ==================== TRACER ====================== +2021-12-29 03:19:47,880 Channel (server worker num[20]): +2021-12-29 03:19:47,881 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:19:47,881 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:20:17,912 ==================== TRACER ====================== +2021-12-29 03:20:17,912 Channel (server worker num[20]): +2021-12-29 03:20:17,913 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:20:17,914 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:20:47,944 ==================== TRACER ====================== +2021-12-29 03:20:47,945 Channel (server worker num[20]): +2021-12-29 03:20:47,946 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:20:47,947 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:21:17,977 ==================== TRACER ====================== +2021-12-29 03:21:17,978 Channel (server worker num[20]): +2021-12-29 03:21:17,978 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:21:17,979 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:21:48,009 ==================== TRACER ====================== +2021-12-29 03:21:48,010 Channel (server worker num[20]): +2021-12-29 03:21:48,011 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:21:48,012 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:22:18,042 ==================== TRACER ====================== +2021-12-29 03:22:18,043 Channel (server worker num[20]): +2021-12-29 03:22:18,044 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:22:18,045 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:22:48,075 ==================== TRACER ====================== +2021-12-29 03:22:48,076 Channel (server worker num[20]): +2021-12-29 03:22:48,076 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:22:48,077 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:23:18,107 ==================== TRACER ====================== +2021-12-29 03:23:18,108 Channel (server worker num[20]): +2021-12-29 03:23:18,109 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:23:18,110 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:23:48,140 ==================== TRACER ====================== +2021-12-29 03:23:48,141 Channel (server worker num[20]): +2021-12-29 03:23:48,142 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:23:48,143 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:24:18,173 ==================== TRACER ====================== +2021-12-29 03:24:18,174 Channel (server worker num[20]): +2021-12-29 03:24:18,175 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:24:18,175 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:24:48,206 ==================== TRACER ====================== +2021-12-29 03:24:48,206 Channel (server worker num[20]): +2021-12-29 03:24:48,207 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:24:48,208 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:25:18,238 ==================== TRACER ====================== +2021-12-29 03:25:18,239 Channel (server worker num[20]): +2021-12-29 03:25:18,242 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:25:18,242 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:25:48,273 ==================== TRACER ====================== +2021-12-29 03:25:48,273 Channel (server worker num[20]): +2021-12-29 03:25:48,274 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:25:48,275 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:26:18,305 ==================== TRACER ====================== +2021-12-29 03:26:18,306 Channel (server worker num[20]): +2021-12-29 03:26:18,307 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:26:18,308 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:26:48,338 ==================== TRACER ====================== +2021-12-29 03:26:48,339 Channel (server worker num[20]): +2021-12-29 03:26:48,340 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:26:48,340 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:27:18,371 ==================== TRACER ====================== +2021-12-29 03:27:18,371 Channel (server worker num[20]): +2021-12-29 03:27:18,372 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:27:18,373 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:27:48,403 ==================== TRACER ====================== +2021-12-29 03:27:48,404 Channel (server worker num[20]): +2021-12-29 03:27:48,405 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:27:48,406 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:28:18,436 ==================== TRACER ====================== +2021-12-29 03:28:18,437 Channel (server worker num[20]): +2021-12-29 03:28:18,438 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:28:18,438 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:28:48,469 ==================== TRACER ====================== +2021-12-29 03:28:48,470 Channel (server worker num[20]): +2021-12-29 03:28:48,470 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:28:48,471 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:29:18,501 ==================== TRACER ====================== +2021-12-29 03:29:18,502 Channel (server worker num[20]): +2021-12-29 03:29:18,503 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:29:18,504 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:29:48,534 ==================== TRACER ====================== +2021-12-29 03:29:48,535 Channel (server worker num[20]): +2021-12-29 03:29:48,536 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:29:48,536 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:30:18,567 ==================== TRACER ====================== +2021-12-29 03:30:18,568 Channel (server worker num[20]): +2021-12-29 03:30:18,568 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:30:18,569 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:30:48,600 ==================== TRACER ====================== +2021-12-29 03:30:48,600 Channel (server worker num[20]): +2021-12-29 03:30:48,601 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:30:48,602 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:31:18,632 ==================== TRACER ====================== +2021-12-29 03:31:18,633 Channel (server worker num[20]): +2021-12-29 03:31:18,634 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:31:18,634 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:31:48,665 ==================== TRACER ====================== +2021-12-29 03:31:48,665 Channel (server worker num[20]): +2021-12-29 03:31:48,666 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:31:48,667 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:32:18,697 ==================== TRACER ====================== +2021-12-29 03:32:18,698 Channel (server worker num[20]): +2021-12-29 03:32:18,699 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:32:18,700 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:32:48,730 ==================== TRACER ====================== +2021-12-29 03:32:48,731 Channel (server worker num[20]): +2021-12-29 03:32:48,731 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:32:48,732 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:33:18,762 ==================== TRACER ====================== +2021-12-29 03:33:18,763 Channel (server worker num[20]): +2021-12-29 03:33:18,764 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:33:18,765 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:33:48,776 ==================== TRACER ====================== +2021-12-29 03:33:48,777 Channel (server worker num[20]): +2021-12-29 03:33:48,778 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:33:48,779 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:34:18,809 ==================== TRACER ====================== +2021-12-29 03:34:18,810 Channel (server worker num[20]): +2021-12-29 03:34:18,810 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:34:18,811 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:34:48,842 ==================== TRACER ====================== +2021-12-29 03:34:48,842 Channel (server worker num[20]): +2021-12-29 03:34:48,843 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:34:48,844 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:35:18,874 ==================== TRACER ====================== +2021-12-29 03:35:18,875 Channel (server worker num[20]): +2021-12-29 03:35:18,876 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:35:18,877 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:35:48,907 ==================== TRACER ====================== +2021-12-29 03:35:48,908 Channel (server worker num[20]): +2021-12-29 03:35:48,909 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:35:48,909 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:36:18,940 ==================== TRACER ====================== +2021-12-29 03:36:18,941 Channel (server worker num[20]): +2021-12-29 03:36:18,941 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:36:18,942 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:36:48,965 ==================== TRACER ====================== +2021-12-29 03:36:48,965 Channel (server worker num[20]): +2021-12-29 03:36:48,966 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:36:48,967 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:37:18,997 ==================== TRACER ====================== +2021-12-29 03:37:18,998 Channel (server worker num[20]): +2021-12-29 03:37:18,999 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:37:19,000 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:37:49,030 ==================== TRACER ====================== +2021-12-29 03:37:49,031 Channel (server worker num[20]): +2021-12-29 03:37:49,031 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:37:49,032 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:38:19,035 ==================== TRACER ====================== +2021-12-29 03:38:19,036 Channel (server worker num[20]): +2021-12-29 03:38:19,037 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:38:19,037 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:38:49,068 ==================== TRACER ====================== +2021-12-29 03:38:49,069 Channel (server worker num[20]): +2021-12-29 03:38:49,069 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:38:49,070 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:39:19,100 ==================== TRACER ====================== +2021-12-29 03:39:19,101 Channel (server worker num[20]): +2021-12-29 03:39:19,102 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:39:19,103 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:39:49,133 ==================== TRACER ====================== +2021-12-29 03:39:49,134 Channel (server worker num[20]): +2021-12-29 03:39:49,135 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:39:49,135 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:40:19,166 ==================== TRACER ====================== +2021-12-29 03:40:19,166 Channel (server worker num[20]): +2021-12-29 03:40:19,167 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:40:19,168 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:40:49,198 ==================== TRACER ====================== +2021-12-29 03:40:49,199 Channel (server worker num[20]): +2021-12-29 03:40:49,200 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:40:49,201 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:41:19,231 ==================== TRACER ====================== +2021-12-29 03:41:19,232 Channel (server worker num[20]): +2021-12-29 03:41:19,233 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:41:19,233 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:41:49,264 ==================== TRACER ====================== +2021-12-29 03:41:49,264 Channel (server worker num[20]): +2021-12-29 03:41:49,265 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:41:49,266 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:42:19,296 ==================== TRACER ====================== +2021-12-29 03:42:19,297 Channel (server worker num[20]): +2021-12-29 03:42:19,298 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:42:19,299 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:42:49,329 ==================== TRACER ====================== +2021-12-29 03:42:49,330 Channel (server worker num[20]): +2021-12-29 03:42:49,331 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:42:49,332 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:43:19,362 ==================== TRACER ====================== +2021-12-29 03:43:19,363 Channel (server worker num[20]): +2021-12-29 03:43:19,364 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:43:19,364 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:43:49,395 ==================== TRACER ====================== +2021-12-29 03:43:49,395 Channel (server worker num[20]): +2021-12-29 03:43:49,396 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:43:49,397 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:44:19,427 ==================== TRACER ====================== +2021-12-29 03:44:19,428 Channel (server worker num[20]): +2021-12-29 03:44:19,429 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:44:19,430 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:44:49,460 ==================== TRACER ====================== +2021-12-29 03:44:49,461 Channel (server worker num[20]): +2021-12-29 03:44:49,462 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:44:49,462 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:45:19,467 ==================== TRACER ====================== +2021-12-29 03:45:19,468 Channel (server worker num[20]): +2021-12-29 03:45:19,469 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:45:19,470 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:45:49,500 ==================== TRACER ====================== +2021-12-29 03:45:49,501 Channel (server worker num[20]): +2021-12-29 03:45:49,502 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:45:49,502 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:46:19,533 ==================== TRACER ====================== +2021-12-29 03:46:19,533 Channel (server worker num[20]): +2021-12-29 03:46:19,534 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:46:19,535 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:46:49,565 ==================== TRACER ====================== +2021-12-29 03:46:49,566 Channel (server worker num[20]): +2021-12-29 03:46:49,567 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:46:49,567 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:47:19,589 ==================== TRACER ====================== +2021-12-29 03:47:19,590 Channel (server worker num[20]): +2021-12-29 03:47:19,591 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:47:19,592 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:47:49,609 ==================== TRACER ====================== +2021-12-29 03:47:49,610 Channel (server worker num[20]): +2021-12-29 03:47:49,610 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:47:49,611 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:48:19,641 ==================== TRACER ====================== +2021-12-29 03:48:19,642 Channel (server worker num[20]): +2021-12-29 03:48:19,643 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:48:19,644 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:48:49,674 ==================== TRACER ====================== +2021-12-29 03:48:49,675 Channel (server worker num[20]): +2021-12-29 03:48:49,676 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:48:49,677 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:49:19,707 ==================== TRACER ====================== +2021-12-29 03:49:19,708 Channel (server worker num[20]): +2021-12-29 03:49:19,709 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:49:19,709 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:49:49,740 ==================== TRACER ====================== +2021-12-29 03:49:49,740 Channel (server worker num[20]): +2021-12-29 03:49:49,741 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:49:49,742 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:50:19,772 ==================== TRACER ====================== +2021-12-29 03:50:19,773 Channel (server worker num[20]): +2021-12-29 03:50:19,774 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:50:19,775 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:50:49,805 ==================== TRACER ====================== +2021-12-29 03:50:49,806 Channel (server worker num[20]): +2021-12-29 03:50:49,806 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:50:49,807 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:51:19,837 ==================== TRACER ====================== +2021-12-29 03:51:19,838 Channel (server worker num[20]): +2021-12-29 03:51:19,839 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:51:19,840 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:51:49,870 ==================== TRACER ====================== +2021-12-29 03:51:49,871 Channel (server worker num[20]): +2021-12-29 03:51:49,872 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:51:49,872 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:52:19,880 ==================== TRACER ====================== +2021-12-29 03:52:19,881 Channel (server worker num[20]): +2021-12-29 03:52:19,882 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:52:19,882 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:52:49,913 ==================== TRACER ====================== +2021-12-29 03:52:49,913 Channel (server worker num[20]): +2021-12-29 03:52:49,914 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:52:49,915 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:53:19,945 ==================== TRACER ====================== +2021-12-29 03:53:19,946 Channel (server worker num[20]): +2021-12-29 03:53:19,947 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:53:19,947 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:53:49,978 ==================== TRACER ====================== +2021-12-29 03:53:49,979 Channel (server worker num[20]): +2021-12-29 03:53:49,980 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:53:49,980 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:54:20,011 ==================== TRACER ====================== +2021-12-29 03:54:20,011 Channel (server worker num[20]): +2021-12-29 03:54:20,012 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:54:20,013 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:54:50,043 ==================== TRACER ====================== +2021-12-29 03:54:50,044 Channel (server worker num[20]): +2021-12-29 03:54:50,045 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:54:50,046 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:55:20,076 ==================== TRACER ====================== +2021-12-29 03:55:20,077 Channel (server worker num[20]): +2021-12-29 03:55:20,078 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:55:20,078 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:55:50,109 ==================== TRACER ====================== +2021-12-29 03:55:50,110 Channel (server worker num[20]): +2021-12-29 03:55:50,110 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:55:50,111 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:56:20,141 ==================== TRACER ====================== +2021-12-29 03:56:20,142 Channel (server worker num[20]): +2021-12-29 03:56:20,143 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:56:20,144 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:56:50,174 ==================== TRACER ====================== +2021-12-29 03:56:50,175 Channel (server worker num[20]): +2021-12-29 03:56:50,175 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:56:50,176 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:57:20,206 ==================== TRACER ====================== +2021-12-29 03:57:20,207 Channel (server worker num[20]): +2021-12-29 03:57:20,208 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:57:20,209 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:57:50,239 ==================== TRACER ====================== +2021-12-29 03:57:50,240 Channel (server worker num[20]): +2021-12-29 03:57:50,241 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:57:50,241 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:58:20,272 ==================== TRACER ====================== +2021-12-29 03:58:20,273 Channel (server worker num[20]): +2021-12-29 03:58:20,273 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:58:20,274 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:58:50,304 ==================== TRACER ====================== +2021-12-29 03:58:50,305 Channel (server worker num[20]): +2021-12-29 03:58:50,306 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:58:50,307 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:59:20,337 ==================== TRACER ====================== +2021-12-29 03:59:20,338 Channel (server worker num[20]): +2021-12-29 03:59:20,339 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:59:20,340 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 03:59:50,370 ==================== TRACER ====================== +2021-12-29 03:59:50,371 Channel (server worker num[20]): +2021-12-29 03:59:50,372 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 03:59:50,372 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:00:20,403 ==================== TRACER ====================== +2021-12-29 04:00:20,404 Channel (server worker num[20]): +2021-12-29 04:00:20,404 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:00:20,405 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:00:50,435 ==================== TRACER ====================== +2021-12-29 04:00:50,436 Channel (server worker num[20]): +2021-12-29 04:00:50,437 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:00:50,438 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:01:20,468 ==================== TRACER ====================== +2021-12-29 04:01:20,469 Channel (server worker num[20]): +2021-12-29 04:01:20,470 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:01:20,471 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:01:50,501 ==================== TRACER ====================== +2021-12-29 04:01:50,502 Channel (server worker num[20]): +2021-12-29 04:01:50,503 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:01:50,503 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:02:20,534 ==================== TRACER ====================== +2021-12-29 04:02:20,535 Channel (server worker num[20]): +2021-12-29 04:02:20,535 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:02:20,536 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:02:50,567 ==================== TRACER ====================== +2021-12-29 04:02:50,567 Channel (server worker num[20]): +2021-12-29 04:02:50,568 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:02:50,569 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:03:20,596 ==================== TRACER ====================== +2021-12-29 04:03:20,597 Channel (server worker num[20]): +2021-12-29 04:03:20,598 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:03:20,598 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:03:50,629 ==================== TRACER ====================== +2021-12-29 04:03:50,630 Channel (server worker num[20]): +2021-12-29 04:03:50,631 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:03:50,631 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:04:20,660 ==================== TRACER ====================== +2021-12-29 04:04:20,661 Channel (server worker num[20]): +2021-12-29 04:04:20,662 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:04:20,662 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:04:50,693 ==================== TRACER ====================== +2021-12-29 04:04:50,693 Channel (server worker num[20]): +2021-12-29 04:04:50,694 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:04:50,695 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:05:20,725 ==================== TRACER ====================== +2021-12-29 04:05:20,726 Channel (server worker num[20]): +2021-12-29 04:05:20,727 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:05:20,728 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:05:50,758 ==================== TRACER ====================== +2021-12-29 04:05:50,759 Channel (server worker num[20]): +2021-12-29 04:05:50,760 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:05:50,761 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:06:20,791 ==================== TRACER ====================== +2021-12-29 04:06:20,792 Channel (server worker num[20]): +2021-12-29 04:06:20,793 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:06:20,793 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:06:50,824 ==================== TRACER ====================== +2021-12-29 04:06:50,825 Channel (server worker num[20]): +2021-12-29 04:06:50,826 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:06:50,826 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:07:20,857 ==================== TRACER ====================== +2021-12-29 04:07:20,858 Channel (server worker num[20]): +2021-12-29 04:07:20,858 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:07:20,859 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:07:50,889 ==================== TRACER ====================== +2021-12-29 04:07:50,890 Channel (server worker num[20]): +2021-12-29 04:07:50,891 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:07:50,892 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:08:20,908 ==================== TRACER ====================== +2021-12-29 04:08:20,909 Channel (server worker num[20]): +2021-12-29 04:08:20,910 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:08:20,911 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:08:50,941 ==================== TRACER ====================== +2021-12-29 04:08:50,942 Channel (server worker num[20]): +2021-12-29 04:08:50,943 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:08:50,943 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:09:20,974 ==================== TRACER ====================== +2021-12-29 04:09:20,974 Channel (server worker num[20]): +2021-12-29 04:09:20,975 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:09:20,976 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:09:51,006 ==================== TRACER ====================== +2021-12-29 04:09:51,007 Channel (server worker num[20]): +2021-12-29 04:09:51,008 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:09:51,009 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:10:21,039 ==================== TRACER ====================== +2021-12-29 04:10:21,040 Channel (server worker num[20]): +2021-12-29 04:10:21,041 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:10:21,042 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:10:51,072 ==================== TRACER ====================== +2021-12-29 04:10:51,073 Channel (server worker num[20]): +2021-12-29 04:10:51,074 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:10:51,074 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:11:21,105 ==================== TRACER ====================== +2021-12-29 04:11:21,106 Channel (server worker num[20]): +2021-12-29 04:11:21,107 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:11:21,107 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:11:51,138 ==================== TRACER ====================== +2021-12-29 04:11:51,138 Channel (server worker num[20]): +2021-12-29 04:11:51,139 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:11:51,140 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:12:21,170 ==================== TRACER ====================== +2021-12-29 04:12:21,171 Channel (server worker num[20]): +2021-12-29 04:12:21,172 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:12:21,173 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:12:51,200 ==================== TRACER ====================== +2021-12-29 04:12:51,201 Channel (server worker num[20]): +2021-12-29 04:12:51,202 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:12:51,202 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:13:21,233 ==================== TRACER ====================== +2021-12-29 04:13:21,234 Channel (server worker num[20]): +2021-12-29 04:13:21,235 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:13:21,235 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:13:51,266 ==================== TRACER ====================== +2021-12-29 04:13:51,266 Channel (server worker num[20]): +2021-12-29 04:13:51,267 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:13:51,268 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:14:21,298 ==================== TRACER ====================== +2021-12-29 04:14:21,299 Channel (server worker num[20]): +2021-12-29 04:14:21,300 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:14:21,301 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:14:51,331 ==================== TRACER ====================== +2021-12-29 04:14:51,332 Channel (server worker num[20]): +2021-12-29 04:14:51,333 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:14:51,334 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:15:21,364 ==================== TRACER ====================== +2021-12-29 04:15:21,365 Channel (server worker num[20]): +2021-12-29 04:15:21,366 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:15:21,366 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:15:51,397 ==================== TRACER ====================== +2021-12-29 04:15:51,398 Channel (server worker num[20]): +2021-12-29 04:15:51,398 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:15:51,399 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:16:21,430 ==================== TRACER ====================== +2021-12-29 04:16:21,430 Channel (server worker num[20]): +2021-12-29 04:16:21,431 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:16:21,432 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:16:51,462 ==================== TRACER ====================== +2021-12-29 04:16:51,463 Channel (server worker num[20]): +2021-12-29 04:16:51,464 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:16:51,465 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:17:21,491 ==================== TRACER ====================== +2021-12-29 04:17:21,492 Channel (server worker num[20]): +2021-12-29 04:17:21,493 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:17:21,494 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:17:51,524 ==================== TRACER ====================== +2021-12-29 04:17:51,525 Channel (server worker num[20]): +2021-12-29 04:17:51,526 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:17:51,527 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:18:21,557 ==================== TRACER ====================== +2021-12-29 04:18:21,558 Channel (server worker num[20]): +2021-12-29 04:18:21,559 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:18:21,559 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:18:51,590 ==================== TRACER ====================== +2021-12-29 04:18:51,591 Channel (server worker num[20]): +2021-12-29 04:18:51,592 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:18:51,592 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:19:21,623 ==================== TRACER ====================== +2021-12-29 04:19:21,624 Channel (server worker num[20]): +2021-12-29 04:19:21,624 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:19:21,625 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:19:51,655 ==================== TRACER ====================== +2021-12-29 04:19:51,656 Channel (server worker num[20]): +2021-12-29 04:19:51,657 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:19:51,658 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:20:21,688 ==================== TRACER ====================== +2021-12-29 04:20:21,689 Channel (server worker num[20]): +2021-12-29 04:20:21,690 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:20:21,691 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:20:51,721 ==================== TRACER ====================== +2021-12-29 04:20:51,722 Channel (server worker num[20]): +2021-12-29 04:20:51,723 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:20:51,724 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:21:21,754 ==================== TRACER ====================== +2021-12-29 04:21:21,755 Channel (server worker num[20]): +2021-12-29 04:21:21,756 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:21:21,756 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:21:51,787 ==================== TRACER ====================== +2021-12-29 04:21:51,788 Channel (server worker num[20]): +2021-12-29 04:21:51,788 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:21:51,789 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:22:21,819 ==================== TRACER ====================== +2021-12-29 04:22:21,820 Channel (server worker num[20]): +2021-12-29 04:22:21,821 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:22:21,822 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:22:51,852 ==================== TRACER ====================== +2021-12-29 04:22:51,853 Channel (server worker num[20]): +2021-12-29 04:22:51,854 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:22:51,855 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:23:21,858 ==================== TRACER ====================== +2021-12-29 04:23:21,859 Channel (server worker num[20]): +2021-12-29 04:23:21,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:23:21,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:23:51,891 ==================== TRACER ====================== +2021-12-29 04:23:51,891 Channel (server worker num[20]): +2021-12-29 04:23:51,892 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:23:51,893 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:24:21,923 ==================== TRACER ====================== +2021-12-29 04:24:21,924 Channel (server worker num[20]): +2021-12-29 04:24:21,925 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:24:21,926 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:24:51,952 ==================== TRACER ====================== +2021-12-29 04:24:51,953 Channel (server worker num[20]): +2021-12-29 04:24:51,953 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:24:51,954 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:25:21,984 ==================== TRACER ====================== +2021-12-29 04:25:21,985 Channel (server worker num[20]): +2021-12-29 04:25:21,986 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:25:21,987 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:25:52,017 ==================== TRACER ====================== +2021-12-29 04:25:52,018 Channel (server worker num[20]): +2021-12-29 04:25:52,019 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:25:52,019 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:26:22,050 ==================== TRACER ====================== +2021-12-29 04:26:22,051 Channel (server worker num[20]): +2021-12-29 04:26:22,051 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:26:22,052 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:26:52,082 ==================== TRACER ====================== +2021-12-29 04:26:52,083 Channel (server worker num[20]): +2021-12-29 04:26:52,084 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:26:52,085 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:27:22,115 ==================== TRACER ====================== +2021-12-29 04:27:22,116 Channel (server worker num[20]): +2021-12-29 04:27:22,117 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:27:22,118 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:27:52,148 ==================== TRACER ====================== +2021-12-29 04:27:52,149 Channel (server worker num[20]): +2021-12-29 04:27:52,150 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:27:52,151 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:28:22,181 ==================== TRACER ====================== +2021-12-29 04:28:22,182 Channel (server worker num[20]): +2021-12-29 04:28:22,183 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:28:22,183 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:28:52,214 ==================== TRACER ====================== +2021-12-29 04:28:52,215 Channel (server worker num[20]): +2021-12-29 04:28:52,215 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:28:52,216 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:29:22,247 ==================== TRACER ====================== +2021-12-29 04:29:22,247 Channel (server worker num[20]): +2021-12-29 04:29:22,248 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:29:22,249 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:29:52,279 ==================== TRACER ====================== +2021-12-29 04:29:52,280 Channel (server worker num[20]): +2021-12-29 04:29:52,281 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:29:52,282 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:30:22,291 ==================== TRACER ====================== +2021-12-29 04:30:22,292 Channel (server worker num[20]): +2021-12-29 04:30:22,293 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:30:22,293 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:30:52,324 ==================== TRACER ====================== +2021-12-29 04:30:52,325 Channel (server worker num[20]): +2021-12-29 04:30:52,326 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:30:52,326 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:31:22,357 ==================== TRACER ====================== +2021-12-29 04:31:22,357 Channel (server worker num[20]): +2021-12-29 04:31:22,358 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:31:22,359 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:31:52,388 ==================== TRACER ====================== +2021-12-29 04:31:52,389 Channel (server worker num[20]): +2021-12-29 04:31:52,390 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:31:52,391 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:32:22,421 ==================== TRACER ====================== +2021-12-29 04:32:22,422 Channel (server worker num[20]): +2021-12-29 04:32:22,423 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:32:22,423 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:32:52,437 ==================== TRACER ====================== +2021-12-29 04:32:52,438 Channel (server worker num[20]): +2021-12-29 04:32:52,439 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:32:52,440 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:33:22,470 ==================== TRACER ====================== +2021-12-29 04:33:22,471 Channel (server worker num[20]): +2021-12-29 04:33:22,472 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:33:22,472 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:33:52,503 ==================== TRACER ====================== +2021-12-29 04:33:52,504 Channel (server worker num[20]): +2021-12-29 04:33:52,504 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:33:52,505 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:34:22,536 ==================== TRACER ====================== +2021-12-29 04:34:22,536 Channel (server worker num[20]): +2021-12-29 04:34:22,537 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:34:22,538 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:34:52,552 ==================== TRACER ====================== +2021-12-29 04:34:52,553 Channel (server worker num[20]): +2021-12-29 04:34:52,554 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:34:52,555 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:35:22,585 ==================== TRACER ====================== +2021-12-29 04:35:22,586 Channel (server worker num[20]): +2021-12-29 04:35:22,587 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:35:22,587 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:35:52,596 ==================== TRACER ====================== +2021-12-29 04:35:52,597 Channel (server worker num[20]): +2021-12-29 04:35:52,598 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:35:52,599 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:36:22,629 ==================== TRACER ====================== +2021-12-29 04:36:22,630 Channel (server worker num[20]): +2021-12-29 04:36:22,631 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:36:22,632 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:36:52,662 ==================== TRACER ====================== +2021-12-29 04:36:52,663 Channel (server worker num[20]): +2021-12-29 04:36:52,664 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:36:52,664 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:37:22,695 ==================== TRACER ====================== +2021-12-29 04:37:22,696 Channel (server worker num[20]): +2021-12-29 04:37:22,696 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:37:22,697 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:37:52,728 ==================== TRACER ====================== +2021-12-29 04:37:52,728 Channel (server worker num[20]): +2021-12-29 04:37:52,729 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:37:52,730 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:38:22,760 ==================== TRACER ====================== +2021-12-29 04:38:22,761 Channel (server worker num[20]): +2021-12-29 04:38:22,762 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:38:22,763 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:38:52,793 ==================== TRACER ====================== +2021-12-29 04:38:52,794 Channel (server worker num[20]): +2021-12-29 04:38:52,795 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:38:52,796 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:39:22,826 ==================== TRACER ====================== +2021-12-29 04:39:22,827 Channel (server worker num[20]): +2021-12-29 04:39:22,828 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:39:22,828 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:39:52,832 ==================== TRACER ====================== +2021-12-29 04:39:52,833 Channel (server worker num[20]): +2021-12-29 04:39:52,834 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:39:52,834 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:40:22,849 ==================== TRACER ====================== +2021-12-29 04:40:22,850 Channel (server worker num[20]): +2021-12-29 04:40:22,851 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:40:22,851 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:40:52,876 ==================== TRACER ====================== +2021-12-29 04:40:52,877 Channel (server worker num[20]): +2021-12-29 04:40:52,878 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:40:52,878 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:41:22,908 ==================== TRACER ====================== +2021-12-29 04:41:22,909 Channel (server worker num[20]): +2021-12-29 04:41:22,910 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:41:22,911 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:41:52,916 ==================== TRACER ====================== +2021-12-29 04:41:52,917 Channel (server worker num[20]): +2021-12-29 04:41:52,918 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:41:52,919 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:42:22,949 ==================== TRACER ====================== +2021-12-29 04:42:22,950 Channel (server worker num[20]): +2021-12-29 04:42:22,951 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:42:22,952 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:42:52,982 ==================== TRACER ====================== +2021-12-29 04:42:52,983 Channel (server worker num[20]): +2021-12-29 04:42:52,984 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:42:52,985 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:43:23,014 ==================== TRACER ====================== +2021-12-29 04:43:23,014 Channel (server worker num[20]): +2021-12-29 04:43:23,015 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:43:23,016 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:43:53,044 ==================== TRACER ====================== +2021-12-29 04:43:53,045 Channel (server worker num[20]): +2021-12-29 04:43:53,046 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:43:53,047 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:44:23,077 ==================== TRACER ====================== +2021-12-29 04:44:23,078 Channel (server worker num[20]): +2021-12-29 04:44:23,079 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:44:23,079 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:44:53,110 ==================== TRACER ====================== +2021-12-29 04:44:53,111 Channel (server worker num[20]): +2021-12-29 04:44:53,111 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:44:53,112 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:45:23,142 ==================== TRACER ====================== +2021-12-29 04:45:23,143 Channel (server worker num[20]): +2021-12-29 04:45:23,144 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:45:23,145 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:45:53,175 ==================== TRACER ====================== +2021-12-29 04:45:53,176 Channel (server worker num[20]): +2021-12-29 04:45:53,177 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:45:53,178 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:46:23,208 ==================== TRACER ====================== +2021-12-29 04:46:23,209 Channel (server worker num[20]): +2021-12-29 04:46:23,210 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:46:23,211 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:46:53,241 ==================== TRACER ====================== +2021-12-29 04:46:53,242 Channel (server worker num[20]): +2021-12-29 04:46:53,243 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:46:53,244 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:47:23,274 ==================== TRACER ====================== +2021-12-29 04:47:23,275 Channel (server worker num[20]): +2021-12-29 04:47:23,276 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:47:23,277 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:47:53,307 ==================== TRACER ====================== +2021-12-29 04:47:53,308 Channel (server worker num[20]): +2021-12-29 04:47:53,309 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:47:53,309 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:48:23,340 ==================== TRACER ====================== +2021-12-29 04:48:23,341 Channel (server worker num[20]): +2021-12-29 04:48:23,342 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:48:23,342 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:48:53,373 ==================== TRACER ====================== +2021-12-29 04:48:53,373 Channel (server worker num[20]): +2021-12-29 04:48:53,374 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:48:53,375 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:49:23,405 ==================== TRACER ====================== +2021-12-29 04:49:23,406 Channel (server worker num[20]): +2021-12-29 04:49:23,407 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:49:23,408 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:49:53,438 ==================== TRACER ====================== +2021-12-29 04:49:53,439 Channel (server worker num[20]): +2021-12-29 04:49:53,440 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:49:53,440 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:50:23,471 ==================== TRACER ====================== +2021-12-29 04:50:23,472 Channel (server worker num[20]): +2021-12-29 04:50:23,473 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:50:23,473 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:50:53,504 ==================== TRACER ====================== +2021-12-29 04:50:53,505 Channel (server worker num[20]): +2021-12-29 04:50:53,505 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:50:53,506 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:51:23,536 ==================== TRACER ====================== +2021-12-29 04:51:23,537 Channel (server worker num[20]): +2021-12-29 04:51:23,538 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:51:23,539 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:51:53,569 ==================== TRACER ====================== +2021-12-29 04:51:53,570 Channel (server worker num[20]): +2021-12-29 04:51:53,571 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:51:53,572 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:52:23,602 ==================== TRACER ====================== +2021-12-29 04:52:23,603 Channel (server worker num[20]): +2021-12-29 04:52:23,604 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:52:23,605 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:52:53,635 ==================== TRACER ====================== +2021-12-29 04:52:53,636 Channel (server worker num[20]): +2021-12-29 04:52:53,637 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:52:53,637 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:53:23,668 ==================== TRACER ====================== +2021-12-29 04:53:23,669 Channel (server worker num[20]): +2021-12-29 04:53:23,669 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:53:23,670 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:53:53,700 ==================== TRACER ====================== +2021-12-29 04:53:53,701 Channel (server worker num[20]): +2021-12-29 04:53:53,702 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:53:53,703 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:54:23,733 ==================== TRACER ====================== +2021-12-29 04:54:23,734 Channel (server worker num[20]): +2021-12-29 04:54:23,735 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:54:23,736 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:54:53,766 ==================== TRACER ====================== +2021-12-29 04:54:53,767 Channel (server worker num[20]): +2021-12-29 04:54:53,768 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:54:53,768 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:55:23,799 ==================== TRACER ====================== +2021-12-29 04:55:23,800 Channel (server worker num[20]): +2021-12-29 04:55:23,801 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:55:23,801 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:55:53,832 ==================== TRACER ====================== +2021-12-29 04:55:53,833 Channel (server worker num[20]): +2021-12-29 04:55:53,833 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:55:53,834 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:56:23,864 ==================== TRACER ====================== +2021-12-29 04:56:23,865 Channel (server worker num[20]): +2021-12-29 04:56:23,866 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:56:23,867 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:56:53,897 ==================== TRACER ====================== +2021-12-29 04:56:53,898 Channel (server worker num[20]): +2021-12-29 04:56:53,899 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:56:53,900 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:57:23,930 ==================== TRACER ====================== +2021-12-29 04:57:23,931 Channel (server worker num[20]): +2021-12-29 04:57:23,932 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:57:23,932 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:57:53,963 ==================== TRACER ====================== +2021-12-29 04:57:53,964 Channel (server worker num[20]): +2021-12-29 04:57:53,964 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:57:53,965 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:58:23,996 ==================== TRACER ====================== +2021-12-29 04:58:23,996 Channel (server worker num[20]): +2021-12-29 04:58:23,997 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:58:23,998 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:58:54,028 ==================== TRACER ====================== +2021-12-29 04:58:54,029 Channel (server worker num[20]): +2021-12-29 04:58:54,030 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:58:54,030 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:59:24,061 ==================== TRACER ====================== +2021-12-29 04:59:24,062 Channel (server worker num[20]): +2021-12-29 04:59:24,063 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:59:24,063 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 04:59:54,094 ==================== TRACER ====================== +2021-12-29 04:59:54,094 Channel (server worker num[20]): +2021-12-29 04:59:54,095 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 04:59:54,096 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:00:24,126 ==================== TRACER ====================== +2021-12-29 05:00:24,127 Channel (server worker num[20]): +2021-12-29 05:00:24,128 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:00:24,129 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:00:54,159 ==================== TRACER ====================== +2021-12-29 05:00:54,160 Channel (server worker num[20]): +2021-12-29 05:00:54,161 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:00:54,162 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:01:24,190 ==================== TRACER ====================== +2021-12-29 05:01:24,191 Channel (server worker num[20]): +2021-12-29 05:01:24,192 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:01:24,192 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:01:54,223 ==================== TRACER ====================== +2021-12-29 05:01:54,223 Channel (server worker num[20]): +2021-12-29 05:01:54,224 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:01:54,225 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:02:24,255 ==================== TRACER ====================== +2021-12-29 05:02:24,256 Channel (server worker num[20]): +2021-12-29 05:02:24,257 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:02:24,258 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:02:54,288 ==================== TRACER ====================== +2021-12-29 05:02:54,289 Channel (server worker num[20]): +2021-12-29 05:02:54,290 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:02:54,291 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:03:24,321 ==================== TRACER ====================== +2021-12-29 05:03:24,322 Channel (server worker num[20]): +2021-12-29 05:03:24,323 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:03:24,323 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:03:54,354 ==================== TRACER ====================== +2021-12-29 05:03:54,355 Channel (server worker num[20]): +2021-12-29 05:03:54,355 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:03:54,356 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:04:24,387 ==================== TRACER ====================== +2021-12-29 05:04:24,387 Channel (server worker num[20]): +2021-12-29 05:04:24,388 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:04:24,389 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:04:54,419 ==================== TRACER ====================== +2021-12-29 05:04:54,420 Channel (server worker num[20]): +2021-12-29 05:04:54,421 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:04:54,422 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:05:24,434 ==================== TRACER ====================== +2021-12-29 05:05:24,435 Channel (server worker num[20]): +2021-12-29 05:05:24,436 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:05:24,437 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:05:54,467 ==================== TRACER ====================== +2021-12-29 05:05:54,468 Channel (server worker num[20]): +2021-12-29 05:05:54,469 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:05:54,469 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:06:24,500 ==================== TRACER ====================== +2021-12-29 05:06:24,501 Channel (server worker num[20]): +2021-12-29 05:06:24,501 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:06:24,502 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:06:54,532 ==================== TRACER ====================== +2021-12-29 05:06:54,533 Channel (server worker num[20]): +2021-12-29 05:06:54,534 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:06:54,535 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:07:24,565 ==================== TRACER ====================== +2021-12-29 05:07:24,566 Channel (server worker num[20]): +2021-12-29 05:07:24,567 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:07:24,568 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:07:54,598 ==================== TRACER ====================== +2021-12-29 05:07:54,599 Channel (server worker num[20]): +2021-12-29 05:07:54,600 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:07:54,600 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:08:24,631 ==================== TRACER ====================== +2021-12-29 05:08:24,632 Channel (server worker num[20]): +2021-12-29 05:08:24,633 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:08:24,633 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:08:54,664 ==================== TRACER ====================== +2021-12-29 05:08:54,665 Channel (server worker num[20]): +2021-12-29 05:08:54,665 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:08:54,666 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:09:24,696 ==================== TRACER ====================== +2021-12-29 05:09:24,697 Channel (server worker num[20]): +2021-12-29 05:09:24,698 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:09:24,699 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:09:54,729 ==================== TRACER ====================== +2021-12-29 05:09:54,730 Channel (server worker num[20]): +2021-12-29 05:09:54,731 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:09:54,732 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:10:24,762 ==================== TRACER ====================== +2021-12-29 05:10:24,763 Channel (server worker num[20]): +2021-12-29 05:10:24,764 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:10:24,765 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:10:54,795 ==================== TRACER ====================== +2021-12-29 05:10:54,796 Channel (server worker num[20]): +2021-12-29 05:10:54,798 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:10:54,799 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:11:24,829 ==================== TRACER ====================== +2021-12-29 05:11:24,830 Channel (server worker num[20]): +2021-12-29 05:11:24,831 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:11:24,832 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:11:54,858 ==================== TRACER ====================== +2021-12-29 05:11:54,859 Channel (server worker num[20]): +2021-12-29 05:11:54,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:11:54,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:12:24,890 ==================== TRACER ====================== +2021-12-29 05:12:24,891 Channel (server worker num[20]): +2021-12-29 05:12:24,892 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:12:24,893 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:12:54,923 ==================== TRACER ====================== +2021-12-29 05:12:54,924 Channel (server worker num[20]): +2021-12-29 05:12:54,925 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:12:54,926 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:13:24,935 ==================== TRACER ====================== +2021-12-29 05:13:24,936 Channel (server worker num[20]): +2021-12-29 05:13:24,937 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:13:24,937 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:13:54,968 ==================== TRACER ====================== +2021-12-29 05:13:54,969 Channel (server worker num[20]): +2021-12-29 05:13:54,970 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:13:54,970 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:14:25,001 ==================== TRACER ====================== +2021-12-29 05:14:25,001 Channel (server worker num[20]): +2021-12-29 05:14:25,002 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:14:25,003 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:14:55,033 ==================== TRACER ====================== +2021-12-29 05:14:55,034 Channel (server worker num[20]): +2021-12-29 05:14:55,035 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:14:55,036 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:15:25,066 ==================== TRACER ====================== +2021-12-29 05:15:25,067 Channel (server worker num[20]): +2021-12-29 05:15:25,068 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:15:25,069 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:15:55,099 ==================== TRACER ====================== +2021-12-29 05:15:55,100 Channel (server worker num[20]): +2021-12-29 05:15:55,101 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:15:55,101 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:16:25,132 ==================== TRACER ====================== +2021-12-29 05:16:25,133 Channel (server worker num[20]): +2021-12-29 05:16:25,134 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:16:25,134 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:16:55,165 ==================== TRACER ====================== +2021-12-29 05:16:55,165 Channel (server worker num[20]): +2021-12-29 05:16:55,166 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:16:55,167 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:17:25,197 ==================== TRACER ====================== +2021-12-29 05:17:25,198 Channel (server worker num[20]): +2021-12-29 05:17:25,199 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:17:25,200 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:17:55,230 ==================== TRACER ====================== +2021-12-29 05:17:55,231 Channel (server worker num[20]): +2021-12-29 05:17:55,232 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:17:55,233 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:18:25,263 ==================== TRACER ====================== +2021-12-29 05:18:25,264 Channel (server worker num[20]): +2021-12-29 05:18:25,265 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:18:25,265 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:18:55,296 ==================== TRACER ====================== +2021-12-29 05:18:55,297 Channel (server worker num[20]): +2021-12-29 05:18:55,298 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:18:55,298 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:19:25,329 ==================== TRACER ====================== +2021-12-29 05:19:25,330 Channel (server worker num[20]): +2021-12-29 05:19:25,330 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:19:25,331 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:19:55,361 ==================== TRACER ====================== +2021-12-29 05:19:55,362 Channel (server worker num[20]): +2021-12-29 05:19:55,363 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:19:55,364 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:20:25,394 ==================== TRACER ====================== +2021-12-29 05:20:25,395 Channel (server worker num[20]): +2021-12-29 05:20:25,396 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:20:25,397 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:20:55,427 ==================== TRACER ====================== +2021-12-29 05:20:55,428 Channel (server worker num[20]): +2021-12-29 05:20:55,429 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:20:55,430 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:21:25,444 ==================== TRACER ====================== +2021-12-29 05:21:25,445 Channel (server worker num[20]): +2021-12-29 05:21:25,445 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:21:25,446 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:21:55,476 ==================== TRACER ====================== +2021-12-29 05:21:55,477 Channel (server worker num[20]): +2021-12-29 05:21:55,478 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:21:55,479 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:22:25,509 ==================== TRACER ====================== +2021-12-29 05:22:25,510 Channel (server worker num[20]): +2021-12-29 05:22:25,511 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:22:25,512 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:22:55,542 ==================== TRACER ====================== +2021-12-29 05:22:55,543 Channel (server worker num[20]): +2021-12-29 05:22:55,544 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:22:55,544 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:23:25,574 ==================== TRACER ====================== +2021-12-29 05:23:25,575 Channel (server worker num[20]): +2021-12-29 05:23:25,576 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:23:25,577 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:23:55,607 ==================== TRACER ====================== +2021-12-29 05:23:55,608 Channel (server worker num[20]): +2021-12-29 05:23:55,609 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:23:55,610 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:24:25,640 ==================== TRACER ====================== +2021-12-29 05:24:25,641 Channel (server worker num[20]): +2021-12-29 05:24:25,642 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:24:25,642 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:24:55,673 ==================== TRACER ====================== +2021-12-29 05:24:55,674 Channel (server worker num[20]): +2021-12-29 05:24:55,674 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:24:55,675 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:25:25,706 ==================== TRACER ====================== +2021-12-29 05:25:25,706 Channel (server worker num[20]): +2021-12-29 05:25:25,707 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:25:25,708 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:25:55,738 ==================== TRACER ====================== +2021-12-29 05:25:55,739 Channel (server worker num[20]): +2021-12-29 05:25:55,740 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:25:55,740 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:26:25,771 ==================== TRACER ====================== +2021-12-29 05:26:25,772 Channel (server worker num[20]): +2021-12-29 05:26:25,772 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:26:25,773 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:26:55,804 ==================== TRACER ====================== +2021-12-29 05:26:55,804 Channel (server worker num[20]): +2021-12-29 05:26:55,805 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:26:55,806 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:27:25,825 ==================== TRACER ====================== +2021-12-29 05:27:25,826 Channel (server worker num[20]): +2021-12-29 05:27:25,826 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:27:25,827 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:27:55,857 ==================== TRACER ====================== +2021-12-29 05:27:55,858 Channel (server worker num[20]): +2021-12-29 05:27:55,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:27:55,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:28:25,890 ==================== TRACER ====================== +2021-12-29 05:28:25,891 Channel (server worker num[20]): +2021-12-29 05:28:25,892 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:28:25,892 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:28:55,923 ==================== TRACER ====================== +2021-12-29 05:28:55,924 Channel (server worker num[20]): +2021-12-29 05:28:55,924 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:28:55,925 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:29:25,955 ==================== TRACER ====================== +2021-12-29 05:29:25,956 Channel (server worker num[20]): +2021-12-29 05:29:25,957 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:29:25,958 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:29:55,990 ==================== TRACER ====================== +2021-12-29 05:29:56,009 Channel (server worker num[20]): +2021-12-29 05:29:56,050 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:29:56,079 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:30:26,119 ==================== TRACER ====================== +2021-12-29 05:30:26,119 Channel (server worker num[20]): +2021-12-29 05:30:26,120 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:30:26,121 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:30:56,151 ==================== TRACER ====================== +2021-12-29 05:30:56,152 Channel (server worker num[20]): +2021-12-29 05:30:56,153 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:30:56,154 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:31:26,184 ==================== TRACER ====================== +2021-12-29 05:31:26,185 Channel (server worker num[20]): +2021-12-29 05:31:26,186 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:31:26,187 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:31:56,217 ==================== TRACER ====================== +2021-12-29 05:31:56,218 Channel (server worker num[20]): +2021-12-29 05:31:56,219 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:31:56,219 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:32:26,250 ==================== TRACER ====================== +2021-12-29 05:32:26,251 Channel (server worker num[20]): +2021-12-29 05:32:26,251 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:32:26,252 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:32:56,281 ==================== TRACER ====================== +2021-12-29 05:32:56,282 Channel (server worker num[20]): +2021-12-29 05:32:56,283 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:32:56,284 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:33:26,312 ==================== TRACER ====================== +2021-12-29 05:33:26,313 Channel (server worker num[20]): +2021-12-29 05:33:26,314 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:33:26,315 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:33:56,345 ==================== TRACER ====================== +2021-12-29 05:33:56,346 Channel (server worker num[20]): +2021-12-29 05:33:56,347 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:33:56,347 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:34:26,378 ==================== TRACER ====================== +2021-12-29 05:34:26,378 Channel (server worker num[20]): +2021-12-29 05:34:26,379 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:34:26,380 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:34:56,408 ==================== TRACER ====================== +2021-12-29 05:34:56,409 Channel (server worker num[20]): +2021-12-29 05:34:56,410 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:34:56,410 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:35:26,440 ==================== TRACER ====================== +2021-12-29 05:35:26,441 Channel (server worker num[20]): +2021-12-29 05:35:26,442 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:35:26,442 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:35:56,472 ==================== TRACER ====================== +2021-12-29 05:35:56,473 Channel (server worker num[20]): +2021-12-29 05:35:56,474 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:35:56,475 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:36:26,505 ==================== TRACER ====================== +2021-12-29 05:36:26,506 Channel (server worker num[20]): +2021-12-29 05:36:26,507 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:36:26,508 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:36:56,538 ==================== TRACER ====================== +2021-12-29 05:36:56,539 Channel (server worker num[20]): +2021-12-29 05:36:56,540 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:36:56,540 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:37:04,999 ==================== TRACER ====================== +2021-12-29 05:37:05,001 Channel (server worker num[20]): +2021-12-29 05:37:05,003 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:37:05,004 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:37:26,571 ==================== TRACER ====================== +2021-12-29 05:37:26,572 Channel (server worker num[20]): +2021-12-29 05:37:26,573 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:37:26,573 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:37:35,034 ==================== TRACER ====================== +2021-12-29 05:37:35,035 DAGExecutor: +2021-12-29 05:37:35,036 Query count[1] +2021-12-29 05:37:35,036 QPS[0.03333333333333333 q/s] +2021-12-29 05:37:35,036 Succ[0.0] +2021-12-29 05:37:35,036 Error req[0] +2021-12-29 05:37:35,037 Latency: +2021-12-29 05:37:35,037 ave[1922.774 ms] +2021-12-29 05:37:35,037 .50[1922.774 ms] +2021-12-29 05:37:35,037 .60[1922.774 ms] +2021-12-29 05:37:35,037 .70[1922.774 ms] +2021-12-29 05:37:35,037 .80[1922.774 ms] +2021-12-29 05:37:35,038 .90[1922.774 ms] +2021-12-29 05:37:35,038 .95[1922.774 ms] +2021-12-29 05:37:35,038 .99[1922.774 ms] +2021-12-29 05:37:35,038 Channel (server worker num[20]): +2021-12-29 05:37:35,039 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:37:35,040 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:37:56,604 ==================== TRACER ====================== +2021-12-29 05:37:56,604 Channel (server worker num[20]): +2021-12-29 05:37:56,605 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:37:56,606 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:38:05,070 ==================== TRACER ====================== +2021-12-29 05:38:05,071 Channel (server worker num[20]): +2021-12-29 05:38:05,072 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:38:05,072 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:38:26,636 ==================== TRACER ====================== +2021-12-29 05:38:26,637 Channel (server worker num[20]): +2021-12-29 05:38:26,638 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:38:26,639 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:38:35,103 ==================== TRACER ====================== +2021-12-29 05:38:35,104 Channel (server worker num[20]): +2021-12-29 05:38:35,104 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:38:35,105 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:38:56,669 ==================== TRACER ====================== +2021-12-29 05:38:56,670 Channel (server worker num[20]): +2021-12-29 05:38:56,671 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:38:56,672 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:39:05,135 ==================== TRACER ====================== +2021-12-29 05:39:05,136 Channel (server worker num[20]): +2021-12-29 05:39:05,137 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:39:05,138 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:39:26,702 ==================== TRACER ====================== +2021-12-29 05:39:26,703 Channel (server worker num[20]): +2021-12-29 05:39:26,704 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:39:26,704 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:39:35,168 ==================== TRACER ====================== +2021-12-29 05:39:35,169 Channel (server worker num[20]): +2021-12-29 05:39:35,170 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:39:35,170 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:39:56,735 ==================== TRACER ====================== +2021-12-29 05:39:56,735 Channel (server worker num[20]): +2021-12-29 05:39:56,736 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:39:56,737 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:40:05,171 ==================== TRACER ====================== +2021-12-29 05:40:05,172 Channel (server worker num[20]): +2021-12-29 05:40:05,173 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:40:05,173 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:40:11,916 ==================== TRACER ====================== +2021-12-29 05:40:11,918 Channel (server worker num[20]): +2021-12-29 05:40:11,922 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:40:11,922 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:40:26,767 ==================== TRACER ====================== +2021-12-29 05:40:26,768 Channel (server worker num[20]): +2021-12-29 05:40:26,769 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:40:26,770 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:40:41,953 ==================== TRACER ====================== +2021-12-29 05:40:41,954 DAGExecutor: +2021-12-29 05:40:41,954 Query count[1] +2021-12-29 05:40:41,954 QPS[0.03333333333333333 q/s] +2021-12-29 05:40:41,955 Succ[0.0] +2021-12-29 05:40:41,955 Error req[0] +2021-12-29 05:40:41,955 Latency: +2021-12-29 05:40:41,955 ave[1827.423 ms] +2021-12-29 05:40:41,955 .50[1827.423 ms] +2021-12-29 05:40:41,956 .60[1827.423 ms] +2021-12-29 05:40:41,956 .70[1827.423 ms] +2021-12-29 05:40:41,956 .80[1827.423 ms] +2021-12-29 05:40:41,956 .90[1827.423 ms] +2021-12-29 05:40:41,956 .95[1827.423 ms] +2021-12-29 05:40:41,957 .99[1827.423 ms] +2021-12-29 05:40:41,957 Channel (server worker num[20]): +2021-12-29 05:40:41,957 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:40:41,958 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:40:56,776 ==================== TRACER ====================== +2021-12-29 05:40:56,777 Channel (server worker num[20]): +2021-12-29 05:40:56,778 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:40:56,779 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:41:11,989 ==================== TRACER ====================== +2021-12-29 05:41:11,989 Channel (server worker num[20]): +2021-12-29 05:41:11,990 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:41:11,991 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:41:26,808 ==================== TRACER ====================== +2021-12-29 05:41:26,809 Channel (server worker num[20]): +2021-12-29 05:41:26,810 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:41:26,810 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:41:42,021 ==================== TRACER ====================== +2021-12-29 05:41:42,022 Channel (server worker num[20]): +2021-12-29 05:41:42,023 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:41:42,024 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:41:56,841 ==================== TRACER ====================== +2021-12-29 05:41:56,842 Channel (server worker num[20]): +2021-12-29 05:41:56,842 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:41:56,843 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:42:11,656 ==================== TRACER ====================== +2021-12-29 05:42:11,657 Channel (server worker num[20]): +2021-12-29 05:42:11,659 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:42:11,660 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:42:26,872 ==================== TRACER ====================== +2021-12-29 05:42:26,873 Channel (server worker num[20]): +2021-12-29 05:42:26,874 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:42:26,874 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:42:41,691 ==================== TRACER ====================== +2021-12-29 05:42:41,692 DAGExecutor: +2021-12-29 05:42:41,692 Query count[1] +2021-12-29 05:42:41,693 QPS[0.03333333333333333 q/s] +2021-12-29 05:42:41,693 Succ[0.0] +2021-12-29 05:42:41,693 Error req[0] +2021-12-29 05:42:41,693 Latency: +2021-12-29 05:42:41,693 ave[1872.395 ms] +2021-12-29 05:42:41,694 .50[1872.395 ms] +2021-12-29 05:42:41,694 .60[1872.395 ms] +2021-12-29 05:42:41,694 .70[1872.395 ms] +2021-12-29 05:42:41,694 .80[1872.395 ms] +2021-12-29 05:42:41,694 .90[1872.395 ms] +2021-12-29 05:42:41,695 .95[1872.395 ms] +2021-12-29 05:42:41,695 .99[1872.395 ms] +2021-12-29 05:42:41,695 Channel (server worker num[20]): +2021-12-29 05:42:41,696 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:42:41,697 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:42:56,904 ==================== TRACER ====================== +2021-12-29 05:42:56,905 Channel (server worker num[20]): +2021-12-29 05:42:56,906 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:42:56,906 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:43:11,727 ==================== TRACER ====================== +2021-12-29 05:43:11,728 Channel (server worker num[20]): +2021-12-29 05:43:11,729 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:43:11,730 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:43:26,937 ==================== TRACER ====================== +2021-12-29 05:43:26,938 Channel (server worker num[20]): +2021-12-29 05:43:26,939 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:43:26,939 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:43:41,760 ==================== TRACER ====================== +2021-12-29 05:43:41,761 Channel (server worker num[20]): +2021-12-29 05:43:41,761 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:43:41,762 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:43:56,943 ==================== TRACER ====================== +2021-12-29 05:43:56,944 Channel (server worker num[20]): +2021-12-29 05:43:56,944 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:43:56,945 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:44:11,792 ==================== TRACER ====================== +2021-12-29 05:44:11,793 Channel (server worker num[20]): +2021-12-29 05:44:11,794 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:44:11,795 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:44:26,976 ==================== TRACER ====================== +2021-12-29 05:44:26,976 Channel (server worker num[20]): +2021-12-29 05:44:26,977 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:44:26,978 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:44:41,800 ==================== TRACER ====================== +2021-12-29 05:44:41,801 Channel (server worker num[20]): +2021-12-29 05:44:41,802 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:44:41,803 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:44:57,008 ==================== TRACER ====================== +2021-12-29 05:44:57,009 Channel (server worker num[20]): +2021-12-29 05:44:57,010 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:44:57,011 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:45:11,833 ==================== TRACER ====================== +2021-12-29 05:45:11,834 Channel (server worker num[20]): +2021-12-29 05:45:11,835 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:45:11,835 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:45:27,041 ==================== TRACER ====================== +2021-12-29 05:45:27,042 Channel (server worker num[20]): +2021-12-29 05:45:27,043 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:45:27,043 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:45:41,866 ==================== TRACER ====================== +2021-12-29 05:45:41,867 Channel (server worker num[20]): +2021-12-29 05:45:41,867 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:45:41,868 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:45:57,074 ==================== TRACER ====================== +2021-12-29 05:45:57,075 Channel (server worker num[20]): +2021-12-29 05:45:57,075 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:45:57,076 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:46:11,898 ==================== TRACER ====================== +2021-12-29 05:46:11,899 Channel (server worker num[20]): +2021-12-29 05:46:11,900 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:46:11,901 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:46:27,092 ==================== TRACER ====================== +2021-12-29 05:46:27,093 Channel (server worker num[20]): +2021-12-29 05:46:27,094 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:46:27,095 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:46:41,931 ==================== TRACER ====================== +2021-12-29 05:46:41,932 Channel (server worker num[20]): +2021-12-29 05:46:41,933 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:46:41,934 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:46:57,108 ==================== TRACER ====================== +2021-12-29 05:46:57,109 Channel (server worker num[20]): +2021-12-29 05:46:57,110 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:46:57,111 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:47:11,964 ==================== TRACER ====================== +2021-12-29 05:47:11,965 Channel (server worker num[20]): +2021-12-29 05:47:11,966 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:47:11,967 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:47:27,141 ==================== TRACER ====================== +2021-12-29 05:47:27,142 Channel (server worker num[20]): +2021-12-29 05:47:27,143 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:47:27,143 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:47:41,974 ==================== TRACER ====================== +2021-12-29 05:47:41,975 Channel (server worker num[20]): +2021-12-29 05:47:41,976 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:47:41,977 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:47:57,174 ==================== TRACER ====================== +2021-12-29 05:47:57,174 Channel (server worker num[20]): +2021-12-29 05:47:57,175 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:47:57,176 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:48:12,007 ==================== TRACER ====================== +2021-12-29 05:48:12,008 Channel (server worker num[20]): +2021-12-29 05:48:12,009 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:48:12,010 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:48:27,206 ==================== TRACER ====================== +2021-12-29 05:48:27,207 Channel (server worker num[20]): +2021-12-29 05:48:27,208 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:48:27,209 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:48:42,040 ==================== TRACER ====================== +2021-12-29 05:48:42,041 Channel (server worker num[20]): +2021-12-29 05:48:42,042 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:48:42,042 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:48:57,239 ==================== TRACER ====================== +2021-12-29 05:48:57,240 Channel (server worker num[20]): +2021-12-29 05:48:57,241 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:48:57,242 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:49:12,073 ==================== TRACER ====================== +2021-12-29 05:49:12,074 Channel (server worker num[20]): +2021-12-29 05:49:12,074 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:49:12,075 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:49:27,272 ==================== TRACER ====================== +2021-12-29 05:49:27,273 Channel (server worker num[20]): +2021-12-29 05:49:27,274 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:49:27,274 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:49:42,105 ==================== TRACER ====================== +2021-12-29 05:49:42,106 Channel (server worker num[20]): +2021-12-29 05:49:42,107 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:49:42,108 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:49:57,305 ==================== TRACER ====================== +2021-12-29 05:49:57,305 Channel (server worker num[20]): +2021-12-29 05:49:57,306 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:49:57,307 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:50:12,138 ==================== TRACER ====================== +2021-12-29 05:50:12,139 Channel (server worker num[20]): +2021-12-29 05:50:12,140 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:50:12,141 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:50:27,337 ==================== TRACER ====================== +2021-12-29 05:50:27,338 Channel (server worker num[20]): +2021-12-29 05:50:27,339 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:50:27,340 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:50:42,171 ==================== TRACER ====================== +2021-12-29 05:50:42,172 Channel (server worker num[20]): +2021-12-29 05:50:42,173 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:50:42,174 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:50:57,370 ==================== TRACER ====================== +2021-12-29 05:50:57,371 Channel (server worker num[20]): +2021-12-29 05:50:57,372 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:50:57,373 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:51:12,204 ==================== TRACER ====================== +2021-12-29 05:51:12,205 Channel (server worker num[20]): +2021-12-29 05:51:12,206 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:51:12,206 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:51:27,381 ==================== TRACER ====================== +2021-12-29 05:51:27,382 Channel (server worker num[20]): +2021-12-29 05:51:27,383 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:51:27,383 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:51:42,237 ==================== TRACER ====================== +2021-12-29 05:51:42,238 Channel (server worker num[20]): +2021-12-29 05:51:42,238 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:51:42,239 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:51:57,414 ==================== TRACER ====================== +2021-12-29 05:51:57,415 Channel (server worker num[20]): +2021-12-29 05:51:57,415 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:51:57,416 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:52:12,254 ==================== TRACER ====================== +2021-12-29 05:52:12,255 Channel (server worker num[20]): +2021-12-29 05:52:12,256 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:52:12,256 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:52:27,446 ==================== TRACER ====================== +2021-12-29 05:52:27,447 Channel (server worker num[20]): +2021-12-29 05:52:27,448 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:52:27,449 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:52:42,287 ==================== TRACER ====================== +2021-12-29 05:52:42,288 Channel (server worker num[20]): +2021-12-29 05:52:42,288 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:52:42,289 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:52:57,479 ==================== TRACER ====================== +2021-12-29 05:52:57,480 Channel (server worker num[20]): +2021-12-29 05:52:57,481 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:52:57,482 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:53:12,319 ==================== TRACER ====================== +2021-12-29 05:53:12,320 Channel (server worker num[20]): +2021-12-29 05:53:12,321 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:53:12,322 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:53:27,512 ==================== TRACER ====================== +2021-12-29 05:53:27,513 Channel (server worker num[20]): +2021-12-29 05:53:27,514 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:53:27,514 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:53:42,352 ==================== TRACER ====================== +2021-12-29 05:53:42,353 Channel (server worker num[20]): +2021-12-29 05:53:42,354 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:53:42,355 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:53:57,520 ==================== TRACER ====================== +2021-12-29 05:53:57,521 Channel (server worker num[20]): +2021-12-29 05:53:57,521 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:53:57,522 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:54:12,369 ==================== TRACER ====================== +2021-12-29 05:54:12,370 Channel (server worker num[20]): +2021-12-29 05:54:12,371 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:54:12,371 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:54:27,552 ==================== TRACER ====================== +2021-12-29 05:54:27,553 Channel (server worker num[20]): +2021-12-29 05:54:27,554 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:54:27,555 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:54:42,400 ==================== TRACER ====================== +2021-12-29 05:54:42,401 Channel (server worker num[20]): +2021-12-29 05:54:42,402 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:54:42,403 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:54:57,585 ==================== TRACER ====================== +2021-12-29 05:54:57,586 Channel (server worker num[20]): +2021-12-29 05:54:57,587 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:54:57,588 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:55:12,433 ==================== TRACER ====================== +2021-12-29 05:55:12,434 Channel (server worker num[20]): +2021-12-29 05:55:12,435 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:55:12,435 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:55:27,618 ==================== TRACER ====================== +2021-12-29 05:55:27,619 Channel (server worker num[20]): +2021-12-29 05:55:27,620 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:55:27,621 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:55:42,466 ==================== TRACER ====================== +2021-12-29 05:55:42,467 Channel (server worker num[20]): +2021-12-29 05:55:42,467 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:55:42,468 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:55:57,648 ==================== TRACER ====================== +2021-12-29 05:55:57,649 Channel (server worker num[20]): +2021-12-29 05:55:57,650 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:55:57,650 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:56:12,498 ==================== TRACER ====================== +2021-12-29 05:56:12,499 Channel (server worker num[20]): +2021-12-29 05:56:12,500 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:56:12,501 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:56:27,680 ==================== TRACER ====================== +2021-12-29 05:56:27,681 Channel (server worker num[20]): +2021-12-29 05:56:27,683 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:56:27,683 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:56:42,531 ==================== TRACER ====================== +2021-12-29 05:56:42,532 Channel (server worker num[20]): +2021-12-29 05:56:42,533 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:56:42,534 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:56:57,714 ==================== TRACER ====================== +2021-12-29 05:56:57,715 Channel (server worker num[20]): +2021-12-29 05:56:57,716 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:56:57,716 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:57:12,564 ==================== TRACER ====================== +2021-12-29 05:57:12,565 Channel (server worker num[20]): +2021-12-29 05:57:12,566 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:57:12,567 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:57:27,747 ==================== TRACER ====================== +2021-12-29 05:57:27,748 Channel (server worker num[20]): +2021-12-29 05:57:27,748 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:57:27,749 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:57:42,585 ==================== TRACER ====================== +2021-12-29 05:57:42,586 Channel (server worker num[20]): +2021-12-29 05:57:42,586 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:57:42,587 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:57:57,779 ==================== TRACER ====================== +2021-12-29 05:57:57,780 Channel (server worker num[20]): +2021-12-29 05:57:57,781 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:57:57,782 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:58:12,617 ==================== TRACER ====================== +2021-12-29 05:58:12,618 Channel (server worker num[20]): +2021-12-29 05:58:12,619 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:58:12,620 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:58:27,812 ==================== TRACER ====================== +2021-12-29 05:58:27,813 Channel (server worker num[20]): +2021-12-29 05:58:27,814 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:58:27,815 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:58:42,650 ==================== TRACER ====================== +2021-12-29 05:58:42,651 Channel (server worker num[20]): +2021-12-29 05:58:42,652 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:58:42,653 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:58:57,815 ==================== TRACER ====================== +2021-12-29 05:58:57,816 Channel (server worker num[20]): +2021-12-29 05:58:57,817 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:58:57,818 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:59:12,683 ==================== TRACER ====================== +2021-12-29 05:59:12,684 Channel (server worker num[20]): +2021-12-29 05:59:12,685 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:59:12,686 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:59:27,848 ==================== TRACER ====================== +2021-12-29 05:59:27,849 Channel (server worker num[20]): +2021-12-29 05:59:27,850 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:59:27,850 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:59:42,716 ==================== TRACER ====================== +2021-12-29 05:59:42,717 Channel (server worker num[20]): +2021-12-29 05:59:42,718 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:59:42,718 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 05:59:57,880 ==================== TRACER ====================== +2021-12-29 05:59:57,881 Channel (server worker num[20]): +2021-12-29 05:59:57,882 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 05:59:57,882 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:00:12,749 ==================== TRACER ====================== +2021-12-29 06:00:12,750 Channel (server worker num[20]): +2021-12-29 06:00:12,751 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:00:12,751 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:00:27,913 ==================== TRACER ====================== +2021-12-29 06:00:27,914 Channel (server worker num[20]): +2021-12-29 06:00:27,914 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:00:27,915 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:00:42,782 ==================== TRACER ====================== +2021-12-29 06:00:42,782 Channel (server worker num[20]): +2021-12-29 06:00:42,783 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:00:42,784 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:00:57,946 ==================== TRACER ====================== +2021-12-29 06:00:57,946 Channel (server worker num[20]): +2021-12-29 06:00:57,947 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:00:57,948 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:01:12,814 ==================== TRACER ====================== +2021-12-29 06:01:12,815 Channel (server worker num[20]): +2021-12-29 06:01:12,816 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:01:12,817 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:01:27,978 ==================== TRACER ====================== +2021-12-29 06:01:27,979 Channel (server worker num[20]): +2021-12-29 06:01:27,980 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:01:27,981 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:01:42,847 ==================== TRACER ====================== +2021-12-29 06:01:42,848 Channel (server worker num[20]): +2021-12-29 06:01:42,849 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:01:42,850 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:01:58,011 ==================== TRACER ====================== +2021-12-29 06:01:58,012 Channel (server worker num[20]): +2021-12-29 06:01:58,013 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:01:58,013 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:02:12,880 ==================== TRACER ====================== +2021-12-29 06:02:12,881 Channel (server worker num[20]): +2021-12-29 06:02:12,882 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:02:12,883 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:02:28,044 ==================== TRACER ====================== +2021-12-29 06:02:28,045 Channel (server worker num[20]): +2021-12-29 06:02:28,045 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:02:28,046 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:02:42,913 ==================== TRACER ====================== +2021-12-29 06:02:42,914 Channel (server worker num[20]): +2021-12-29 06:02:42,915 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:02:42,915 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:02:58,077 ==================== TRACER ====================== +2021-12-29 06:02:58,077 Channel (server worker num[20]): +2021-12-29 06:02:58,078 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:02:58,079 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:03:12,946 ==================== TRACER ====================== +2021-12-29 06:03:12,946 Channel (server worker num[20]): +2021-12-29 06:03:12,947 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:03:12,948 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:03:28,109 ==================== TRACER ====================== +2021-12-29 06:03:28,110 Channel (server worker num[20]): +2021-12-29 06:03:28,111 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:03:28,112 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:03:58,142 ==================== TRACER ====================== +2021-12-29 06:03:58,143 Channel (server worker num[20]): +2021-12-29 06:03:58,144 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:03:58,144 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:04:28,175 ==================== TRACER ====================== +2021-12-29 06:04:28,176 Channel (server worker num[20]): +2021-12-29 06:04:28,176 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:04:28,177 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:04:58,207 ==================== TRACER ====================== +2021-12-29 06:04:58,208 Channel (server worker num[20]): +2021-12-29 06:04:58,209 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:04:58,210 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:05:28,240 ==================== TRACER ====================== +2021-12-29 06:05:28,241 Channel (server worker num[20]): +2021-12-29 06:05:28,242 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:05:28,243 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:05:58,273 ==================== TRACER ====================== +2021-12-29 06:05:58,274 Channel (server worker num[20]): +2021-12-29 06:05:58,275 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:05:58,275 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:06:28,306 ==================== TRACER ====================== +2021-12-29 06:06:28,307 Channel (server worker num[20]): +2021-12-29 06:06:28,307 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:06:28,308 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:06:58,338 ==================== TRACER ====================== +2021-12-29 06:06:58,339 Channel (server worker num[20]): +2021-12-29 06:06:58,340 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:06:58,341 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:07:28,371 ==================== TRACER ====================== +2021-12-29 06:07:28,372 Channel (server worker num[20]): +2021-12-29 06:07:28,373 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:07:28,374 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:07:58,404 ==================== TRACER ====================== +2021-12-29 06:07:58,405 Channel (server worker num[20]): +2021-12-29 06:07:58,406 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:07:58,406 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:08:28,426 ==================== TRACER ====================== +2021-12-29 06:08:28,427 Channel (server worker num[20]): +2021-12-29 06:08:28,428 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:08:28,428 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:08:54,465 ==================== TRACER ====================== +2021-12-29 06:08:54,466 Channel (server worker num[20]): +2021-12-29 06:08:54,468 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:08:54,468 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:08:58,450 ==================== TRACER ====================== +2021-12-29 06:08:58,451 Channel (server worker num[20]): +2021-12-29 06:08:58,452 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:08:58,452 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:09:24,499 ==================== TRACER ====================== +2021-12-29 06:09:24,500 Channel (server worker num[20]): +2021-12-29 06:09:24,500 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:09:24,501 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:09:28,483 ==================== TRACER ====================== +2021-12-29 06:09:28,484 Channel (server worker num[20]): +2021-12-29 06:09:28,484 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:09:28,485 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:09:54,532 ==================== TRACER ====================== +2021-12-29 06:09:54,532 Channel (server worker num[20]): +2021-12-29 06:09:54,533 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:09:54,534 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:09:58,516 ==================== TRACER ====================== +2021-12-29 06:09:58,516 Channel (server worker num[20]): +2021-12-29 06:09:58,517 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:09:58,518 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:10:19,913 ==================== TRACER ====================== +2021-12-29 06:10:19,914 Channel (server worker num[20]): +2021-12-29 06:10:19,917 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:10:19,917 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:10:28,520 ==================== TRACER ====================== +2021-12-29 06:10:28,521 Channel (server worker num[20]): +2021-12-29 06:10:28,522 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:10:28,522 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:10:49,948 ==================== TRACER ====================== +2021-12-29 06:10:49,949 Channel (server worker num[20]): +2021-12-29 06:10:49,950 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:10:49,950 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:10:58,553 ==================== TRACER ====================== +2021-12-29 06:10:58,554 Channel (server worker num[20]): +2021-12-29 06:10:58,555 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:10:58,555 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:11:19,981 ==================== TRACER ====================== +2021-12-29 06:11:19,981 Channel (server worker num[20]): +2021-12-29 06:11:19,982 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:11:19,983 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:11:28,586 ==================== TRACER ====================== +2021-12-29 06:11:28,586 Channel (server worker num[20]): +2021-12-29 06:11:28,587 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:11:28,588 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:11:49,985 ==================== TRACER ====================== +2021-12-29 06:11:49,986 Channel (server worker num[20]): +2021-12-29 06:11:49,986 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:11:49,987 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:11:58,608 ==================== TRACER ====================== +2021-12-29 06:11:58,608 Channel (server worker num[20]): +2021-12-29 06:11:58,609 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:11:58,610 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:12:09,038 ==================== TRACER ====================== +2021-12-29 06:12:09,039 Channel (server worker num[20]): +2021-12-29 06:12:09,041 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:12:09,042 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:12:28,640 ==================== TRACER ====================== +2021-12-29 06:12:28,641 Channel (server worker num[20]): +2021-12-29 06:12:28,642 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:12:28,643 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:12:39,072 ==================== TRACER ====================== +2021-12-29 06:12:39,073 Channel (server worker num[20]): +2021-12-29 06:12:39,074 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:12:39,075 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:12:47,298 ==================== TRACER ====================== +2021-12-29 06:12:47,299 Channel (server worker num[20]): +2021-12-29 06:12:47,302 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:12:47,302 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:12:58,673 ==================== TRACER ====================== +2021-12-29 06:12:58,674 Channel (server worker num[20]): +2021-12-29 06:12:58,675 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:12:58,675 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:13:17,333 ==================== TRACER ====================== +2021-12-29 06:13:17,334 Channel (server worker num[20]): +2021-12-29 06:13:17,334 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:13:17,335 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:13:28,706 ==================== TRACER ====================== +2021-12-29 06:13:28,707 Channel (server worker num[20]): +2021-12-29 06:13:28,707 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:13:28,708 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:13:47,365 ==================== TRACER ====================== +2021-12-29 06:13:47,366 Channel (server worker num[20]): +2021-12-29 06:13:47,367 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:13:47,368 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:13:58,739 ==================== TRACER ====================== +2021-12-29 06:13:58,739 Channel (server worker num[20]): +2021-12-29 06:13:58,740 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:13:58,741 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:14:17,399 ==================== TRACER ====================== +2021-12-29 06:14:17,399 Channel (server worker num[20]): +2021-12-29 06:14:17,400 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:14:17,401 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:14:28,771 ==================== TRACER ====================== +2021-12-29 06:14:28,772 Channel (server worker num[20]): +2021-12-29 06:14:28,773 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:14:28,774 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:14:47,409 ==================== TRACER ====================== +2021-12-29 06:14:47,410 Channel (server worker num[20]): +2021-12-29 06:14:47,410 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[1/0] +2021-12-29 06:14:47,411 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:14:58,804 ==================== TRACER ====================== +2021-12-29 06:14:58,805 Channel (server worker num[20]): +2021-12-29 06:14:58,806 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:14:58,807 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:15:10,573 ==================== TRACER ====================== +2021-12-29 06:15:10,574 Channel (server worker num[20]): +2021-12-29 06:15:10,577 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:15:10,577 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:15:28,837 ==================== TRACER ====================== +2021-12-29 06:15:28,838 Channel (server worker num[20]): +2021-12-29 06:15:28,839 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:15:28,839 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:15:40,599 ==================== TRACER ====================== +2021-12-29 06:15:40,600 Channel (server worker num[20]): +2021-12-29 06:15:40,601 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:15:40,601 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:15:58,870 ==================== TRACER ====================== +2021-12-29 06:15:58,871 Channel (server worker num[20]): +2021-12-29 06:15:58,871 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:15:58,872 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:16:10,614 ==================== TRACER ====================== +2021-12-29 06:16:10,615 Channel (server worker num[20]): +2021-12-29 06:16:10,616 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:16:10,616 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:16:28,892 ==================== TRACER ====================== +2021-12-29 06:16:28,893 Channel (server worker num[20]): +2021-12-29 06:16:28,894 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:16:28,894 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:16:40,623 ==================== TRACER ====================== +2021-12-29 06:16:40,624 Channel (server worker num[20]): +2021-12-29 06:16:40,625 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:16:40,626 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:16:58,924 ==================== TRACER ====================== +2021-12-29 06:16:58,925 Channel (server worker num[20]): +2021-12-29 06:16:58,926 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:16:58,927 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:17:10,656 ==================== TRACER ====================== +2021-12-29 06:17:10,657 Channel (server worker num[20]): +2021-12-29 06:17:10,658 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:17:10,659 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:17:28,957 ==================== TRACER ====================== +2021-12-29 06:17:28,958 Channel (server worker num[20]): +2021-12-29 06:17:28,959 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:17:28,959 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:17:36,431 ==================== TRACER ====================== +2021-12-29 06:17:36,433 Channel (server worker num[20]): +2021-12-29 06:17:36,435 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:17:36,436 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:17:58,990 ==================== TRACER ====================== +2021-12-29 06:17:58,991 Channel (server worker num[20]): +2021-12-29 06:17:58,991 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:17:58,992 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:18:06,466 ==================== TRACER ====================== +2021-12-29 06:18:06,467 Channel (server worker num[20]): +2021-12-29 06:18:06,468 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:18:06,469 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:18:17,514 ==================== TRACER ====================== +2021-12-29 06:18:17,516 Channel (server worker num[20]): +2021-12-29 06:18:17,518 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:18:17,518 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:18:29,022 ==================== TRACER ====================== +2021-12-29 06:18:29,023 Channel (server worker num[20]): +2021-12-29 06:18:29,024 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:18:29,025 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:18:47,549 ==================== TRACER ====================== +2021-12-29 06:18:47,550 Channel (server worker num[20]): +2021-12-29 06:18:47,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:18:47,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:18:59,055 ==================== TRACER ====================== +2021-12-29 06:18:59,056 Channel (server worker num[20]): +2021-12-29 06:18:59,057 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:18:59,058 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:19:17,582 ==================== TRACER ====================== +2021-12-29 06:19:17,583 Channel (server worker num[20]): +2021-12-29 06:19:17,583 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:19:17,584 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:19:29,064 ==================== TRACER ====================== +2021-12-29 06:19:29,065 Channel (server worker num[20]): +2021-12-29 06:19:29,066 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:19:29,067 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:19:47,614 ==================== TRACER ====================== +2021-12-29 06:19:47,615 Channel (server worker num[20]): +2021-12-29 06:19:47,616 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:19:47,617 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:19:57,984 ==================== TRACER ====================== +2021-12-29 06:19:57,986 Channel (server worker num[20]): +2021-12-29 06:19:57,988 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:19:57,989 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:19:59,080 ==================== TRACER ====================== +2021-12-29 06:19:59,081 Channel (server worker num[20]): +2021-12-29 06:19:59,082 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:19:59,082 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:20:28,000 ==================== TRACER ====================== +2021-12-29 06:20:28,001 Channel (server worker num[20]): +2021-12-29 06:20:28,002 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:20:28,003 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:20:29,093 ==================== TRACER ====================== +2021-12-29 06:20:29,094 Channel (server worker num[20]): +2021-12-29 06:20:29,095 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:20:29,095 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:20:58,032 ==================== TRACER ====================== +2021-12-29 06:20:58,033 Channel (server worker num[20]): +2021-12-29 06:20:58,034 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:20:58,035 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:20:59,126 ==================== TRACER ====================== +2021-12-29 06:20:59,126 Channel (server worker num[20]): +2021-12-29 06:20:59,127 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:20:59,128 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:21:28,732 ==================== TRACER ====================== +2021-12-29 06:21:28,733 Channel (server worker num[20]): +2021-12-29 06:21:28,735 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:21:28,736 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:21:29,158 ==================== TRACER ====================== +2021-12-29 06:21:29,159 Channel (server worker num[20]): +2021-12-29 06:21:29,160 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:21:29,161 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:21:58,766 ==================== TRACER ====================== +2021-12-29 06:21:58,767 Channel (server worker num[20]): +2021-12-29 06:21:58,768 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:21:58,768 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:21:59,175 ==================== TRACER ====================== +2021-12-29 06:21:59,175 Channel (server worker num[20]): +2021-12-29 06:21:59,176 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:21:59,177 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:22:28,799 ==================== TRACER ====================== +2021-12-29 06:22:28,800 Channel (server worker num[20]): +2021-12-29 06:22:28,801 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:22:28,801 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:22:29,178 ==================== TRACER ====================== +2021-12-29 06:22:29,179 Channel (server worker num[20]): +2021-12-29 06:22:29,180 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:22:29,180 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:22:58,832 ==================== TRACER ====================== +2021-12-29 06:22:58,832 Channel (server worker num[20]): +2021-12-29 06:22:58,833 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:22:58,834 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:22:59,211 ==================== TRACER ====================== +2021-12-29 06:22:59,211 Channel (server worker num[20]): +2021-12-29 06:22:59,212 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:22:59,213 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:23:28,864 ==================== TRACER ====================== +2021-12-29 06:23:28,865 Channel (server worker num[20]): +2021-12-29 06:23:28,866 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:23:28,867 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:23:29,243 ==================== TRACER ====================== +2021-12-29 06:23:29,244 Channel (server worker num[20]): +2021-12-29 06:23:29,245 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:23:29,246 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:23:58,897 ==================== TRACER ====================== +2021-12-29 06:23:58,898 Channel (server worker num[20]): +2021-12-29 06:23:58,899 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:23:58,900 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:23:59,270 ==================== TRACER ====================== +2021-12-29 06:23:59,270 Channel (server worker num[20]): +2021-12-29 06:23:59,271 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:23:59,272 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:24:29,302 ==================== TRACER ====================== +2021-12-29 06:24:29,303 Channel (server worker num[20]): +2021-12-29 06:24:29,304 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:24:29,305 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:24:59,335 ==================== TRACER ====================== +2021-12-29 06:24:59,336 Channel (server worker num[20]): +2021-12-29 06:24:59,337 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:24:59,337 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:25:12,161 ==================== TRACER ====================== +2021-12-29 06:25:12,162 Channel (server worker num[20]): +2021-12-29 06:25:12,165 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:25:12,166 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:25:29,368 ==================== TRACER ====================== +2021-12-29 06:25:29,368 Channel (server worker num[20]): +2021-12-29 06:25:29,369 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:25:29,370 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:25:29,547 ==================== TRACER ====================== +2021-12-29 06:25:29,548 Channel (server worker num[20]): +2021-12-29 06:25:29,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:25:29,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:25:59,376 ==================== TRACER ====================== +2021-12-29 06:25:59,377 Channel (server worker num[20]): +2021-12-29 06:25:59,377 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:25:59,378 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:25:59,581 ==================== TRACER ====================== +2021-12-29 06:25:59,582 Channel (server worker num[20]): +2021-12-29 06:25:59,583 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:25:59,584 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:26:29,409 ==================== TRACER ====================== +2021-12-29 06:26:29,409 Channel (server worker num[20]): +2021-12-29 06:26:29,410 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:26:29,411 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:26:29,614 ==================== TRACER ====================== +2021-12-29 06:26:29,615 Channel (server worker num[20]): +2021-12-29 06:26:29,616 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:26:29,617 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:26:59,441 ==================== TRACER ====================== +2021-12-29 06:26:59,442 Channel (server worker num[20]): +2021-12-29 06:26:59,443 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:26:59,444 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:26:59,648 ==================== TRACER ====================== +2021-12-29 06:26:59,648 Channel (server worker num[20]): +2021-12-29 06:26:59,649 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:26:59,650 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:27:29,474 ==================== TRACER ====================== +2021-12-29 06:27:29,475 Channel (server worker num[20]): +2021-12-29 06:27:29,476 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:27:29,476 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:27:29,680 ==================== TRACER ====================== +2021-12-29 06:27:29,681 Channel (server worker num[20]): +2021-12-29 06:27:29,682 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:27:29,683 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:27:59,507 ==================== TRACER ====================== +2021-12-29 06:27:59,507 Channel (server worker num[20]): +2021-12-29 06:27:59,508 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:27:59,509 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:27:59,713 ==================== TRACER ====================== +2021-12-29 06:27:59,714 Channel (server worker num[20]): +2021-12-29 06:27:59,715 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:27:59,716 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:28:29,539 ==================== TRACER ====================== +2021-12-29 06:28:29,540 Channel (server worker num[20]): +2021-12-29 06:28:29,541 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:28:29,542 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:28:29,746 ==================== TRACER ====================== +2021-12-29 06:28:29,747 Channel (server worker num[20]): +2021-12-29 06:28:29,748 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:28:29,748 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:28:59,572 ==================== TRACER ====================== +2021-12-29 06:28:59,573 Channel (server worker num[20]): +2021-12-29 06:28:59,574 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:28:59,574 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:28:59,779 ==================== TRACER ====================== +2021-12-29 06:28:59,780 Channel (server worker num[20]): +2021-12-29 06:28:59,780 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:28:59,781 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:29:29,605 ==================== TRACER ====================== +2021-12-29 06:29:29,606 Channel (server worker num[20]): +2021-12-29 06:29:29,606 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:29:29,607 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:29:29,812 ==================== TRACER ====================== +2021-12-29 06:29:29,812 Channel (server worker num[20]): +2021-12-29 06:29:29,813 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:29:29,814 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:29:59,638 ==================== TRACER ====================== +2021-12-29 06:29:59,638 Channel (server worker num[20]): +2021-12-29 06:29:59,639 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:29:59,640 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:29:59,844 ==================== TRACER ====================== +2021-12-29 06:29:59,845 Channel (server worker num[20]): +2021-12-29 06:29:59,846 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:29:59,847 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:30:29,670 ==================== TRACER ====================== +2021-12-29 06:30:29,671 Channel (server worker num[20]): +2021-12-29 06:30:29,672 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:30:29,673 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:30:29,877 ==================== TRACER ====================== +2021-12-29 06:30:29,878 Channel (server worker num[20]): +2021-12-29 06:30:29,879 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:30:29,880 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:30:59,703 ==================== TRACER ====================== +2021-12-29 06:30:59,704 Channel (server worker num[20]): +2021-12-29 06:30:59,705 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:30:59,706 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:30:59,910 ==================== TRACER ====================== +2021-12-29 06:30:59,911 Channel (server worker num[20]): +2021-12-29 06:30:59,912 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:30:59,912 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:31:29,736 ==================== TRACER ====================== +2021-12-29 06:31:29,737 Channel (server worker num[20]): +2021-12-29 06:31:29,738 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:31:29,738 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:31:29,943 ==================== TRACER ====================== +2021-12-29 06:31:29,944 Channel (server worker num[20]): +2021-12-29 06:31:29,945 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:31:29,945 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:31:43,566 ==================== TRACER ====================== +2021-12-29 06:31:43,568 Channel (server worker num[20]): +2021-12-29 06:31:43,570 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:31:43,571 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:31:59,769 ==================== TRACER ====================== +2021-12-29 06:31:59,770 Channel (server worker num[20]): +2021-12-29 06:31:59,770 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:31:59,771 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:32:13,591 ==================== TRACER ====================== +2021-12-29 06:32:13,592 Channel (server worker num[20]): +2021-12-29 06:32:13,593 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:32:13,594 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:32:29,801 ==================== TRACER ====================== +2021-12-29 06:32:29,802 Channel (server worker num[20]): +2021-12-29 06:32:29,803 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:32:29,804 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:32:31,132 ==================== TRACER ====================== +2021-12-29 06:32:31,134 Channel (server worker num[20]): +2021-12-29 06:32:31,138 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:32:31,138 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:32:59,834 ==================== TRACER ====================== +2021-12-29 06:32:59,835 Channel (server worker num[20]): +2021-12-29 06:32:59,836 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:32:59,837 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:33:01,169 ==================== TRACER ====================== +2021-12-29 06:33:01,170 Channel (server worker num[20]): +2021-12-29 06:33:01,171 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:33:01,171 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:33:05,365 ==================== TRACER ====================== +2021-12-29 06:33:05,368 Channel (server worker num[20]): +2021-12-29 06:33:05,370 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:33:05,370 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:33:29,848 ==================== TRACER ====================== +2021-12-29 06:33:29,849 Channel (server worker num[20]): +2021-12-29 06:33:29,850 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:33:29,850 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:33:35,401 ==================== TRACER ====================== +2021-12-29 06:33:35,402 Channel (server worker num[20]): +2021-12-29 06:33:35,402 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:33:35,403 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:33:59,858 ==================== TRACER ====================== +2021-12-29 06:33:59,859 Channel (server worker num[20]): +2021-12-29 06:33:59,860 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:33:59,861 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:34:05,424 ==================== TRACER ====================== +2021-12-29 06:34:05,425 Channel (server worker num[20]): +2021-12-29 06:34:05,428 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:34:05,429 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:34:29,891 ==================== TRACER ====================== +2021-12-29 06:34:29,892 Channel (server worker num[20]): +2021-12-29 06:34:29,893 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:34:29,894 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:34:35,434 ==================== TRACER ====================== +2021-12-29 06:34:35,435 Channel (server worker num[20]): +2021-12-29 06:34:35,436 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:34:35,437 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:34:59,924 ==================== TRACER ====================== +2021-12-29 06:34:59,925 Channel (server worker num[20]): +2021-12-29 06:34:59,925 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:34:59,926 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:35:05,467 ==================== TRACER ====================== +2021-12-29 06:35:05,468 Channel (server worker num[20]): +2021-12-29 06:35:05,469 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:35:05,470 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:35:29,937 ==================== TRACER ====================== +2021-12-29 06:35:29,938 Channel (server worker num[20]): +2021-12-29 06:35:29,939 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:35:29,940 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:35:35,500 ==================== TRACER ====================== +2021-12-29 06:35:35,501 Channel (server worker num[20]): +2021-12-29 06:35:35,502 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:35:35,503 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:35:59,970 ==================== TRACER ====================== +2021-12-29 06:35:59,971 Channel (server worker num[20]): +2021-12-29 06:35:59,972 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:35:59,972 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:36:05,533 ==================== TRACER ====================== +2021-12-29 06:36:05,534 Channel (server worker num[20]): +2021-12-29 06:36:05,535 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:36:05,536 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:36:30,003 ==================== TRACER ====================== +2021-12-29 06:36:30,004 Channel (server worker num[20]): +2021-12-29 06:36:30,005 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:36:30,005 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:36:35,566 ==================== TRACER ====================== +2021-12-29 06:36:35,567 Channel (server worker num[20]): +2021-12-29 06:36:35,568 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:36:35,569 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:37:00,036 ==================== TRACER ====================== +2021-12-29 06:37:00,036 Channel (server worker num[20]): +2021-12-29 06:37:00,037 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:37:00,038 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:37:05,599 ==================== TRACER ====================== +2021-12-29 06:37:05,600 Channel (server worker num[20]): +2021-12-29 06:37:05,601 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:37:05,602 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:37:30,068 ==================== TRACER ====================== +2021-12-29 06:37:30,069 Channel (server worker num[20]): +2021-12-29 06:37:30,070 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:37:30,071 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:37:35,632 ==================== TRACER ====================== +2021-12-29 06:37:35,633 Channel (server worker num[20]): +2021-12-29 06:37:35,634 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:37:35,635 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:38:00,100 ==================== TRACER ====================== +2021-12-29 06:38:00,101 Channel (server worker num[20]): +2021-12-29 06:38:00,101 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:38:00,102 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:38:05,665 ==================== TRACER ====================== +2021-12-29 06:38:05,666 Channel (server worker num[20]): +2021-12-29 06:38:05,667 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:38:05,668 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:38:30,128 ==================== TRACER ====================== +2021-12-29 06:38:30,129 Channel (server worker num[20]): +2021-12-29 06:38:30,130 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:38:30,131 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:38:35,697 ==================== TRACER ====================== +2021-12-29 06:38:35,698 Channel (server worker num[20]): +2021-12-29 06:38:35,699 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:38:35,700 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:39:00,161 ==================== TRACER ====================== +2021-12-29 06:39:00,162 Channel (server worker num[20]): +2021-12-29 06:39:00,163 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:39:00,163 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:39:05,730 ==================== TRACER ====================== +2021-12-29 06:39:05,731 Channel (server worker num[20]): +2021-12-29 06:39:05,734 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:39:05,734 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:39:30,194 ==================== TRACER ====================== +2021-12-29 06:39:30,195 Channel (server worker num[20]): +2021-12-29 06:39:30,195 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:39:30,196 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:39:35,765 ==================== TRACER ====================== +2021-12-29 06:39:35,766 Channel (server worker num[20]): +2021-12-29 06:39:35,767 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:39:35,767 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:40:00,226 ==================== TRACER ====================== +2021-12-29 06:40:00,227 Channel (server worker num[20]): +2021-12-29 06:40:00,228 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:40:00,229 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:40:05,798 ==================== TRACER ====================== +2021-12-29 06:40:05,799 Channel (server worker num[20]): +2021-12-29 06:40:05,799 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:40:05,800 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:40:19,364 ==================== TRACER ====================== +2021-12-29 06:40:19,365 Channel (server worker num[20]): +2021-12-29 06:40:19,367 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:40:19,368 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:40:30,259 ==================== TRACER ====================== +2021-12-29 06:40:30,260 Channel (server worker num[20]): +2021-12-29 06:40:30,261 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:40:30,261 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:40:49,398 ==================== TRACER ====================== +2021-12-29 06:40:49,399 DAGExecutor: +2021-12-29 06:40:49,400 Query count[1] +2021-12-29 06:40:49,400 QPS[0.03333333333333333 q/s] +2021-12-29 06:40:49,400 Succ[0.0] +2021-12-29 06:40:49,400 Error req[0] +2021-12-29 06:40:49,401 Latency: +2021-12-29 06:40:49,401 ave[50.848 ms] +2021-12-29 06:40:49,401 .50[50.848 ms] +2021-12-29 06:40:49,401 .60[50.848 ms] +2021-12-29 06:40:49,401 .70[50.848 ms] +2021-12-29 06:40:49,402 .80[50.848 ms] +2021-12-29 06:40:49,402 .90[50.848 ms] +2021-12-29 06:40:49,402 .95[50.848 ms] +2021-12-29 06:40:49,402 .99[50.848 ms] +2021-12-29 06:40:49,402 Channel (server worker num[20]): +2021-12-29 06:40:49,403 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:40:49,404 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:41:00,292 ==================== TRACER ====================== +2021-12-29 06:41:00,292 Channel (server worker num[20]): +2021-12-29 06:41:00,293 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:41:00,294 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:41:19,434 ==================== TRACER ====================== +2021-12-29 06:41:19,435 Channel (server worker num[20]): +2021-12-29 06:41:19,436 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:41:19,437 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:41:30,300 ==================== TRACER ====================== +2021-12-29 06:41:30,301 Channel (server worker num[20]): +2021-12-29 06:41:30,302 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:41:30,303 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:42:00,333 ==================== TRACER ====================== +2021-12-29 06:42:00,334 Channel (server worker num[20]): +2021-12-29 06:42:00,335 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:42:00,335 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:42:03,488 ==================== TRACER ====================== +2021-12-29 06:42:03,489 Channel (server worker num[20]): +2021-12-29 06:42:03,491 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:42:03,492 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:42:30,366 ==================== TRACER ====================== +2021-12-29 06:42:30,367 Channel (server worker num[20]): +2021-12-29 06:42:30,367 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:42:30,368 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:42:33,522 ==================== TRACER ====================== +2021-12-29 06:42:33,523 DAGExecutor: +2021-12-29 06:42:33,524 Query count[1] +2021-12-29 06:42:33,524 QPS[0.03333333333333333 q/s] +2021-12-29 06:42:33,524 Succ[0.0] +2021-12-29 06:42:33,524 Error req[0] +2021-12-29 06:42:33,525 Latency: +2021-12-29 06:42:33,525 ave[44.165 ms] +2021-12-29 06:42:33,525 .50[44.165 ms] +2021-12-29 06:42:33,525 .60[44.165 ms] +2021-12-29 06:42:33,525 .70[44.165 ms] +2021-12-29 06:42:33,526 .80[44.165 ms] +2021-12-29 06:42:33,526 .90[44.165 ms] +2021-12-29 06:42:33,526 .95[44.165 ms] +2021-12-29 06:42:33,526 .99[44.165 ms] +2021-12-29 06:42:33,526 Channel (server worker num[20]): +2021-12-29 06:42:33,527 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:42:33,528 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:42:44,285 ==================== TRACER ====================== +2021-12-29 06:42:44,287 Channel (server worker num[20]): +2021-12-29 06:42:44,289 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:42:44,289 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:43:00,398 ==================== TRACER ====================== +2021-12-29 06:43:00,399 Channel (server worker num[20]): +2021-12-29 06:43:00,400 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:43:00,401 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:43:14,300 ==================== TRACER ====================== +2021-12-29 06:43:14,301 DAGExecutor: +2021-12-29 06:43:14,302 Query count[1] +2021-12-29 06:43:14,302 QPS[0.03333333333333333 q/s] +2021-12-29 06:43:14,302 Succ[0.0] +2021-12-29 06:43:14,302 Error req[0] +2021-12-29 06:43:14,302 Latency: +2021-12-29 06:43:14,303 ave[47.774 ms] +2021-12-29 06:43:14,303 .50[47.774 ms] +2021-12-29 06:43:14,303 .60[47.774 ms] +2021-12-29 06:43:14,303 .70[47.774 ms] +2021-12-29 06:43:14,303 .80[47.774 ms] +2021-12-29 06:43:14,303 .90[47.774 ms] +2021-12-29 06:43:14,304 .95[47.774 ms] +2021-12-29 06:43:14,304 .99[47.774 ms] +2021-12-29 06:43:14,304 Channel (server worker num[20]): +2021-12-29 06:43:14,305 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:43:14,306 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:43:30,408 ==================== TRACER ====================== +2021-12-29 06:43:30,409 Channel (server worker num[20]): +2021-12-29 06:43:30,410 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:43:30,410 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:43:44,336 ==================== TRACER ====================== +2021-12-29 06:43:44,337 Channel (server worker num[20]): +2021-12-29 06:43:44,338 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:43:44,338 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:44:00,416 ==================== TRACER ====================== +2021-12-29 06:44:00,417 Channel (server worker num[20]): +2021-12-29 06:44:00,418 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:44:00,418 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:44:14,368 ==================== TRACER ====================== +2021-12-29 06:44:14,369 Channel (server worker num[20]): +2021-12-29 06:44:14,370 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:44:14,371 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:44:30,449 ==================== TRACER ====================== +2021-12-29 06:44:30,450 Channel (server worker num[20]): +2021-12-29 06:44:30,450 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:44:30,451 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:44:34,336 ==================== TRACER ====================== +2021-12-29 06:44:34,338 Channel (server worker num[20]): +2021-12-29 06:44:34,339 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:44:34,340 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:45:00,481 ==================== TRACER ====================== +2021-12-29 06:45:00,482 Channel (server worker num[20]): +2021-12-29 06:45:00,483 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:45:00,484 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:45:04,370 ==================== TRACER ====================== +2021-12-29 06:45:04,371 DAGExecutor: +2021-12-29 06:45:04,372 Query count[1] +2021-12-29 06:45:04,372 QPS[0.03333333333333333 q/s] +2021-12-29 06:45:04,372 Succ[0.0] +2021-12-29 06:45:04,372 Error req[0] +2021-12-29 06:45:04,373 Latency: +2021-12-29 06:45:04,373 ave[52.253 ms] +2021-12-29 06:45:04,373 .50[52.253 ms] +2021-12-29 06:45:04,373 .60[52.253 ms] +2021-12-29 06:45:04,373 .70[52.253 ms] +2021-12-29 06:45:04,373 .80[52.253 ms] +2021-12-29 06:45:04,374 .90[52.253 ms] +2021-12-29 06:45:04,374 .95[52.253 ms] +2021-12-29 06:45:04,374 .99[52.253 ms] +2021-12-29 06:45:04,374 Channel (server worker num[20]): +2021-12-29 06:45:04,375 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:45:04,376 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:45:30,514 ==================== TRACER ====================== +2021-12-29 06:45:30,515 Channel (server worker num[20]): +2021-12-29 06:45:30,516 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:45:30,517 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:45:34,406 ==================== TRACER ====================== +2021-12-29 06:45:34,407 Channel (server worker num[20]): +2021-12-29 06:45:34,408 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:45:34,408 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:46:00,547 ==================== TRACER ====================== +2021-12-29 06:46:00,548 Channel (server worker num[20]): +2021-12-29 06:46:00,548 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:46:00,549 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:46:04,425 ==================== TRACER ====================== +2021-12-29 06:46:04,426 Channel (server worker num[20]): +2021-12-29 06:46:04,427 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:46:04,428 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:46:19,150 ==================== TRACER ====================== +2021-12-29 06:46:19,152 Channel (server worker num[20]): +2021-12-29 06:46:19,155 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:46:19,155 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:46:30,579 ==================== TRACER ====================== +2021-12-29 06:46:30,580 Channel (server worker num[20]): +2021-12-29 06:46:30,581 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:46:30,582 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:46:49,186 ==================== TRACER ====================== +2021-12-29 06:46:49,187 DAGExecutor: +2021-12-29 06:46:49,187 Query count[1] +2021-12-29 06:46:49,188 QPS[0.03333333333333333 q/s] +2021-12-29 06:46:49,188 Succ[0.0] +2021-12-29 06:46:49,188 Error req[0] +2021-12-29 06:46:49,188 Latency: +2021-12-29 06:46:49,188 ave[98.181 ms] +2021-12-29 06:46:49,189 .50[98.181 ms] +2021-12-29 06:46:49,189 .60[98.181 ms] +2021-12-29 06:46:49,189 .70[98.181 ms] +2021-12-29 06:46:49,189 .80[98.181 ms] +2021-12-29 06:46:49,189 .90[98.181 ms] +2021-12-29 06:46:49,190 .95[98.181 ms] +2021-12-29 06:46:49,190 .99[98.181 ms] +2021-12-29 06:46:49,190 Channel (server worker num[20]): +2021-12-29 06:46:49,191 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:46:49,191 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:47:00,612 ==================== TRACER ====================== +2021-12-29 06:47:00,613 Channel (server worker num[20]): +2021-12-29 06:47:00,614 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:47:00,614 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:47:19,222 ==================== TRACER ====================== +2021-12-29 06:47:19,223 Channel (server worker num[20]): +2021-12-29 06:47:19,223 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:47:19,224 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:47:30,639 ==================== TRACER ====================== +2021-12-29 06:47:30,640 Channel (server worker num[20]): +2021-12-29 06:47:30,641 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:47:30,641 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:47:49,255 ==================== TRACER ====================== +2021-12-29 06:47:49,255 Channel (server worker num[20]): +2021-12-29 06:47:49,256 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:47:49,257 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:48:00,672 ==================== TRACER ====================== +2021-12-29 06:48:00,673 Channel (server worker num[20]): +2021-12-29 06:48:00,674 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:48:00,674 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:48:19,287 ==================== TRACER ====================== +2021-12-29 06:48:19,288 Channel (server worker num[20]): +2021-12-29 06:48:19,289 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:48:19,290 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:48:30,705 ==================== TRACER ====================== +2021-12-29 06:48:30,705 Channel (server worker num[20]): +2021-12-29 06:48:30,706 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:48:30,707 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:48:49,320 ==================== TRACER ====================== +2021-12-29 06:48:49,321 Channel (server worker num[20]): +2021-12-29 06:48:49,322 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:48:49,323 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:49:00,737 ==================== TRACER ====================== +2021-12-29 06:49:00,738 Channel (server worker num[20]): +2021-12-29 06:49:00,739 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:49:00,740 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:49:19,353 ==================== TRACER ====================== +2021-12-29 06:49:19,354 Channel (server worker num[20]): +2021-12-29 06:49:19,355 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:49:19,355 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:49:30,770 ==================== TRACER ====================== +2021-12-29 06:49:30,771 Channel (server worker num[20]): +2021-12-29 06:49:30,772 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:49:30,772 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:49:49,386 ==================== TRACER ====================== +2021-12-29 06:49:49,386 Channel (server worker num[20]): +2021-12-29 06:49:49,387 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:49:49,388 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:50:00,803 ==================== TRACER ====================== +2021-12-29 06:50:00,804 Channel (server worker num[20]): +2021-12-29 06:50:00,805 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:50:00,805 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:50:19,418 ==================== TRACER ====================== +2021-12-29 06:50:19,419 Channel (server worker num[20]): +2021-12-29 06:50:19,420 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:50:19,421 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:50:30,808 ==================== TRACER ====================== +2021-12-29 06:50:30,809 Channel (server worker num[20]): +2021-12-29 06:50:30,810 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:50:30,810 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:50:49,451 ==================== TRACER ====================== +2021-12-29 06:50:49,452 Channel (server worker num[20]): +2021-12-29 06:50:49,453 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:50:49,454 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:51:00,841 ==================== TRACER ====================== +2021-12-29 06:51:00,841 Channel (server worker num[20]): +2021-12-29 06:51:00,842 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:51:00,843 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:51:01,177 ==================== TRACER ====================== +2021-12-29 06:51:01,179 Channel (server worker num[20]): +2021-12-29 06:51:01,181 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:51:01,182 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:51:30,858 ==================== TRACER ====================== +2021-12-29 06:51:30,859 Channel (server worker num[20]): +2021-12-29 06:51:30,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:51:30,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:51:31,212 ==================== TRACER ====================== +2021-12-29 06:51:31,213 DAGExecutor: +2021-12-29 06:51:31,213 Query count[1] +2021-12-29 06:51:31,214 QPS[0.03333333333333333 q/s] +2021-12-29 06:51:31,214 Succ[0.0] +2021-12-29 06:51:31,214 Error req[0] +2021-12-29 06:51:31,214 Latency: +2021-12-29 06:51:31,214 ave[1632.321 ms] +2021-12-29 06:51:31,215 .50[1632.321 ms] +2021-12-29 06:51:31,215 .60[1632.321 ms] +2021-12-29 06:51:31,215 .70[1632.321 ms] +2021-12-29 06:51:31,215 .80[1632.321 ms] +2021-12-29 06:51:31,215 .90[1632.321 ms] +2021-12-29 06:51:31,216 .95[1632.321 ms] +2021-12-29 06:51:31,216 .99[1632.321 ms] +2021-12-29 06:51:31,216 Channel (server worker num[20]): +2021-12-29 06:51:31,217 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:51:31,217 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:51:57,055 ==================== TRACER ====================== +2021-12-29 06:51:57,056 Channel (server worker num[20]): +2021-12-29 06:51:57,059 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:51:57,059 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:52:00,865 ==================== TRACER ====================== +2021-12-29 06:52:00,867 Channel (server worker num[20]): +2021-12-29 06:52:00,867 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:52:00,868 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:52:13,972 ==================== TRACER ====================== +2021-12-29 06:52:13,974 Channel (server worker num[20]): +2021-12-29 06:52:13,977 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:52:13,977 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:52:30,883 ==================== TRACER ====================== +2021-12-29 06:52:30,884 Channel (server worker num[20]): +2021-12-29 06:52:30,885 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:52:30,886 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:52:44,008 ==================== TRACER ====================== +2021-12-29 06:52:44,009 DAGExecutor: +2021-12-29 06:52:44,009 Query count[1] +2021-12-29 06:52:44,009 QPS[0.03333333333333333 q/s] +2021-12-29 06:52:44,009 Succ[0.0] +2021-12-29 06:52:44,010 Error req[0] +2021-12-29 06:52:44,010 Latency: +2021-12-29 06:52:44,010 ave[1637.304 ms] +2021-12-29 06:52:44,010 .50[1637.304 ms] +2021-12-29 06:52:44,010 .60[1637.304 ms] +2021-12-29 06:52:44,011 .70[1637.304 ms] +2021-12-29 06:52:44,011 .80[1637.304 ms] +2021-12-29 06:52:44,011 .90[1637.304 ms] +2021-12-29 06:52:44,011 .95[1637.304 ms] +2021-12-29 06:52:44,011 .99[1637.304 ms] +2021-12-29 06:52:44,012 Channel (server worker num[20]): +2021-12-29 06:52:44,012 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:52:44,013 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:53:00,892 ==================== TRACER ====================== +2021-12-29 06:53:00,893 Channel (server worker num[20]): +2021-12-29 06:53:00,894 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:53:00,895 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:53:14,044 ==================== TRACER ====================== +2021-12-29 06:53:14,044 Channel (server worker num[20]): +2021-12-29 06:53:14,045 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:53:14,046 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:53:30,906 ==================== TRACER ====================== +2021-12-29 06:53:30,907 Channel (server worker num[20]): +2021-12-29 06:53:30,908 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:53:30,909 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:53:44,067 ==================== TRACER ====================== +2021-12-29 06:53:44,068 Channel (server worker num[20]): +2021-12-29 06:53:44,069 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:53:44,070 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:54:00,738 ==================== TRACER ====================== +2021-12-29 06:54:00,740 Channel (server worker num[20]): +2021-12-29 06:54:00,742 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:54:00,743 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:54:00,939 ==================== TRACER ====================== +2021-12-29 06:54:00,940 Channel (server worker num[20]): +2021-12-29 06:54:00,941 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:54:00,941 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:54:30,768 ==================== TRACER ====================== +2021-12-29 06:54:30,769 DAGExecutor: +2021-12-29 06:54:30,769 Query count[1] +2021-12-29 06:54:30,769 QPS[0.03333333333333333 q/s] +2021-12-29 06:54:30,770 Succ[0.0] +2021-12-29 06:54:30,770 Error req[0] +2021-12-29 06:54:30,770 Latency: +2021-12-29 06:54:30,770 ave[1731.024 ms] +2021-12-29 06:54:30,770 .50[1731.024 ms] +2021-12-29 06:54:30,770 .60[1731.024 ms] +2021-12-29 06:54:30,771 .70[1731.024 ms] +2021-12-29 06:54:30,771 .80[1731.024 ms] +2021-12-29 06:54:30,771 .90[1731.024 ms] +2021-12-29 06:54:30,771 .95[1731.024 ms] +2021-12-29 06:54:30,771 .99[1731.024 ms] +2021-12-29 06:54:30,772 Channel (server worker num[20]): +2021-12-29 06:54:30,772 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:54:30,773 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:54:30,972 ==================== TRACER ====================== +2021-12-29 06:54:30,972 Channel (server worker num[20]): +2021-12-29 06:54:30,973 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:54:30,974 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:55:00,800 ==================== TRACER ====================== +2021-12-29 06:55:00,801 Channel (server worker num[20]): +2021-12-29 06:55:00,804 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:55:00,804 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:55:01,004 ==================== TRACER ====================== +2021-12-29 06:55:01,005 Channel (server worker num[20]): +2021-12-29 06:55:01,006 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:55:01,007 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:55:30,835 ==================== TRACER ====================== +2021-12-29 06:55:30,836 Channel (server worker num[20]): +2021-12-29 06:55:30,836 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:55:30,837 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:55:31,037 ==================== TRACER ====================== +2021-12-29 06:55:31,038 Channel (server worker num[20]): +2021-12-29 06:55:31,039 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:55:31,040 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:56:00,858 ==================== TRACER ====================== +2021-12-29 06:56:00,859 Channel (server worker num[20]): +2021-12-29 06:56:00,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:56:00,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:56:01,063 ==================== TRACER ====================== +2021-12-29 06:56:01,064 Channel (server worker num[20]): +2021-12-29 06:56:01,065 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:56:01,065 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:56:30,888 ==================== TRACER ====================== +2021-12-29 06:56:30,889 Channel (server worker num[20]): +2021-12-29 06:56:30,890 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:56:30,891 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:56:31,083 ==================== TRACER ====================== +2021-12-29 06:56:31,084 Channel (server worker num[20]): +2021-12-29 06:56:31,084 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:56:31,085 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:57:00,912 ==================== TRACER ====================== +2021-12-29 06:57:00,913 Channel (server worker num[20]): +2021-12-29 06:57:00,914 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:57:00,915 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:57:01,116 ==================== TRACER ====================== +2021-12-29 06:57:01,116 Channel (server worker num[20]): +2021-12-29 06:57:01,117 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:57:01,118 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:57:30,945 ==================== TRACER ====================== +2021-12-29 06:57:30,946 Channel (server worker num[20]): +2021-12-29 06:57:30,947 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:57:30,947 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:57:31,148 ==================== TRACER ====================== +2021-12-29 06:57:31,149 Channel (server worker num[20]): +2021-12-29 06:57:31,150 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:57:31,151 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:58:00,978 ==================== TRACER ====================== +2021-12-29 06:58:00,979 Channel (server worker num[20]): +2021-12-29 06:58:00,981 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:58:00,982 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:58:01,181 ==================== TRACER ====================== +2021-12-29 06:58:01,182 Channel (server worker num[20]): +2021-12-29 06:58:01,183 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:58:01,183 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:58:31,012 ==================== TRACER ====================== +2021-12-29 06:58:31,013 Channel (server worker num[20]): +2021-12-29 06:58:31,014 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:58:31,014 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:58:31,214 ==================== TRACER ====================== +2021-12-29 06:58:31,214 Channel (server worker num[20]): +2021-12-29 06:58:31,215 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:58:31,216 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:59:01,045 ==================== TRACER ====================== +2021-12-29 06:59:01,046 Channel (server worker num[20]): +2021-12-29 06:59:01,047 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:59:01,047 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:59:01,246 ==================== TRACER ====================== +2021-12-29 06:59:01,247 Channel (server worker num[20]): +2021-12-29 06:59:01,248 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:59:01,249 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:59:31,076 ==================== TRACER ====================== +2021-12-29 06:59:31,077 Channel (server worker num[20]): +2021-12-29 06:59:31,078 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:59:31,079 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 06:59:31,260 ==================== TRACER ====================== +2021-12-29 06:59:31,261 Channel (server worker num[20]): +2021-12-29 06:59:31,262 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 06:59:31,262 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:00:01,088 ==================== TRACER ====================== +2021-12-29 07:00:01,088 Channel (server worker num[20]): +2021-12-29 07:00:01,089 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:00:01,090 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:00:01,293 ==================== TRACER ====================== +2021-12-29 07:00:01,294 Channel (server worker num[20]): +2021-12-29 07:00:01,294 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:00:01,295 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:00:31,120 ==================== TRACER ====================== +2021-12-29 07:00:31,121 Channel (server worker num[20]): +2021-12-29 07:00:31,122 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:00:31,123 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:00:31,324 ==================== TRACER ====================== +2021-12-29 07:00:31,325 Channel (server worker num[20]): +2021-12-29 07:00:31,326 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:00:31,326 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:01:01,153 ==================== TRACER ====================== +2021-12-29 07:01:01,154 Channel (server worker num[20]): +2021-12-29 07:01:01,155 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:01:01,156 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:01:01,357 ==================== TRACER ====================== +2021-12-29 07:01:01,358 Channel (server worker num[20]): +2021-12-29 07:01:01,358 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:01:01,359 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:01:31,186 ==================== TRACER ====================== +2021-12-29 07:01:31,187 Channel (server worker num[20]): +2021-12-29 07:01:31,188 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:01:31,188 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:01:31,389 ==================== TRACER ====================== +2021-12-29 07:01:31,390 Channel (server worker num[20]): +2021-12-29 07:01:31,391 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:01:31,392 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:02:01,219 ==================== TRACER ====================== +2021-12-29 07:02:01,220 Channel (server worker num[20]): +2021-12-29 07:02:01,220 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:02:01,221 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:02:01,416 ==================== TRACER ====================== +2021-12-29 07:02:01,417 Channel (server worker num[20]): +2021-12-29 07:02:01,418 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:02:01,418 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:02:31,251 ==================== TRACER ====================== +2021-12-29 07:02:31,252 Channel (server worker num[20]): +2021-12-29 07:02:31,253 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:02:31,254 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:02:31,448 ==================== TRACER ====================== +2021-12-29 07:02:31,449 Channel (server worker num[20]): +2021-12-29 07:02:31,450 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:02:31,450 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:03:01,284 ==================== TRACER ====================== +2021-12-29 07:03:01,285 Channel (server worker num[20]): +2021-12-29 07:03:01,286 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:03:01,287 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:03:01,481 ==================== TRACER ====================== +2021-12-29 07:03:01,481 Channel (server worker num[20]): +2021-12-29 07:03:01,482 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:03:01,483 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:03:31,316 ==================== TRACER ====================== +2021-12-29 07:03:31,317 Channel (server worker num[20]): +2021-12-29 07:03:31,318 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:03:31,318 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:03:31,513 ==================== TRACER ====================== +2021-12-29 07:03:31,514 Channel (server worker num[20]): +2021-12-29 07:03:31,515 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:03:31,516 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:04:01,349 ==================== TRACER ====================== +2021-12-29 07:04:01,350 Channel (server worker num[20]): +2021-12-29 07:04:01,351 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:04:01,351 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:04:01,546 ==================== TRACER ====================== +2021-12-29 07:04:01,547 Channel (server worker num[20]): +2021-12-29 07:04:01,548 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:04:01,548 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:04:31,381 ==================== TRACER ====================== +2021-12-29 07:04:31,382 Channel (server worker num[20]): +2021-12-29 07:04:31,383 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:04:31,384 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:04:31,579 ==================== TRACER ====================== +2021-12-29 07:04:31,579 Channel (server worker num[20]): +2021-12-29 07:04:31,580 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:04:31,581 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:05:01,414 ==================== TRACER ====================== +2021-12-29 07:05:01,415 Channel (server worker num[20]): +2021-12-29 07:05:01,416 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:05:01,416 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:05:01,611 ==================== TRACER ====================== +2021-12-29 07:05:01,612 Channel (server worker num[20]): +2021-12-29 07:05:01,613 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:05:01,614 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:05:31,447 ==================== TRACER ====================== +2021-12-29 07:05:31,448 Channel (server worker num[20]): +2021-12-29 07:05:31,448 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:05:31,449 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:05:31,644 ==================== TRACER ====================== +2021-12-29 07:05:31,645 Channel (server worker num[20]): +2021-12-29 07:05:31,646 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:05:31,646 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:06:01,479 ==================== TRACER ====================== +2021-12-29 07:06:01,480 Channel (server worker num[20]): +2021-12-29 07:06:01,481 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:06:01,482 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:06:01,649 ==================== TRACER ====================== +2021-12-29 07:06:01,650 Channel (server worker num[20]): +2021-12-29 07:06:01,651 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:06:01,652 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:06:31,512 ==================== TRACER ====================== +2021-12-29 07:06:31,513 Channel (server worker num[20]): +2021-12-29 07:06:31,514 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:06:31,515 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:06:31,680 ==================== TRACER ====================== +2021-12-29 07:06:31,681 Channel (server worker num[20]): +2021-12-29 07:06:31,682 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:06:31,682 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:07:01,545 ==================== TRACER ====================== +2021-12-29 07:07:01,546 Channel (server worker num[20]): +2021-12-29 07:07:01,547 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:07:01,547 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:07:01,713 ==================== TRACER ====================== +2021-12-29 07:07:01,713 Channel (server worker num[20]): +2021-12-29 07:07:01,714 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:07:01,715 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:07:31,578 ==================== TRACER ====================== +2021-12-29 07:07:31,579 Channel (server worker num[20]): +2021-12-29 07:07:31,579 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:07:31,580 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:07:31,745 ==================== TRACER ====================== +2021-12-29 07:07:31,746 Channel (server worker num[20]): +2021-12-29 07:07:31,747 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:07:31,748 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:08:01,610 ==================== TRACER ====================== +2021-12-29 07:08:01,611 Channel (server worker num[20]): +2021-12-29 07:08:01,612 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:08:01,613 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:08:01,778 ==================== TRACER ====================== +2021-12-29 07:08:01,779 Channel (server worker num[20]): +2021-12-29 07:08:01,780 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:08:01,780 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:08:31,643 ==================== TRACER ====================== +2021-12-29 07:08:31,644 Channel (server worker num[20]): +2021-12-29 07:08:31,645 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:08:31,646 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:08:31,803 ==================== TRACER ====================== +2021-12-29 07:08:31,804 Channel (server worker num[20]): +2021-12-29 07:08:31,804 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:08:31,805 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:09:01,676 ==================== TRACER ====================== +2021-12-29 07:09:01,677 Channel (server worker num[20]): +2021-12-29 07:09:01,678 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:09:01,678 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:09:01,836 ==================== TRACER ====================== +2021-12-29 07:09:01,836 Channel (server worker num[20]): +2021-12-29 07:09:01,837 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:09:01,838 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:09:31,709 ==================== TRACER ====================== +2021-12-29 07:09:31,710 Channel (server worker num[20]): +2021-12-29 07:09:31,710 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:09:31,711 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:09:31,868 ==================== TRACER ====================== +2021-12-29 07:09:31,869 Channel (server worker num[20]): +2021-12-29 07:09:31,870 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:09:31,871 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:10:01,742 ==================== TRACER ====================== +2021-12-29 07:10:01,742 Channel (server worker num[20]): +2021-12-29 07:10:01,745 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:10:01,746 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:10:01,888 ==================== TRACER ====================== +2021-12-29 07:10:01,888 Channel (server worker num[20]): +2021-12-29 07:10:01,889 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:10:01,890 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:10:31,776 ==================== TRACER ====================== +2021-12-29 07:10:31,777 Channel (server worker num[20]): +2021-12-29 07:10:31,778 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:10:31,778 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:10:31,920 ==================== TRACER ====================== +2021-12-29 07:10:31,921 Channel (server worker num[20]): +2021-12-29 07:10:31,922 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:10:31,922 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:11:01,809 ==================== TRACER ====================== +2021-12-29 07:11:01,810 Channel (server worker num[20]): +2021-12-29 07:11:01,810 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:11:01,811 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:11:01,953 ==================== TRACER ====================== +2021-12-29 07:11:01,953 Channel (server worker num[20]): +2021-12-29 07:11:01,954 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:11:01,955 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:11:31,841 ==================== TRACER ====================== +2021-12-29 07:11:31,842 Channel (server worker num[20]): +2021-12-29 07:11:31,843 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:11:31,844 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:11:31,985 ==================== TRACER ====================== +2021-12-29 07:11:31,986 Channel (server worker num[20]): +2021-12-29 07:11:31,987 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:11:31,988 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:12:01,874 ==================== TRACER ====================== +2021-12-29 07:12:01,875 Channel (server worker num[20]): +2021-12-29 07:12:01,876 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:12:01,876 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:12:02,018 ==================== TRACER ====================== +2021-12-29 07:12:02,019 Channel (server worker num[20]): +2021-12-29 07:12:02,019 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:12:02,020 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:12:31,885 ==================== TRACER ====================== +2021-12-29 07:12:31,886 Channel (server worker num[20]): +2021-12-29 07:12:31,887 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:12:31,888 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:12:32,035 ==================== TRACER ====================== +2021-12-29 07:12:32,036 Channel (server worker num[20]): +2021-12-29 07:12:32,037 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:12:32,038 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:13:01,918 ==================== TRACER ====================== +2021-12-29 07:13:01,919 Channel (server worker num[20]): +2021-12-29 07:13:01,920 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:13:01,920 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:13:02,068 ==================== TRACER ====================== +2021-12-29 07:13:02,069 Channel (server worker num[20]): +2021-12-29 07:13:02,069 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:13:02,070 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:13:12,292 ==================== TRACER ====================== +2021-12-29 07:13:12,294 Channel (server worker num[20]): +2021-12-29 07:13:12,296 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:13:12,297 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:13:32,101 ==================== TRACER ====================== +2021-12-29 07:13:32,101 Channel (server worker num[20]): +2021-12-29 07:13:32,102 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:13:32,103 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:13:42,327 ==================== TRACER ====================== +2021-12-29 07:13:42,329 DAGExecutor: +2021-12-29 07:13:42,329 Query count[1] +2021-12-29 07:13:42,329 QPS[0.03333333333333333 q/s] +2021-12-29 07:13:42,329 Succ[0.0] +2021-12-29 07:13:42,329 Error req[0] +2021-12-29 07:13:42,330 Latency: +2021-12-29 07:13:42,330 ave[1661.686 ms] +2021-12-29 07:13:42,330 .50[1661.686 ms] +2021-12-29 07:13:42,330 .60[1661.686 ms] +2021-12-29 07:13:42,330 .70[1661.686 ms] +2021-12-29 07:13:42,331 .80[1661.686 ms] +2021-12-29 07:13:42,331 .90[1661.686 ms] +2021-12-29 07:13:42,331 .95[1661.686 ms] +2021-12-29 07:13:42,331 .99[1661.686 ms] +2021-12-29 07:13:42,331 Channel (server worker num[20]): +2021-12-29 07:13:42,332 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:13:42,333 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:14:02,133 ==================== TRACER ====================== +2021-12-29 07:14:02,134 Channel (server worker num[20]): +2021-12-29 07:14:02,135 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:14:02,136 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:14:12,363 ==================== TRACER ====================== +2021-12-29 07:14:12,364 Channel (server worker num[20]): +2021-12-29 07:14:12,365 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:14:12,366 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:14:32,166 ==================== TRACER ====================== +2021-12-29 07:14:32,167 Channel (server worker num[20]): +2021-12-29 07:14:32,168 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:14:32,168 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:14:42,396 ==================== TRACER ====================== +2021-12-29 07:14:42,397 Channel (server worker num[20]): +2021-12-29 07:14:42,398 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:14:42,398 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:15:02,199 ==================== TRACER ====================== +2021-12-29 07:15:02,200 Channel (server worker num[20]): +2021-12-29 07:15:02,200 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:15:02,201 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:15:12,429 ==================== TRACER ====================== +2021-12-29 07:15:12,430 Channel (server worker num[20]): +2021-12-29 07:15:12,430 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:15:12,431 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:15:32,223 ==================== TRACER ====================== +2021-12-29 07:15:32,224 Channel (server worker num[20]): +2021-12-29 07:15:32,225 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:15:32,226 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:15:42,461 ==================== TRACER ====================== +2021-12-29 07:15:42,462 Channel (server worker num[20]): +2021-12-29 07:15:42,463 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:15:42,464 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:16:02,228 ==================== TRACER ====================== +2021-12-29 07:16:02,229 Channel (server worker num[20]): +2021-12-29 07:16:02,230 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:16:02,230 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:16:12,494 ==================== TRACER ====================== +2021-12-29 07:16:12,495 Channel (server worker num[20]): +2021-12-29 07:16:12,496 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:16:12,496 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:16:27,943 ==================== TRACER ====================== +2021-12-29 07:16:27,945 Channel (server worker num[20]): +2021-12-29 07:16:27,947 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:16:27,948 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:16:32,261 ==================== TRACER ====================== +2021-12-29 07:16:32,261 Channel (server worker num[20]): +2021-12-29 07:16:32,262 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:16:32,263 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:16:57,960 ==================== TRACER ====================== +2021-12-29 07:16:57,961 DAGExecutor: +2021-12-29 07:16:57,962 Query count[1] +2021-12-29 07:16:57,962 QPS[0.03333333333333333 q/s] +2021-12-29 07:16:57,962 Succ[0.0] +2021-12-29 07:16:57,962 Error req[0] +2021-12-29 07:16:57,962 Latency: +2021-12-29 07:16:57,962 ave[76.653 ms] +2021-12-29 07:16:57,963 .50[76.653 ms] +2021-12-29 07:16:57,963 .60[76.653 ms] +2021-12-29 07:16:57,963 .70[76.653 ms] +2021-12-29 07:16:57,963 .80[76.653 ms] +2021-12-29 07:16:57,963 .90[76.653 ms] +2021-12-29 07:16:57,963 .95[76.653 ms] +2021-12-29 07:16:57,964 .99[76.653 ms] +2021-12-29 07:16:57,964 Channel (server worker num[20]): +2021-12-29 07:16:57,965 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:16:57,965 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:17:02,293 ==================== TRACER ====================== +2021-12-29 07:17:02,294 Channel (server worker num[20]): +2021-12-29 07:17:02,295 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:17:02,295 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:17:27,996 ==================== TRACER ====================== +2021-12-29 07:17:27,997 Channel (server worker num[20]): +2021-12-29 07:17:27,997 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:17:27,998 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:17:32,326 ==================== TRACER ====================== +2021-12-29 07:17:32,327 Channel (server worker num[20]): +2021-12-29 07:17:32,327 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:17:32,329 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:17:33,622 ==================== TRACER ====================== +2021-12-29 07:17:33,624 Channel (server worker num[20]): +2021-12-29 07:17:33,626 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:17:33,626 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:18:02,359 ==================== TRACER ====================== +2021-12-29 07:18:02,360 Channel (server worker num[20]): +2021-12-29 07:18:02,361 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:18:02,362 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:18:03,656 ==================== TRACER ====================== +2021-12-29 07:18:03,658 DAGExecutor: +2021-12-29 07:18:03,658 Query count[1] +2021-12-29 07:18:03,658 QPS[0.03333333333333333 q/s] +2021-12-29 07:18:03,658 Succ[0.0] +2021-12-29 07:18:03,658 Error req[0] +2021-12-29 07:18:03,659 Latency: +2021-12-29 07:18:03,659 ave[1646.997 ms] +2021-12-29 07:18:03,659 .50[1646.997 ms] +2021-12-29 07:18:03,659 .60[1646.997 ms] +2021-12-29 07:18:03,659 .70[1646.997 ms] +2021-12-29 07:18:03,659 .80[1646.997 ms] +2021-12-29 07:18:03,660 .90[1646.997 ms] +2021-12-29 07:18:03,660 .95[1646.997 ms] +2021-12-29 07:18:03,660 .99[1646.997 ms] +2021-12-29 07:18:03,660 Channel (server worker num[20]): +2021-12-29 07:18:03,661 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:18:03,662 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:18:32,392 ==================== TRACER ====================== +2021-12-29 07:18:32,393 Channel (server worker num[20]): +2021-12-29 07:18:32,394 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:18:32,395 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:18:33,692 ==================== TRACER ====================== +2021-12-29 07:18:33,693 Channel (server worker num[20]): +2021-12-29 07:18:33,694 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:18:33,694 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:19:02,425 ==================== TRACER ====================== +2021-12-29 07:19:02,426 Channel (server worker num[20]): +2021-12-29 07:19:02,427 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:19:02,427 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:19:03,718 ==================== TRACER ====================== +2021-12-29 07:19:03,718 Channel (server worker num[20]): +2021-12-29 07:19:03,719 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:19:03,720 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:19:06,252 ==================== TRACER ====================== +2021-12-29 07:19:06,254 Channel (server worker num[20]): +2021-12-29 07:19:06,256 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:19:06,256 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:19:32,457 ==================== TRACER ====================== +2021-12-29 07:19:32,458 Channel (server worker num[20]): +2021-12-29 07:19:32,459 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:19:32,459 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:19:36,278 ==================== TRACER ====================== +2021-12-29 07:19:36,279 DAGExecutor: +2021-12-29 07:19:36,279 Query count[1] +2021-12-29 07:19:36,279 QPS[0.03333333333333333 q/s] +2021-12-29 07:19:36,280 Succ[0.0] +2021-12-29 07:19:36,280 Error req[0] +2021-12-29 07:19:36,280 Latency: +2021-12-29 07:19:36,280 ave[1630.707 ms] +2021-12-29 07:19:36,280 .50[1630.707 ms] +2021-12-29 07:19:36,280 .60[1630.707 ms] +2021-12-29 07:19:36,281 .70[1630.707 ms] +2021-12-29 07:19:36,281 .80[1630.707 ms] +2021-12-29 07:19:36,281 .90[1630.707 ms] +2021-12-29 07:19:36,281 .95[1630.707 ms] +2021-12-29 07:19:36,281 .99[1630.707 ms] +2021-12-29 07:19:36,282 Channel (server worker num[20]): +2021-12-29 07:19:36,282 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:19:36,283 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:20:02,490 ==================== TRACER ====================== +2021-12-29 07:20:02,490 Channel (server worker num[20]): +2021-12-29 07:20:02,491 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:20:02,492 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:20:06,313 ==================== TRACER ====================== +2021-12-29 07:20:06,314 Channel (server worker num[20]): +2021-12-29 07:20:06,315 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:20:06,316 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:20:32,520 ==================== TRACER ====================== +2021-12-29 07:20:32,521 Channel (server worker num[20]): +2021-12-29 07:20:32,522 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:20:32,523 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:20:36,346 ==================== TRACER ====================== +2021-12-29 07:20:36,347 Channel (server worker num[20]): +2021-12-29 07:20:36,348 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:20:36,349 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:21:02,553 ==================== TRACER ====================== +2021-12-29 07:21:02,554 Channel (server worker num[20]): +2021-12-29 07:21:02,555 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:21:02,555 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:21:06,379 ==================== TRACER ====================== +2021-12-29 07:21:06,380 Channel (server worker num[20]): +2021-12-29 07:21:06,381 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:21:06,381 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:21:32,586 ==================== TRACER ====================== +2021-12-29 07:21:32,587 Channel (server worker num[20]): +2021-12-29 07:21:32,587 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:21:32,588 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:21:36,412 ==================== TRACER ====================== +2021-12-29 07:21:36,413 Channel (server worker num[20]): +2021-12-29 07:21:36,413 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:21:36,414 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:22:02,618 ==================== TRACER ====================== +2021-12-29 07:22:02,619 Channel (server worker num[20]): +2021-12-29 07:22:02,620 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:22:02,621 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:22:06,445 ==================== TRACER ====================== +2021-12-29 07:22:06,445 Channel (server worker num[20]): +2021-12-29 07:22:06,446 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:22:06,447 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:22:32,651 ==================== TRACER ====================== +2021-12-29 07:22:32,652 Channel (server worker num[20]): +2021-12-29 07:22:32,653 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:22:32,654 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:22:36,477 ==================== TRACER ====================== +2021-12-29 07:22:36,478 Channel (server worker num[20]): +2021-12-29 07:22:36,479 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:22:36,480 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:23:02,664 ==================== TRACER ====================== +2021-12-29 07:23:02,665 Channel (server worker num[20]): +2021-12-29 07:23:02,666 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:23:02,666 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:23:06,510 ==================== TRACER ====================== +2021-12-29 07:23:06,511 Channel (server worker num[20]): +2021-12-29 07:23:06,512 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:23:06,513 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:23:32,680 ==================== TRACER ====================== +2021-12-29 07:23:32,681 Channel (server worker num[20]): +2021-12-29 07:23:32,682 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:23:32,683 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:23:36,543 ==================== TRACER ====================== +2021-12-29 07:23:36,544 Channel (server worker num[20]): +2021-12-29 07:23:36,545 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:23:36,545 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:24:02,713 ==================== TRACER ====================== +2021-12-29 07:24:02,714 Channel (server worker num[20]): +2021-12-29 07:24:02,715 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:24:02,715 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:24:06,576 ==================== TRACER ====================== +2021-12-29 07:24:06,577 Channel (server worker num[20]): +2021-12-29 07:24:06,577 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:24:06,578 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:24:32,746 ==================== TRACER ====================== +2021-12-29 07:24:32,746 Channel (server worker num[20]): +2021-12-29 07:24:32,747 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:24:32,748 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:24:36,609 ==================== TRACER ====================== +2021-12-29 07:24:36,609 Channel (server worker num[20]): +2021-12-29 07:24:36,610 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:24:36,611 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:25:02,778 ==================== TRACER ====================== +2021-12-29 07:25:02,779 Channel (server worker num[20]): +2021-12-29 07:25:02,780 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:25:02,781 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:25:06,641 ==================== TRACER ====================== +2021-12-29 07:25:06,642 Channel (server worker num[20]): +2021-12-29 07:25:06,643 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:25:06,644 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:25:32,796 ==================== TRACER ====================== +2021-12-29 07:25:32,797 Channel (server worker num[20]): +2021-12-29 07:25:32,798 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:25:32,798 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:25:36,674 ==================== TRACER ====================== +2021-12-29 07:25:36,675 Channel (server worker num[20]): +2021-12-29 07:25:36,676 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:25:36,677 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:26:02,829 ==================== TRACER ====================== +2021-12-29 07:26:02,830 Channel (server worker num[20]): +2021-12-29 07:26:02,830 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:26:02,831 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:26:06,707 ==================== TRACER ====================== +2021-12-29 07:26:06,708 Channel (server worker num[20]): +2021-12-29 07:26:06,708 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:26:06,709 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:26:32,861 ==================== TRACER ====================== +2021-12-29 07:26:32,862 Channel (server worker num[20]): +2021-12-29 07:26:32,863 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:26:32,864 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:26:36,740 ==================== TRACER ====================== +2021-12-29 07:26:36,740 Channel (server worker num[20]): +2021-12-29 07:26:36,741 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:26:36,742 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:27:02,894 ==================== TRACER ====================== +2021-12-29 07:27:02,895 Channel (server worker num[20]): +2021-12-29 07:27:02,896 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:27:02,896 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:27:06,772 ==================== TRACER ====================== +2021-12-29 07:27:06,773 Channel (server worker num[20]): +2021-12-29 07:27:06,774 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:27:06,775 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:27:32,927 ==================== TRACER ====================== +2021-12-29 07:27:32,928 Channel (server worker num[20]): +2021-12-29 07:27:32,928 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:27:32,929 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:27:36,805 ==================== TRACER ====================== +2021-12-29 07:27:36,806 Channel (server worker num[20]): +2021-12-29 07:27:36,807 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:27:36,807 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:28:02,959 ==================== TRACER ====================== +2021-12-29 07:28:02,960 Channel (server worker num[20]): +2021-12-29 07:28:02,961 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:28:02,962 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:28:06,838 ==================== TRACER ====================== +2021-12-29 07:28:06,839 Channel (server worker num[20]): +2021-12-29 07:28:06,839 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:28:06,840 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:28:32,992 ==================== TRACER ====================== +2021-12-29 07:28:32,993 Channel (server worker num[20]): +2021-12-29 07:28:32,993 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:28:32,994 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:28:36,871 ==================== TRACER ====================== +2021-12-29 07:28:36,871 Channel (server worker num[20]): +2021-12-29 07:28:36,872 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:28:36,873 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:29:03,024 ==================== TRACER ====================== +2021-12-29 07:29:03,025 Channel (server worker num[20]): +2021-12-29 07:29:03,026 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:29:03,027 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:29:06,903 ==================== TRACER ====================== +2021-12-29 07:29:06,904 Channel (server worker num[20]): +2021-12-29 07:29:06,905 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:29:06,906 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:29:33,057 ==================== TRACER ====================== +2021-12-29 07:29:33,058 Channel (server worker num[20]): +2021-12-29 07:29:33,059 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:29:33,060 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:29:36,912 ==================== TRACER ====================== +2021-12-29 07:29:36,913 Channel (server worker num[20]): +2021-12-29 07:29:36,914 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:29:36,914 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:30:03,090 ==================== TRACER ====================== +2021-12-29 07:30:03,091 Channel (server worker num[20]): +2021-12-29 07:30:03,092 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:30:03,092 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:30:06,945 ==================== TRACER ====================== +2021-12-29 07:30:06,946 Channel (server worker num[20]): +2021-12-29 07:30:06,946 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:30:06,947 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:30:33,123 ==================== TRACER ====================== +2021-12-29 07:30:33,123 Channel (server worker num[20]): +2021-12-29 07:30:33,124 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:30:33,125 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:30:36,974 ==================== TRACER ====================== +2021-12-29 07:30:36,975 Channel (server worker num[20]): +2021-12-29 07:30:36,976 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:30:36,977 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:31:03,155 ==================== TRACER ====================== +2021-12-29 07:31:03,156 Channel (server worker num[20]): +2021-12-29 07:31:03,157 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:31:03,158 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:31:07,007 ==================== TRACER ====================== +2021-12-29 07:31:07,008 Channel (server worker num[20]): +2021-12-29 07:31:07,009 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:31:07,009 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:31:33,188 ==================== TRACER ====================== +2021-12-29 07:31:33,189 Channel (server worker num[20]): +2021-12-29 07:31:33,190 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:31:33,190 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:31:37,029 ==================== TRACER ====================== +2021-12-29 07:31:37,029 Channel (server worker num[20]): +2021-12-29 07:31:37,030 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:31:37,031 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:32:03,208 ==================== TRACER ====================== +2021-12-29 07:32:03,208 Channel (server worker num[20]): +2021-12-29 07:32:03,209 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:32:03,210 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:32:07,061 ==================== TRACER ====================== +2021-12-29 07:32:07,062 Channel (server worker num[20]): +2021-12-29 07:32:07,063 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:32:07,064 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:32:33,217 ==================== TRACER ====================== +2021-12-29 07:32:33,218 Channel (server worker num[20]): +2021-12-29 07:32:33,219 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:32:33,219 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:32:37,094 ==================== TRACER ====================== +2021-12-29 07:32:37,095 Channel (server worker num[20]): +2021-12-29 07:32:37,095 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:32:37,096 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:33:03,250 ==================== TRACER ====================== +2021-12-29 07:33:03,251 Channel (server worker num[20]): +2021-12-29 07:33:03,251 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:33:03,252 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:33:07,116 ==================== TRACER ====================== +2021-12-29 07:33:07,117 Channel (server worker num[20]): +2021-12-29 07:33:07,118 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:33:07,119 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:33:33,282 ==================== TRACER ====================== +2021-12-29 07:33:33,283 Channel (server worker num[20]): +2021-12-29 07:33:33,284 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:33:33,285 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:33:37,119 ==================== TRACER ====================== +2021-12-29 07:33:37,120 Channel (server worker num[20]): +2021-12-29 07:33:37,121 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:33:37,122 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:34:03,312 ==================== TRACER ====================== +2021-12-29 07:34:03,313 Channel (server worker num[20]): +2021-12-29 07:34:03,314 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:34:03,314 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:34:07,152 ==================== TRACER ====================== +2021-12-29 07:34:07,153 Channel (server worker num[20]): +2021-12-29 07:34:07,154 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:34:07,154 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:34:33,345 ==================== TRACER ====================== +2021-12-29 07:34:33,346 Channel (server worker num[20]): +2021-12-29 07:34:33,346 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:34:33,347 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:34:37,185 ==================== TRACER ====================== +2021-12-29 07:34:37,186 Channel (server worker num[20]): +2021-12-29 07:34:37,186 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:34:37,187 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:35:03,366 ==================== TRACER ====================== +2021-12-29 07:35:03,367 Channel (server worker num[20]): +2021-12-29 07:35:03,368 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:35:03,369 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:35:07,218 ==================== TRACER ====================== +2021-12-29 07:35:07,218 Channel (server worker num[20]): +2021-12-29 07:35:07,221 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:35:07,222 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:35:33,378 ==================== TRACER ====================== +2021-12-29 07:35:33,379 Channel (server worker num[20]): +2021-12-29 07:35:33,380 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:35:33,381 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:35:37,252 ==================== TRACER ====================== +2021-12-29 07:35:37,253 Channel (server worker num[20]): +2021-12-29 07:35:37,254 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:35:37,255 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:36:03,411 ==================== TRACER ====================== +2021-12-29 07:36:03,412 Channel (server worker num[20]): +2021-12-29 07:36:03,413 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:36:03,413 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:36:07,257 ==================== TRACER ====================== +2021-12-29 07:36:07,257 Channel (server worker num[20]): +2021-12-29 07:36:07,260 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:36:07,261 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:36:33,444 ==================== TRACER ====================== +2021-12-29 07:36:33,445 Channel (server worker num[20]): +2021-12-29 07:36:33,445 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:36:33,446 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:36:37,291 ==================== TRACER ====================== +2021-12-29 07:36:37,292 Channel (server worker num[20]): +2021-12-29 07:36:37,293 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:36:37,294 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:37:03,477 ==================== TRACER ====================== +2021-12-29 07:37:03,477 Channel (server worker num[20]): +2021-12-29 07:37:03,478 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:37:03,479 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:37:07,324 ==================== TRACER ====================== +2021-12-29 07:37:07,325 Channel (server worker num[20]): +2021-12-29 07:37:07,326 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:37:07,326 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:37:33,509 ==================== TRACER ====================== +2021-12-29 07:37:33,510 Channel (server worker num[20]): +2021-12-29 07:37:33,511 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:37:33,512 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:37:37,357 ==================== TRACER ====================== +2021-12-29 07:37:37,358 Channel (server worker num[20]): +2021-12-29 07:37:37,358 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:37:37,359 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:38:03,542 ==================== TRACER ====================== +2021-12-29 07:38:03,543 Channel (server worker num[20]): +2021-12-29 07:38:03,544 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:38:03,545 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:38:07,365 ==================== TRACER ====================== +2021-12-29 07:38:07,365 Channel (server worker num[20]): +2021-12-29 07:38:07,366 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:38:07,367 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:38:33,565 ==================== TRACER ====================== +2021-12-29 07:38:33,566 Channel (server worker num[20]): +2021-12-29 07:38:33,567 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:38:33,568 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:38:37,397 ==================== TRACER ====================== +2021-12-29 07:38:37,398 Channel (server worker num[20]): +2021-12-29 07:38:37,399 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:38:37,400 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:39:03,592 ==================== TRACER ====================== +2021-12-29 07:39:03,593 Channel (server worker num[20]): +2021-12-29 07:39:03,594 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:39:03,595 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:39:07,430 ==================== TRACER ====================== +2021-12-29 07:39:07,431 Channel (server worker num[20]): +2021-12-29 07:39:07,432 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:39:07,432 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:39:33,625 ==================== TRACER ====================== +2021-12-29 07:39:33,626 Channel (server worker num[20]): +2021-12-29 07:39:33,627 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:39:33,628 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:39:37,463 ==================== TRACER ====================== +2021-12-29 07:39:37,464 Channel (server worker num[20]): +2021-12-29 07:39:37,465 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:39:37,465 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:40:03,658 ==================== TRACER ====================== +2021-12-29 07:40:03,659 Channel (server worker num[20]): +2021-12-29 07:40:03,660 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:40:03,661 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:40:07,475 ==================== TRACER ====================== +2021-12-29 07:40:07,475 Channel (server worker num[20]): +2021-12-29 07:40:07,476 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:40:07,477 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:40:33,680 ==================== TRACER ====================== +2021-12-29 07:40:33,681 Channel (server worker num[20]): +2021-12-29 07:40:33,681 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:40:33,682 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:40:37,507 ==================== TRACER ====================== +2021-12-29 07:40:37,508 Channel (server worker num[20]): +2021-12-29 07:40:37,509 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:40:37,510 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:41:03,713 ==================== TRACER ====================== +2021-12-29 07:41:03,714 Channel (server worker num[20]): +2021-12-29 07:41:03,714 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:41:03,715 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:41:07,516 ==================== TRACER ====================== +2021-12-29 07:41:07,517 Channel (server worker num[20]): +2021-12-29 07:41:07,518 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:41:07,518 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:41:33,744 ==================== TRACER ====================== +2021-12-29 07:41:33,745 Channel (server worker num[20]): +2021-12-29 07:41:33,746 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:41:33,746 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:41:37,549 ==================== TRACER ====================== +2021-12-29 07:41:37,550 Channel (server worker num[20]): +2021-12-29 07:41:37,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:41:37,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:42:11,746 ==================== TRACER ====================== +2021-12-29 07:42:11,748 Channel (server worker num[20]): +2021-12-29 07:42:11,754 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:42:11,755 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:42:41,785 ==================== TRACER ====================== +2021-12-29 07:42:41,787 DAGExecutor: +2021-12-29 07:42:41,787 Query count[1] +2021-12-29 07:42:41,787 QPS[0.03333333333333333 q/s] +2021-12-29 07:42:41,787 Succ[0.0] +2021-12-29 07:42:41,787 Error req[0] +2021-12-29 07:42:41,788 Latency: +2021-12-29 07:42:41,788 ave[1679.686 ms] +2021-12-29 07:42:41,788 .50[1679.686 ms] +2021-12-29 07:42:41,788 .60[1679.686 ms] +2021-12-29 07:42:41,788 .70[1679.686 ms] +2021-12-29 07:42:41,788 .80[1679.686 ms] +2021-12-29 07:42:41,789 .90[1679.686 ms] +2021-12-29 07:42:41,789 .95[1679.686 ms] +2021-12-29 07:42:41,789 .99[1679.686 ms] +2021-12-29 07:42:41,789 Channel (server worker num[20]): +2021-12-29 07:42:41,790 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:42:41,791 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:43:11,821 ==================== TRACER ====================== +2021-12-29 07:43:11,822 Channel (server worker num[20]): +2021-12-29 07:43:11,823 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:43:11,823 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:43:41,854 ==================== TRACER ====================== +2021-12-29 07:43:41,855 Channel (server worker num[20]): +2021-12-29 07:43:41,856 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:43:41,856 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:44:11,858 ==================== TRACER ====================== +2021-12-29 07:44:11,859 Channel (server worker num[20]): +2021-12-29 07:44:11,859 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:44:11,860 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:44:41,891 ==================== TRACER ====================== +2021-12-29 07:44:41,891 Channel (server worker num[20]): +2021-12-29 07:44:41,892 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:44:41,893 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:45:11,910 ==================== TRACER ====================== +2021-12-29 07:45:11,911 Channel (server worker num[20]): +2021-12-29 07:45:11,912 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:45:11,913 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:45:41,943 ==================== TRACER ====================== +2021-12-29 07:45:41,944 Channel (server worker num[20]): +2021-12-29 07:45:41,945 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:45:41,946 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:46:11,976 ==================== TRACER ====================== +2021-12-29 07:46:11,977 Channel (server worker num[20]): +2021-12-29 07:46:11,978 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:46:11,979 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:46:41,996 ==================== TRACER ====================== +2021-12-29 07:46:41,997 Channel (server worker num[20]): +2021-12-29 07:46:41,998 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:46:41,999 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:47:12,015 ==================== TRACER ====================== +2021-12-29 07:47:12,016 Channel (server worker num[20]): +2021-12-29 07:47:12,016 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:47:12,017 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:47:42,048 ==================== TRACER ====================== +2021-12-29 07:47:42,049 Channel (server worker num[20]): +2021-12-29 07:47:42,049 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:47:42,050 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:48:12,080 ==================== TRACER ====================== +2021-12-29 07:48:12,081 Channel (server worker num[20]): +2021-12-29 07:48:12,082 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:48:12,083 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:48:42,113 ==================== TRACER ====================== +2021-12-29 07:48:42,114 Channel (server worker num[20]): +2021-12-29 07:48:42,115 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:48:42,116 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:49:12,146 ==================== TRACER ====================== +2021-12-29 07:49:12,147 Channel (server worker num[20]): +2021-12-29 07:49:12,148 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:49:12,149 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:49:42,153 ==================== TRACER ====================== +2021-12-29 07:49:42,154 Channel (server worker num[20]): +2021-12-29 07:49:42,155 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:49:42,156 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:49:58,546 ==================== TRACER ====================== +2021-12-29 07:49:58,548 Channel (server worker num[20]): +2021-12-29 07:49:58,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:49:58,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:50:28,581 ==================== TRACER ====================== +2021-12-29 07:50:28,582 DAGExecutor: +2021-12-29 07:50:28,582 Query count[1] +2021-12-29 07:50:28,583 QPS[0.03333333333333333 q/s] +2021-12-29 07:50:28,583 Succ[0.0] +2021-12-29 07:50:28,583 Error req[0] +2021-12-29 07:50:28,583 Latency: +2021-12-29 07:50:28,583 ave[1878.876 ms] +2021-12-29 07:50:28,583 .50[1878.876 ms] +2021-12-29 07:50:28,584 .60[1878.876 ms] +2021-12-29 07:50:28,584 .70[1878.876 ms] +2021-12-29 07:50:28,584 .80[1878.876 ms] +2021-12-29 07:50:28,584 .90[1878.876 ms] +2021-12-29 07:50:28,584 .95[1878.876 ms] +2021-12-29 07:50:28,585 .99[1878.876 ms] +2021-12-29 07:50:28,585 Channel (server worker num[20]): +2021-12-29 07:50:28,586 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:50:28,586 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:50:58,592 ==================== TRACER ====================== +2021-12-29 07:50:58,593 Channel (server worker num[20]): +2021-12-29 07:50:58,594 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:50:58,595 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-29 07:51:28,625 ==================== TRACER ====================== +2021-12-29 07:51:28,626 Channel (server worker num[20]): +2021-12-29 07:51:28,627 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-29 07:51:28,627 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:53:16,350 ==================== TRACER ====================== +2021-12-30 06:53:16,352 Channel (server worker num[20]): +2021-12-30 06:53:16,353 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:53:16,354 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:53:46,384 ==================== TRACER ====================== +2021-12-30 06:53:46,385 DAGExecutor: +2021-12-30 06:53:46,386 Query count[1] +2021-12-30 06:53:46,386 QPS[0.03333333333333333 q/s] +2021-12-30 06:53:46,386 Succ[0.0] +2021-12-30 06:53:46,386 Error req[0] +2021-12-30 06:53:46,386 Latency: +2021-12-30 06:53:46,387 ave[1711.484 ms] +2021-12-30 06:53:46,387 .50[1711.484 ms] +2021-12-30 06:53:46,387 .60[1711.484 ms] +2021-12-30 06:53:46,387 .70[1711.484 ms] +2021-12-30 06:53:46,387 .80[1711.484 ms] +2021-12-30 06:53:46,387 .90[1711.484 ms] +2021-12-30 06:53:46,388 .95[1711.484 ms] +2021-12-30 06:53:46,388 .99[1711.484 ms] +2021-12-30 06:53:46,388 Channel (server worker num[20]): +2021-12-30 06:53:46,389 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:53:46,389 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:54:16,420 ==================== TRACER ====================== +2021-12-30 06:54:16,421 Channel (server worker num[20]): +2021-12-30 06:54:16,421 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:54:16,422 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:54:46,453 ==================== TRACER ====================== +2021-12-30 06:54:46,453 Channel (server worker num[20]): +2021-12-30 06:54:46,454 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:54:46,455 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:55:16,485 ==================== TRACER ====================== +2021-12-30 06:55:16,486 Channel (server worker num[20]): +2021-12-30 06:55:16,487 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:55:16,488 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:55:46,501 ==================== TRACER ====================== +2021-12-30 06:55:46,502 Channel (server worker num[20]): +2021-12-30 06:55:46,503 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:55:46,504 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:56:16,522 ==================== TRACER ====================== +2021-12-30 06:56:16,522 Channel (server worker num[20]): +2021-12-30 06:56:16,523 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:56:16,524 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:56:46,554 ==================== TRACER ====================== +2021-12-30 06:56:46,555 Channel (server worker num[20]): +2021-12-30 06:56:46,556 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:56:46,557 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:57:16,587 ==================== TRACER ====================== +2021-12-30 06:57:16,588 Channel (server worker num[20]): +2021-12-30 06:57:16,589 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:57:16,590 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:57:46,613 ==================== TRACER ====================== +2021-12-30 06:57:46,614 Channel (server worker num[20]): +2021-12-30 06:57:46,615 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:57:46,616 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:58:16,646 ==================== TRACER ====================== +2021-12-30 06:58:16,647 Channel (server worker num[20]): +2021-12-30 06:58:16,648 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:58:16,649 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:58:46,679 ==================== TRACER ====================== +2021-12-30 06:58:46,680 Channel (server worker num[20]): +2021-12-30 06:58:46,681 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:58:46,682 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:59:16,712 ==================== TRACER ====================== +2021-12-30 06:59:16,713 Channel (server worker num[20]): +2021-12-30 06:59:16,714 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:59:16,714 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 06:59:46,745 ==================== TRACER ====================== +2021-12-30 06:59:46,746 Channel (server worker num[20]): +2021-12-30 06:59:46,746 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 06:59:46,747 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:00:16,778 ==================== TRACER ====================== +2021-12-30 07:00:16,778 Channel (server worker num[20]): +2021-12-30 07:00:16,779 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:00:16,780 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:00:46,810 ==================== TRACER ====================== +2021-12-30 07:00:46,811 Channel (server worker num[20]): +2021-12-30 07:00:46,812 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:00:46,813 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:01:16,843 ==================== TRACER ====================== +2021-12-30 07:01:16,844 Channel (server worker num[20]): +2021-12-30 07:01:16,845 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:01:16,846 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:01:46,876 ==================== TRACER ====================== +2021-12-30 07:01:46,877 Channel (server worker num[20]): +2021-12-30 07:01:46,878 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:01:46,878 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:02:16,904 ==================== TRACER ====================== +2021-12-30 07:02:16,905 Channel (server worker num[20]): +2021-12-30 07:02:16,906 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:02:16,907 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:02:46,937 ==================== TRACER ====================== +2021-12-30 07:02:46,938 Channel (server worker num[20]): +2021-12-30 07:02:46,939 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:02:46,939 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:03:16,970 ==================== TRACER ====================== +2021-12-30 07:03:16,971 Channel (server worker num[20]): +2021-12-30 07:03:16,971 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:03:16,972 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:03:46,974 ==================== TRACER ====================== +2021-12-30 07:03:46,975 Channel (server worker num[20]): +2021-12-30 07:03:46,976 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:03:46,977 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:04:17,007 ==================== TRACER ====================== +2021-12-30 07:04:17,008 Channel (server worker num[20]): +2021-12-30 07:04:17,009 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:04:17,010 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:04:47,040 ==================== TRACER ====================== +2021-12-30 07:04:47,041 Channel (server worker num[20]): +2021-12-30 07:04:47,042 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:04:47,043 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:05:17,073 ==================== TRACER ====================== +2021-12-30 07:05:17,074 Channel (server worker num[20]): +2021-12-30 07:05:17,075 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:05:17,075 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:05:47,106 ==================== TRACER ====================== +2021-12-30 07:05:47,107 Channel (server worker num[20]): +2021-12-30 07:05:47,107 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:05:47,108 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:06:17,138 ==================== TRACER ====================== +2021-12-30 07:06:17,139 Channel (server worker num[20]): +2021-12-30 07:06:17,140 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:06:17,141 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:06:47,171 ==================== TRACER ====================== +2021-12-30 07:06:47,172 Channel (server worker num[20]): +2021-12-30 07:06:47,173 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:06:47,174 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:07:17,204 ==================== TRACER ====================== +2021-12-30 07:07:17,205 Channel (server worker num[20]): +2021-12-30 07:07:17,206 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:07:17,207 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:07:47,237 ==================== TRACER ====================== +2021-12-30 07:07:47,238 Channel (server worker num[20]): +2021-12-30 07:07:47,239 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:07:47,239 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:08:17,270 ==================== TRACER ====================== +2021-12-30 07:08:17,270 Channel (server worker num[20]): +2021-12-30 07:08:17,271 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:08:17,272 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:08:47,272 ==================== TRACER ====================== +2021-12-30 07:08:47,273 Channel (server worker num[20]): +2021-12-30 07:08:47,274 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:08:47,275 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:09:17,305 ==================== TRACER ====================== +2021-12-30 07:09:17,306 Channel (server worker num[20]): +2021-12-30 07:09:17,307 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:09:17,308 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:09:47,338 ==================== TRACER ====================== +2021-12-30 07:09:47,339 Channel (server worker num[20]): +2021-12-30 07:09:47,340 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:09:47,340 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:10:17,371 ==================== TRACER ====================== +2021-12-30 07:10:17,372 Channel (server worker num[20]): +2021-12-30 07:10:17,373 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:10:17,373 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:10:47,404 ==================== TRACER ====================== +2021-12-30 07:10:47,405 Channel (server worker num[20]): +2021-12-30 07:10:47,405 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:10:47,406 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:11:17,436 ==================== TRACER ====================== +2021-12-30 07:11:17,437 Channel (server worker num[20]): +2021-12-30 07:11:17,438 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:11:17,439 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:11:47,469 ==================== TRACER ====================== +2021-12-30 07:11:47,470 Channel (server worker num[20]): +2021-12-30 07:11:47,471 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:11:47,472 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:12:17,502 ==================== TRACER ====================== +2021-12-30 07:12:17,503 Channel (server worker num[20]): +2021-12-30 07:12:17,504 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:12:17,505 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:12:47,535 ==================== TRACER ====================== +2021-12-30 07:12:47,536 Channel (server worker num[20]): +2021-12-30 07:12:47,537 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:12:47,537 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:13:17,568 ==================== TRACER ====================== +2021-12-30 07:13:17,569 Channel (server worker num[20]): +2021-12-30 07:13:17,569 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:13:17,570 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:13:47,601 ==================== TRACER ====================== +2021-12-30 07:13:47,601 Channel (server worker num[20]): +2021-12-30 07:13:47,603 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:13:47,603 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:14:17,634 ==================== TRACER ====================== +2021-12-30 07:14:17,635 Channel (server worker num[20]): +2021-12-30 07:14:17,635 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:14:17,636 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:14:47,667 ==================== TRACER ====================== +2021-12-30 07:14:47,667 Channel (server worker num[20]): +2021-12-30 07:14:47,668 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:14:47,669 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:15:17,699 ==================== TRACER ====================== +2021-12-30 07:15:17,700 Channel (server worker num[20]): +2021-12-30 07:15:17,701 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:15:17,702 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:15:47,732 ==================== TRACER ====================== +2021-12-30 07:15:47,733 Channel (server worker num[20]): +2021-12-30 07:15:47,734 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:15:47,735 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:16:17,765 ==================== TRACER ====================== +2021-12-30 07:16:17,766 Channel (server worker num[20]): +2021-12-30 07:16:17,767 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:16:17,767 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:16:47,798 ==================== TRACER ====================== +2021-12-30 07:16:47,799 Channel (server worker num[20]): +2021-12-30 07:16:47,799 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:16:47,800 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:17:17,831 ==================== TRACER ====================== +2021-12-30 07:17:17,831 Channel (server worker num[20]): +2021-12-30 07:17:17,832 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:17:17,833 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:17:47,863 ==================== TRACER ====================== +2021-12-30 07:17:47,864 Channel (server worker num[20]): +2021-12-30 07:17:47,865 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:17:47,865 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:18:17,896 ==================== TRACER ====================== +2021-12-30 07:18:17,897 Channel (server worker num[20]): +2021-12-30 07:18:17,898 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:18:17,898 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:18:47,929 ==================== TRACER ====================== +2021-12-30 07:18:47,929 Channel (server worker num[20]): +2021-12-30 07:18:47,930 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:18:47,931 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:19:17,961 ==================== TRACER ====================== +2021-12-30 07:19:17,962 Channel (server worker num[20]): +2021-12-30 07:19:17,963 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:19:17,964 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:19:47,994 ==================== TRACER ====================== +2021-12-30 07:19:47,995 Channel (server worker num[20]): +2021-12-30 07:19:47,996 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:19:47,996 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:20:18,027 ==================== TRACER ====================== +2021-12-30 07:20:18,028 Channel (server worker num[20]): +2021-12-30 07:20:18,029 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:20:18,029 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:20:48,060 ==================== TRACER ====================== +2021-12-30 07:20:48,061 Channel (server worker num[20]): +2021-12-30 07:20:48,061 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:20:48,062 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:21:18,092 ==================== TRACER ====================== +2021-12-30 07:21:18,093 Channel (server worker num[20]): +2021-12-30 07:21:18,094 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:21:18,095 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:21:48,125 ==================== TRACER ====================== +2021-12-30 07:21:48,126 Channel (server worker num[20]): +2021-12-30 07:21:48,127 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:21:48,128 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:22:18,158 ==================== TRACER ====================== +2021-12-30 07:22:18,159 Channel (server worker num[20]): +2021-12-30 07:22:18,160 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:22:18,161 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:22:48,191 ==================== TRACER ====================== +2021-12-30 07:22:48,192 Channel (server worker num[20]): +2021-12-30 07:22:48,193 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:22:48,193 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:23:18,224 ==================== TRACER ====================== +2021-12-30 07:23:18,225 Channel (server worker num[20]): +2021-12-30 07:23:18,225 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:23:18,226 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:23:48,256 ==================== TRACER ====================== +2021-12-30 07:23:48,257 Channel (server worker num[20]): +2021-12-30 07:23:48,258 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:23:48,259 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:24:18,289 ==================== TRACER ====================== +2021-12-30 07:24:18,290 Channel (server worker num[20]): +2021-12-30 07:24:18,291 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:24:18,292 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:24:48,322 ==================== TRACER ====================== +2021-12-30 07:24:48,323 Channel (server worker num[20]): +2021-12-30 07:24:48,324 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:24:48,325 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:25:18,355 ==================== TRACER ====================== +2021-12-30 07:25:18,356 Channel (server worker num[20]): +2021-12-30 07:25:18,358 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:25:18,359 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:25:48,389 ==================== TRACER ====================== +2021-12-30 07:25:48,390 Channel (server worker num[20]): +2021-12-30 07:25:48,391 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:25:48,392 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:26:18,422 ==================== TRACER ====================== +2021-12-30 07:26:18,423 Channel (server worker num[20]): +2021-12-30 07:26:18,424 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:26:18,425 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:26:48,455 ==================== TRACER ====================== +2021-12-30 07:26:48,456 Channel (server worker num[20]): +2021-12-30 07:26:48,457 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:26:48,457 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:27:18,488 ==================== TRACER ====================== +2021-12-30 07:27:18,489 Channel (server worker num[20]): +2021-12-30 07:27:18,489 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:27:18,490 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:27:48,520 ==================== TRACER ====================== +2021-12-30 07:27:48,521 Channel (server worker num[20]): +2021-12-30 07:27:48,522 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:27:48,523 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:28:18,553 ==================== TRACER ====================== +2021-12-30 07:28:18,554 Channel (server worker num[20]): +2021-12-30 07:28:18,555 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:28:18,556 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:28:48,586 ==================== TRACER ====================== +2021-12-30 07:28:48,587 Channel (server worker num[20]): +2021-12-30 07:28:48,588 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:28:48,588 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:29:18,619 ==================== TRACER ====================== +2021-12-30 07:29:18,620 Channel (server worker num[20]): +2021-12-30 07:29:18,621 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:29:18,621 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:29:48,628 ==================== TRACER ====================== +2021-12-30 07:29:48,629 Channel (server worker num[20]): +2021-12-30 07:29:48,630 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:29:48,630 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:30:18,661 ==================== TRACER ====================== +2021-12-30 07:30:18,662 Channel (server worker num[20]): +2021-12-30 07:30:18,662 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:30:18,663 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:30:48,694 ==================== TRACER ====================== +2021-12-30 07:30:48,695 Channel (server worker num[20]): +2021-12-30 07:30:48,695 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:30:48,696 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:31:18,726 ==================== TRACER ====================== +2021-12-30 07:31:18,727 Channel (server worker num[20]): +2021-12-30 07:31:18,728 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:31:18,729 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:31:48,759 ==================== TRACER ====================== +2021-12-30 07:31:48,760 Channel (server worker num[20]): +2021-12-30 07:31:48,761 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:31:48,762 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:32:18,792 ==================== TRACER ====================== +2021-12-30 07:32:18,793 Channel (server worker num[20]): +2021-12-30 07:32:18,794 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:32:18,794 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:32:48,796 ==================== TRACER ====================== +2021-12-30 07:32:48,797 Channel (server worker num[20]): +2021-12-30 07:32:48,798 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:32:48,799 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:33:18,829 ==================== TRACER ====================== +2021-12-30 07:33:18,830 Channel (server worker num[20]): +2021-12-30 07:33:18,831 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:33:18,831 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:33:48,862 ==================== TRACER ====================== +2021-12-30 07:33:48,863 Channel (server worker num[20]): +2021-12-30 07:33:48,863 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:33:48,864 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:34:18,884 ==================== TRACER ====================== +2021-12-30 07:34:18,885 Channel (server worker num[20]): +2021-12-30 07:34:18,886 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:34:18,887 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:34:48,917 ==================== TRACER ====================== +2021-12-30 07:34:48,918 Channel (server worker num[20]): +2021-12-30 07:34:48,919 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:34:48,919 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:35:18,950 ==================== TRACER ====================== +2021-12-30 07:35:18,951 Channel (server worker num[20]): +2021-12-30 07:35:18,951 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:35:18,952 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:35:48,983 ==================== TRACER ====================== +2021-12-30 07:35:48,983 Channel (server worker num[20]): +2021-12-30 07:35:48,984 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:35:48,985 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:36:19,015 ==================== TRACER ====================== +2021-12-30 07:36:19,016 Channel (server worker num[20]): +2021-12-30 07:36:19,017 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:36:19,018 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:36:49,048 ==================== TRACER ====================== +2021-12-30 07:36:49,049 Channel (server worker num[20]): +2021-12-30 07:36:49,050 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:36:49,051 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:37:19,081 ==================== TRACER ====================== +2021-12-30 07:37:19,082 Channel (server worker num[20]): +2021-12-30 07:37:19,083 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:37:19,083 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:37:49,114 ==================== TRACER ====================== +2021-12-30 07:37:49,115 Channel (server worker num[20]): +2021-12-30 07:37:49,115 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:37:49,116 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:38:19,147 ==================== TRACER ====================== +2021-12-30 07:38:19,147 Channel (server worker num[20]): +2021-12-30 07:38:19,148 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:38:19,149 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:38:49,179 ==================== TRACER ====================== +2021-12-30 07:38:49,180 Channel (server worker num[20]): +2021-12-30 07:38:49,181 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:38:49,182 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:39:19,212 ==================== TRACER ====================== +2021-12-30 07:39:19,213 Channel (server worker num[20]): +2021-12-30 07:39:19,214 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:39:19,215 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:39:49,245 ==================== TRACER ====================== +2021-12-30 07:39:49,246 Channel (server worker num[20]): +2021-12-30 07:39:49,247 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:39:49,248 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:40:19,278 ==================== TRACER ====================== +2021-12-30 07:40:19,279 Channel (server worker num[20]): +2021-12-30 07:40:19,280 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:40:19,280 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:40:49,311 ==================== TRACER ====================== +2021-12-30 07:40:49,312 Channel (server worker num[20]): +2021-12-30 07:40:49,312 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:40:49,313 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:41:19,343 ==================== TRACER ====================== +2021-12-30 07:41:19,344 Channel (server worker num[20]): +2021-12-30 07:41:19,345 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:41:19,346 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:41:49,376 ==================== TRACER ====================== +2021-12-30 07:41:49,377 Channel (server worker num[20]): +2021-12-30 07:41:49,378 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:41:49,379 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:42:19,389 ==================== TRACER ====================== +2021-12-30 07:42:19,390 Channel (server worker num[20]): +2021-12-30 07:42:19,391 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:42:19,392 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:42:49,422 ==================== TRACER ====================== +2021-12-30 07:42:49,423 Channel (server worker num[20]): +2021-12-30 07:42:49,424 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:42:49,425 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:43:19,455 ==================== TRACER ====================== +2021-12-30 07:43:19,456 Channel (server worker num[20]): +2021-12-30 07:43:19,457 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:43:19,458 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:43:49,488 ==================== TRACER ====================== +2021-12-30 07:43:49,489 Channel (server worker num[20]): +2021-12-30 07:43:49,490 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:43:49,491 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:44:19,521 ==================== TRACER ====================== +2021-12-30 07:44:19,522 Channel (server worker num[20]): +2021-12-30 07:44:19,523 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:44:19,523 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:44:49,554 ==================== TRACER ====================== +2021-12-30 07:44:49,555 Channel (server worker num[20]): +2021-12-30 07:44:49,555 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:44:49,556 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:45:19,587 ==================== TRACER ====================== +2021-12-30 07:45:19,587 Channel (server worker num[20]): +2021-12-30 07:45:19,588 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:45:19,589 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:45:49,619 ==================== TRACER ====================== +2021-12-30 07:45:49,620 Channel (server worker num[20]): +2021-12-30 07:45:49,621 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:45:49,622 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:46:19,652 ==================== TRACER ====================== +2021-12-30 07:46:19,653 Channel (server worker num[20]): +2021-12-30 07:46:19,654 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:46:19,655 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:46:49,685 ==================== TRACER ====================== +2021-12-30 07:46:49,686 Channel (server worker num[20]): +2021-12-30 07:46:49,687 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:46:49,687 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:47:19,700 ==================== TRACER ====================== +2021-12-30 07:47:19,701 Channel (server worker num[20]): +2021-12-30 07:47:19,702 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:47:19,703 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:47:49,733 ==================== TRACER ====================== +2021-12-30 07:47:49,734 Channel (server worker num[20]): +2021-12-30 07:47:49,735 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:47:49,735 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:48:19,766 ==================== TRACER ====================== +2021-12-30 07:48:19,767 Channel (server worker num[20]): +2021-12-30 07:48:19,768 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:48:19,768 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:48:49,799 ==================== TRACER ====================== +2021-12-30 07:48:49,800 Channel (server worker num[20]): +2021-12-30 07:48:49,800 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:48:49,801 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:49:19,831 ==================== TRACER ====================== +2021-12-30 07:49:19,832 Channel (server worker num[20]): +2021-12-30 07:49:19,833 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:49:19,834 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:49:49,864 ==================== TRACER ====================== +2021-12-30 07:49:49,865 Channel (server worker num[20]): +2021-12-30 07:49:49,866 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:49:49,867 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:50:19,897 ==================== TRACER ====================== +2021-12-30 07:50:19,898 Channel (server worker num[20]): +2021-12-30 07:50:19,899 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:50:19,900 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:50:49,930 ==================== TRACER ====================== +2021-12-30 07:50:49,931 Channel (server worker num[20]): +2021-12-30 07:50:49,932 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:50:49,932 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:51:19,963 ==================== TRACER ====================== +2021-12-30 07:51:19,964 Channel (server worker num[20]): +2021-12-30 07:51:19,964 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:51:19,965 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:51:49,996 ==================== TRACER ====================== +2021-12-30 07:51:49,996 Channel (server worker num[20]): +2021-12-30 07:51:49,997 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:51:49,998 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:52:20,028 ==================== TRACER ====================== +2021-12-30 07:52:20,029 Channel (server worker num[20]): +2021-12-30 07:52:20,030 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:52:20,031 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:52:50,061 ==================== TRACER ====================== +2021-12-30 07:52:50,062 Channel (server worker num[20]): +2021-12-30 07:52:50,063 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:52:50,064 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:53:20,094 ==================== TRACER ====================== +2021-12-30 07:53:20,095 Channel (server worker num[20]): +2021-12-30 07:53:20,096 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:53:20,096 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:53:50,127 ==================== TRACER ====================== +2021-12-30 07:53:50,128 Channel (server worker num[20]): +2021-12-30 07:53:50,129 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:53:50,129 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:54:20,160 ==================== TRACER ====================== +2021-12-30 07:54:20,160 Channel (server worker num[20]): +2021-12-30 07:54:20,161 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:54:20,162 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:54:50,192 ==================== TRACER ====================== +2021-12-30 07:54:50,193 Channel (server worker num[20]): +2021-12-30 07:54:50,194 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:54:50,195 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:55:20,225 ==================== TRACER ====================== +2021-12-30 07:55:20,226 Channel (server worker num[20]): +2021-12-30 07:55:20,227 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:55:20,227 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:55:50,258 ==================== TRACER ====================== +2021-12-30 07:55:50,259 Channel (server worker num[20]): +2021-12-30 07:55:50,259 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:55:50,260 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:56:20,291 ==================== TRACER ====================== +2021-12-30 07:56:20,291 Channel (server worker num[20]): +2021-12-30 07:56:20,292 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:56:20,293 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:56:50,323 ==================== TRACER ====================== +2021-12-30 07:56:50,324 Channel (server worker num[20]): +2021-12-30 07:56:50,325 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:56:50,326 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:57:06,917 ==================== TRACER ====================== +2021-12-30 07:57:06,918 Channel (server worker num[20]): +2021-12-30 07:57:06,920 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:57:06,921 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:57:36,951 ==================== TRACER ====================== +2021-12-30 07:57:36,953 DAGExecutor: +2021-12-30 07:57:36,953 Query count[1] +2021-12-30 07:57:36,953 QPS[0.03333333333333333 q/s] +2021-12-30 07:57:36,953 Succ[0.0] +2021-12-30 07:57:36,953 Error req[0] +2021-12-30 07:57:36,954 Latency: +2021-12-30 07:57:36,954 ave[1767.715 ms] +2021-12-30 07:57:36,954 .50[1767.715 ms] +2021-12-30 07:57:36,954 .60[1767.715 ms] +2021-12-30 07:57:36,954 .70[1767.715 ms] +2021-12-30 07:57:36,954 .80[1767.715 ms] +2021-12-30 07:57:36,955 .90[1767.715 ms] +2021-12-30 07:57:36,955 .95[1767.715 ms] +2021-12-30 07:57:36,955 .99[1767.715 ms] +2021-12-30 07:57:36,955 Channel (server worker num[20]): +2021-12-30 07:57:36,956 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:57:36,957 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:58:06,987 ==================== TRACER ====================== +2021-12-30 07:58:06,988 Channel (server worker num[20]): +2021-12-30 07:58:06,989 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:58:06,989 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:58:37,020 ==================== TRACER ====================== +2021-12-30 07:58:37,021 Channel (server worker num[20]): +2021-12-30 07:58:37,021 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:58:37,022 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:59:07,053 ==================== TRACER ====================== +2021-12-30 07:59:07,053 Channel (server worker num[20]): +2021-12-30 07:59:07,054 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:59:07,055 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 07:59:37,085 ==================== TRACER ====================== +2021-12-30 07:59:37,086 Channel (server worker num[20]): +2021-12-30 07:59:37,087 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 07:59:37,088 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:00:07,118 ==================== TRACER ====================== +2021-12-30 08:00:07,119 Channel (server worker num[20]): +2021-12-30 08:00:07,120 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:00:07,120 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:00:37,151 ==================== TRACER ====================== +2021-12-30 08:00:37,152 Channel (server worker num[20]): +2021-12-30 08:00:37,152 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:00:37,153 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:01:07,184 ==================== TRACER ====================== +2021-12-30 08:01:07,184 Channel (server worker num[20]): +2021-12-30 08:01:07,185 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:01:07,186 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:01:37,216 ==================== TRACER ====================== +2021-12-30 08:01:37,217 Channel (server worker num[20]): +2021-12-30 08:01:37,218 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:01:37,219 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:02:07,249 ==================== TRACER ====================== +2021-12-30 08:02:07,250 Channel (server worker num[20]): +2021-12-30 08:02:07,251 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:02:07,251 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:02:37,282 ==================== TRACER ====================== +2021-12-30 08:02:37,282 Channel (server worker num[20]): +2021-12-30 08:02:37,283 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:02:37,284 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:03:07,314 ==================== TRACER ====================== +2021-12-30 08:03:07,315 Channel (server worker num[20]): +2021-12-30 08:03:07,316 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:03:07,317 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:03:37,347 ==================== TRACER ====================== +2021-12-30 08:03:37,348 Channel (server worker num[20]): +2021-12-30 08:03:37,349 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:03:37,349 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:04:07,380 ==================== TRACER ====================== +2021-12-30 08:04:07,381 Channel (server worker num[20]): +2021-12-30 08:04:07,382 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:04:07,382 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:04:37,390 ==================== TRACER ====================== +2021-12-30 08:04:37,390 Channel (server worker num[20]): +2021-12-30 08:04:37,391 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:04:37,392 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:05:07,422 ==================== TRACER ====================== +2021-12-30 08:05:07,423 Channel (server worker num[20]): +2021-12-30 08:05:07,424 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:05:07,425 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:05:37,455 ==================== TRACER ====================== +2021-12-30 08:05:37,456 Channel (server worker num[20]): +2021-12-30 08:05:37,457 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:05:37,458 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:06:07,488 ==================== TRACER ====================== +2021-12-30 08:06:07,489 Channel (server worker num[20]): +2021-12-30 08:06:07,490 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:06:07,490 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:06:37,516 ==================== TRACER ====================== +2021-12-30 08:06:37,517 Channel (server worker num[20]): +2021-12-30 08:06:37,518 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:06:37,518 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:07:07,549 ==================== TRACER ====================== +2021-12-30 08:07:07,550 Channel (server worker num[20]): +2021-12-30 08:07:07,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:07:07,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:07:37,581 ==================== TRACER ====================== +2021-12-30 08:07:37,582 Channel (server worker num[20]): +2021-12-30 08:07:37,583 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:07:37,584 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:08:07,614 ==================== TRACER ====================== +2021-12-30 08:08:07,615 Channel (server worker num[20]): +2021-12-30 08:08:07,616 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:08:07,617 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:08:37,647 ==================== TRACER ====================== +2021-12-30 08:08:37,648 Channel (server worker num[20]): +2021-12-30 08:08:37,649 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:08:37,650 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:09:07,680 ==================== TRACER ====================== +2021-12-30 08:09:07,681 Channel (server worker num[20]): +2021-12-30 08:09:07,682 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:09:07,682 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:09:37,709 ==================== TRACER ====================== +2021-12-30 08:09:37,710 Channel (server worker num[20]): +2021-12-30 08:09:37,710 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:09:37,711 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:10:07,741 ==================== TRACER ====================== +2021-12-30 08:10:07,742 Channel (server worker num[20]): +2021-12-30 08:10:07,743 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:10:07,744 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:10:37,774 ==================== TRACER ====================== +2021-12-30 08:10:37,775 Channel (server worker num[20]): +2021-12-30 08:10:37,776 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:10:37,777 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:11:07,807 ==================== TRACER ====================== +2021-12-30 08:11:07,808 Channel (server worker num[20]): +2021-12-30 08:11:07,809 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:11:07,810 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:11:37,840 ==================== TRACER ====================== +2021-12-30 08:11:37,841 Channel (server worker num[20]): +2021-12-30 08:11:37,842 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:11:37,842 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:12:07,244 ==================== TRACER ====================== +2021-12-30 08:12:07,246 Channel (server worker num[20]): +2021-12-30 08:12:07,247 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:12:07,248 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:12:37,278 ==================== TRACER ====================== +2021-12-30 08:12:37,280 DAGExecutor: +2021-12-30 08:12:37,280 Query count[1] +2021-12-30 08:12:37,280 QPS[0.03333333333333333 q/s] +2021-12-30 08:12:37,280 Succ[0.0] +2021-12-30 08:12:37,281 Error req[0] +2021-12-30 08:12:37,281 Latency: +2021-12-30 08:12:37,281 ave[1666.015 ms] +2021-12-30 08:12:37,281 .50[1666.015 ms] +2021-12-30 08:12:37,281 .60[1666.015 ms] +2021-12-30 08:12:37,281 .70[1666.015 ms] +2021-12-30 08:12:37,282 .80[1666.015 ms] +2021-12-30 08:12:37,282 .90[1666.015 ms] +2021-12-30 08:12:37,282 .95[1666.015 ms] +2021-12-30 08:12:37,282 .99[1666.015 ms] +2021-12-30 08:12:37,282 Channel (server worker num[20]): +2021-12-30 08:12:37,283 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:12:37,284 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:13:07,314 ==================== TRACER ====================== +2021-12-30 08:13:07,315 Channel (server worker num[20]): +2021-12-30 08:13:07,316 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:13:07,317 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:13:37,347 ==================== TRACER ====================== +2021-12-30 08:13:37,348 Channel (server worker num[20]): +2021-12-30 08:13:37,349 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:13:37,349 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:13:45,787 ==================== TRACER ====================== +2021-12-30 08:13:45,789 Channel (server worker num[20]): +2021-12-30 08:13:45,792 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:13:45,793 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:14:15,823 ==================== TRACER ====================== +2021-12-30 08:14:15,824 DAGExecutor: +2021-12-30 08:14:15,825 Query count[1] +2021-12-30 08:14:15,825 QPS[0.03333333333333333 q/s] +2021-12-30 08:14:15,825 Succ[0.0] +2021-12-30 08:14:15,825 Error req[0] +2021-12-30 08:14:15,825 Latency: +2021-12-30 08:14:15,826 ave[1659.131 ms] +2021-12-30 08:14:15,826 .50[1659.131 ms] +2021-12-30 08:14:15,826 .60[1659.131 ms] +2021-12-30 08:14:15,826 .70[1659.131 ms] +2021-12-30 08:14:15,826 .80[1659.131 ms] +2021-12-30 08:14:15,826 .90[1659.131 ms] +2021-12-30 08:14:15,827 .95[1659.131 ms] +2021-12-30 08:14:15,827 .99[1659.131 ms] +2021-12-30 08:14:15,827 Channel (server worker num[20]): +2021-12-30 08:14:15,828 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:14:15,828 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:14:45,859 ==================== TRACER ====================== +2021-12-30 08:14:45,860 Channel (server worker num[20]): +2021-12-30 08:14:45,860 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:14:45,861 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:15:15,892 ==================== TRACER ====================== +2021-12-30 08:15:15,892 Channel (server worker num[20]): +2021-12-30 08:15:15,893 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:15:15,894 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:15:45,924 ==================== TRACER ====================== +2021-12-30 08:15:45,925 Channel (server worker num[20]): +2021-12-30 08:15:45,926 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:15:45,927 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:16:15,957 ==================== TRACER ====================== +2021-12-30 08:16:15,958 Channel (server worker num[20]): +2021-12-30 08:16:15,959 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:16:15,959 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:16:45,990 ==================== TRACER ====================== +2021-12-30 08:16:45,991 Channel (server worker num[20]): +2021-12-30 08:16:45,992 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:16:45,992 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:17:15,596 ==================== TRACER ====================== +2021-12-30 08:17:15,598 Channel (server worker num[20]): +2021-12-30 08:17:15,600 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:17:15,601 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:17:31,013 ==================== TRACER ====================== +2021-12-30 08:17:31,015 Channel (server worker num[20]): +2021-12-30 08:17:31,017 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:17:31,018 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:18:01,049 ==================== TRACER ====================== +2021-12-30 08:18:01,050 DAGExecutor: +2021-12-30 08:18:01,050 Query count[1] +2021-12-30 08:18:01,050 QPS[0.03333333333333333 q/s] +2021-12-30 08:18:01,050 Succ[0.0] +2021-12-30 08:18:01,051 Error req[0] +2021-12-30 08:18:01,051 Latency: +2021-12-30 08:18:01,051 ave[1718.021 ms] +2021-12-30 08:18:01,051 .50[1718.021 ms] +2021-12-30 08:18:01,051 .60[1718.021 ms] +2021-12-30 08:18:01,051 .70[1718.021 ms] +2021-12-30 08:18:01,052 .80[1718.021 ms] +2021-12-30 08:18:01,052 .90[1718.021 ms] +2021-12-30 08:18:01,052 .95[1718.021 ms] +2021-12-30 08:18:01,052 .99[1718.021 ms] +2021-12-30 08:18:01,052 Channel (server worker num[20]): +2021-12-30 08:18:01,053 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:18:01,054 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:18:31,084 ==================== TRACER ====================== +2021-12-30 08:18:31,085 Channel (server worker num[20]): +2021-12-30 08:18:31,086 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:18:31,087 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:19:01,117 ==================== TRACER ====================== +2021-12-30 08:19:01,118 Channel (server worker num[20]): +2021-12-30 08:19:01,119 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:19:01,120 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:19:31,150 ==================== TRACER ====================== +2021-12-30 08:19:31,151 Channel (server worker num[20]): +2021-12-30 08:19:31,152 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:19:31,152 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:20:01,183 ==================== TRACER ====================== +2021-12-30 08:20:01,184 Channel (server worker num[20]): +2021-12-30 08:20:01,184 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:20:01,185 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:20:13,303 ==================== TRACER ====================== +2021-12-30 08:20:13,304 Channel (server worker num[20]): +2021-12-30 08:20:13,306 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:20:13,307 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:20:43,337 ==================== TRACER ====================== +2021-12-30 08:20:43,339 DAGExecutor: +2021-12-30 08:20:43,339 Query count[1] +2021-12-30 08:20:43,339 QPS[0.03333333333333333 q/s] +2021-12-30 08:20:43,339 Succ[0.0] +2021-12-30 08:20:43,339 Error req[0] +2021-12-30 08:20:43,340 Latency: +2021-12-30 08:20:43,340 ave[1664.843 ms] +2021-12-30 08:20:43,340 .50[1664.843 ms] +2021-12-30 08:20:43,340 .60[1664.843 ms] +2021-12-30 08:20:43,340 .70[1664.843 ms] +2021-12-30 08:20:43,341 .80[1664.843 ms] +2021-12-30 08:20:43,341 .90[1664.843 ms] +2021-12-30 08:20:43,341 .95[1664.843 ms] +2021-12-30 08:20:43,341 .99[1664.843 ms] +2021-12-30 08:20:43,341 Channel (server worker num[20]): +2021-12-30 08:20:43,342 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:20:43,343 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:21:13,373 ==================== TRACER ====================== +2021-12-30 08:21:13,374 Channel (server worker num[20]): +2021-12-30 08:21:13,375 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:21:13,376 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:31:54,883 ==================== TRACER ====================== +2021-12-30 08:31:54,885 Channel (server worker num[20]): +2021-12-30 08:31:54,887 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:31:54,888 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:32:24,900 ==================== TRACER ====================== +2021-12-30 08:32:24,901 DAGExecutor: +2021-12-30 08:32:24,902 Query count[1] +2021-12-30 08:32:24,902 QPS[0.03333333333333333 q/s] +2021-12-30 08:32:24,902 Succ[0.0] +2021-12-30 08:32:24,902 Error req[0] +2021-12-30 08:32:24,902 Latency: +2021-12-30 08:32:24,903 ave[1698.932 ms] +2021-12-30 08:32:24,903 .50[1698.932 ms] +2021-12-30 08:32:24,903 .60[1698.932 ms] +2021-12-30 08:32:24,903 .70[1698.932 ms] +2021-12-30 08:32:24,903 .80[1698.932 ms] +2021-12-30 08:32:24,903 .90[1698.932 ms] +2021-12-30 08:32:24,904 .95[1698.932 ms] +2021-12-30 08:32:24,904 .99[1698.932 ms] +2021-12-30 08:32:24,904 Channel (server worker num[20]): +2021-12-30 08:32:24,905 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:32:24,906 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:32:54,936 ==================== TRACER ====================== +2021-12-30 08:32:54,937 Channel (server worker num[20]): +2021-12-30 08:32:54,938 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:32:54,938 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:33:24,969 ==================== TRACER ====================== +2021-12-30 08:33:24,970 Channel (server worker num[20]): +2021-12-30 08:33:24,970 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:33:24,971 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:33:55,002 ==================== TRACER ====================== +2021-12-30 08:33:55,002 Channel (server worker num[20]): +2021-12-30 08:33:55,003 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:33:55,004 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:34:21,528 ==================== TRACER ====================== +2021-12-30 08:34:21,530 Channel (server worker num[20]): +2021-12-30 08:34:21,533 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:34:21,534 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:34:51,564 ==================== TRACER ====================== +2021-12-30 08:34:51,565 DAGExecutor: +2021-12-30 08:34:51,566 Query count[1] +2021-12-30 08:34:51,566 QPS[0.03333333333333333 q/s] +2021-12-30 08:34:51,566 Succ[0.0] +2021-12-30 08:34:51,566 Error req[0] +2021-12-30 08:34:51,566 Latency: +2021-12-30 08:34:51,566 ave[1726.27 ms] +2021-12-30 08:34:51,567 .50[1726.27 ms] +2021-12-30 08:34:51,567 .60[1726.27 ms] +2021-12-30 08:34:51,567 .70[1726.27 ms] +2021-12-30 08:34:51,567 .80[1726.27 ms] +2021-12-30 08:34:51,567 .90[1726.27 ms] +2021-12-30 08:34:51,567 .95[1726.27 ms] +2021-12-30 08:34:51,568 .99[1726.27 ms] +2021-12-30 08:34:51,568 Channel (server worker num[20]): +2021-12-30 08:34:51,569 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:34:51,569 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:35:21,600 ==================== TRACER ====================== +2021-12-30 08:35:21,601 Channel (server worker num[20]): +2021-12-30 08:35:21,602 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:35:21,602 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:35:51,633 ==================== TRACER ====================== +2021-12-30 08:35:51,634 Channel (server worker num[20]): +2021-12-30 08:35:51,635 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:35:51,635 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:36:21,666 ==================== TRACER ====================== +2021-12-30 08:36:21,667 Channel (server worker num[20]): +2021-12-30 08:36:21,668 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:36:21,668 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:36:51,699 ==================== TRACER ====================== +2021-12-30 08:36:51,700 Channel (server worker num[20]): +2021-12-30 08:36:51,701 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:36:51,701 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:37:21,732 ==================== TRACER ====================== +2021-12-30 08:37:21,733 Channel (server worker num[20]): +2021-12-30 08:37:21,734 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:37:21,734 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:37:51,765 ==================== TRACER ====================== +2021-12-30 08:37:51,766 Channel (server worker num[20]): +2021-12-30 08:37:51,767 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:37:51,767 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:43:32,381 ==================== TRACER ====================== +2021-12-30 08:43:32,383 Channel (server worker num[20]): +2021-12-30 08:43:32,386 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:43:32,387 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:44:02,417 ==================== TRACER ====================== +2021-12-30 08:44:02,418 DAGExecutor: +2021-12-30 08:44:02,419 Query count[1] +2021-12-30 08:44:02,419 QPS[0.03333333333333333 q/s] +2021-12-30 08:44:02,419 Succ[0.0] +2021-12-30 08:44:02,419 Error req[0] +2021-12-30 08:44:02,419 Latency: +2021-12-30 08:44:02,419 ave[1687.363 ms] +2021-12-30 08:44:02,420 .50[1687.363 ms] +2021-12-30 08:44:02,420 .60[1687.363 ms] +2021-12-30 08:44:02,420 .70[1687.363 ms] +2021-12-30 08:44:02,420 .80[1687.363 ms] +2021-12-30 08:44:02,420 .90[1687.363 ms] +2021-12-30 08:44:02,420 .95[1687.363 ms] +2021-12-30 08:44:02,421 .99[1687.363 ms] +2021-12-30 08:44:02,421 Channel (server worker num[20]): +2021-12-30 08:44:02,422 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:44:02,422 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:44:32,453 ==================== TRACER ====================== +2021-12-30 08:44:32,453 Channel (server worker num[20]): +2021-12-30 08:44:32,454 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:44:32,455 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:45:02,485 ==================== TRACER ====================== +2021-12-30 08:45:02,486 Channel (server worker num[20]): +2021-12-30 08:45:02,487 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:45:02,488 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:45:32,518 ==================== TRACER ====================== +2021-12-30 08:45:32,519 Channel (server worker num[20]): +2021-12-30 08:45:32,520 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:45:32,521 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:46:02,551 ==================== TRACER ====================== +2021-12-30 08:46:02,552 Channel (server worker num[20]): +2021-12-30 08:46:02,553 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:46:02,553 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:46:32,584 ==================== TRACER ====================== +2021-12-30 08:46:32,585 Channel (server worker num[20]): +2021-12-30 08:46:32,585 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:46:32,586 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:47:02,616 ==================== TRACER ====================== +2021-12-30 08:47:02,617 Channel (server worker num[20]): +2021-12-30 08:47:02,618 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:47:02,619 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:47:32,649 ==================== TRACER ====================== +2021-12-30 08:47:32,650 Channel (server worker num[20]): +2021-12-30 08:47:32,651 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:47:32,652 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:48:02,682 ==================== TRACER ====================== +2021-12-30 08:48:02,683 Channel (server worker num[20]): +2021-12-30 08:48:02,684 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:48:02,684 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:48:32,715 ==================== TRACER ====================== +2021-12-30 08:48:32,716 Channel (server worker num[20]): +2021-12-30 08:48:32,717 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:48:32,717 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:49:02,748 ==================== TRACER ====================== +2021-12-30 08:49:02,749 Channel (server worker num[20]): +2021-12-30 08:49:02,749 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:49:02,750 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:49:32,396 ==================== TRACER ====================== +2021-12-30 08:49:32,398 Channel (server worker num[20]): +2021-12-30 08:49:32,400 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:49:32,401 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:50:02,431 ==================== TRACER ====================== +2021-12-30 08:50:02,432 DAGExecutor: +2021-12-30 08:50:02,433 Query count[1] +2021-12-30 08:50:02,433 QPS[0.03333333333333333 q/s] +2021-12-30 08:50:02,433 Succ[0.0] +2021-12-30 08:50:02,433 Error req[0] +2021-12-30 08:50:02,433 Latency: +2021-12-30 08:50:02,433 ave[1679.848 ms] +2021-12-30 08:50:02,434 .50[1679.848 ms] +2021-12-30 08:50:02,434 .60[1679.848 ms] +2021-12-30 08:50:02,434 .70[1679.848 ms] +2021-12-30 08:50:02,434 .80[1679.848 ms] +2021-12-30 08:50:02,434 .90[1679.848 ms] +2021-12-30 08:50:02,434 .95[1679.848 ms] +2021-12-30 08:50:02,435 .99[1679.848 ms] +2021-12-30 08:50:02,435 Channel (server worker num[20]): +2021-12-30 08:50:02,436 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:50:02,436 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:50:32,467 ==================== TRACER ====================== +2021-12-30 08:50:32,468 Channel (server worker num[20]): +2021-12-30 08:50:32,468 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:50:32,469 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:51:02,497 ==================== TRACER ====================== +2021-12-30 08:51:02,498 Channel (server worker num[20]): +2021-12-30 08:51:02,499 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:51:02,500 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:51:32,530 ==================== TRACER ====================== +2021-12-30 08:51:32,531 Channel (server worker num[20]): +2021-12-30 08:51:32,532 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:51:32,532 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:52:02,563 ==================== TRACER ====================== +2021-12-30 08:52:02,564 Channel (server worker num[20]): +2021-12-30 08:52:02,565 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:52:02,565 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:52:32,596 ==================== TRACER ====================== +2021-12-30 08:52:32,596 Channel (server worker num[20]): +2021-12-30 08:52:32,597 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:52:32,598 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:52:42,481 ==================== TRACER ====================== +2021-12-30 08:52:42,484 Channel (server worker num[20]): +2021-12-30 08:52:42,486 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:52:42,486 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:53:12,517 ==================== TRACER ====================== +2021-12-30 08:53:12,518 DAGExecutor: +2021-12-30 08:53:12,518 Query count[1] +2021-12-30 08:53:12,518 QPS[0.03333333333333333 q/s] +2021-12-30 08:53:12,519 Succ[0.0] +2021-12-30 08:53:12,519 Error req[0] +2021-12-30 08:53:12,519 Latency: +2021-12-30 08:53:12,519 ave[1709.757 ms] +2021-12-30 08:53:12,519 .50[1709.757 ms] +2021-12-30 08:53:12,519 .60[1709.757 ms] +2021-12-30 08:53:12,520 .70[1709.757 ms] +2021-12-30 08:53:12,520 .80[1709.757 ms] +2021-12-30 08:53:12,520 .90[1709.757 ms] +2021-12-30 08:53:12,520 .95[1709.757 ms] +2021-12-30 08:53:12,520 .99[1709.757 ms] +2021-12-30 08:53:12,520 Channel (server worker num[20]): +2021-12-30 08:53:12,521 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:53:12,522 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:53:42,552 ==================== TRACER ====================== +2021-12-30 08:53:42,553 Channel (server worker num[20]): +2021-12-30 08:53:42,554 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:53:42,555 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:54:12,585 ==================== TRACER ====================== +2021-12-30 08:54:12,586 Channel (server worker num[20]): +2021-12-30 08:54:12,587 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:54:12,587 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:54:42,618 ==================== TRACER ====================== +2021-12-30 08:54:42,619 Channel (server worker num[20]): +2021-12-30 08:54:42,620 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:54:42,620 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 08:59:43,531 ==================== TRACER ====================== +2021-12-30 08:59:43,533 Channel (server worker num[20]): +2021-12-30 08:59:43,535 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 08:59:43,536 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:00:13,566 ==================== TRACER ====================== +2021-12-30 09:00:13,567 Channel (server worker num[20]): +2021-12-30 09:00:13,568 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:00:13,569 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:00:43,599 ==================== TRACER ====================== +2021-12-30 09:00:43,600 Channel (server worker num[20]): +2021-12-30 09:00:43,601 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:00:43,602 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:01:13,632 ==================== TRACER ====================== +2021-12-30 09:01:13,633 Channel (server worker num[20]): +2021-12-30 09:01:13,634 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:01:13,634 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:01:43,665 ==================== TRACER ====================== +2021-12-30 09:01:43,666 Channel (server worker num[20]): +2021-12-30 09:01:43,666 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:01:43,667 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:02:13,697 ==================== TRACER ====================== +2021-12-30 09:02:13,698 Channel (server worker num[20]): +2021-12-30 09:02:13,699 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:02:13,700 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:02:43,708 ==================== TRACER ====================== +2021-12-30 09:02:43,709 Channel (server worker num[20]): +2021-12-30 09:02:43,710 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:02:43,711 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:03:10,066 ==================== TRACER ====================== +2021-12-30 09:03:10,068 Channel (server worker num[20]): +2021-12-30 09:03:10,071 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:03:10,071 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:03:12,019 ==================== TRACER ====================== +2021-12-30 09:03:12,021 Channel (server worker num[20]): +2021-12-30 09:03:12,023 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:03:12,024 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:03:42,055 ==================== TRACER ====================== +2021-12-30 09:03:42,056 DAGExecutor: +2021-12-30 09:03:42,056 Query count[1] +2021-12-30 09:03:42,056 QPS[0.03333333333333333 q/s] +2021-12-30 09:03:42,057 Succ[0.0] +2021-12-30 09:03:42,057 Error req[0] +2021-12-30 09:03:42,057 Latency: +2021-12-30 09:03:42,057 ave[1915.476 ms] +2021-12-30 09:03:42,057 .50[1915.476 ms] +2021-12-30 09:03:42,057 .60[1915.476 ms] +2021-12-30 09:03:42,058 .70[1915.476 ms] +2021-12-30 09:03:42,058 .80[1915.476 ms] +2021-12-30 09:03:42,058 .90[1915.476 ms] +2021-12-30 09:03:42,058 .95[1915.476 ms] +2021-12-30 09:03:42,058 .99[1915.476 ms] +2021-12-30 09:03:42,058 Channel (server worker num[20]): +2021-12-30 09:03:42,059 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:03:42,060 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:04:12,090 ==================== TRACER ====================== +2021-12-30 09:04:12,091 Channel (server worker num[20]): +2021-12-30 09:04:12,092 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:04:12,093 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:04:42,120 ==================== TRACER ====================== +2021-12-30 09:04:42,121 Channel (server worker num[20]): +2021-12-30 09:04:42,122 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:04:42,123 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:05:12,153 ==================== TRACER ====================== +2021-12-30 09:05:12,154 Channel (server worker num[20]): +2021-12-30 09:05:12,155 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:05:12,155 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:05:42,186 ==================== TRACER ====================== +2021-12-30 09:05:42,187 Channel (server worker num[20]): +2021-12-30 09:05:42,188 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:05:42,188 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:06:12,219 ==================== TRACER ====================== +2021-12-30 09:06:12,220 Channel (server worker num[20]): +2021-12-30 09:06:12,220 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:06:12,221 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:06:42,251 ==================== TRACER ====================== +2021-12-30 09:06:42,252 Channel (server worker num[20]): +2021-12-30 09:06:42,253 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:06:42,254 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:07:12,284 ==================== TRACER ====================== +2021-12-30 09:07:12,285 Channel (server worker num[20]): +2021-12-30 09:07:12,286 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:07:12,287 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:07:27,496 ==================== TRACER ====================== +2021-12-30 09:07:27,498 Channel (server worker num[20]): +2021-12-30 09:07:27,500 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:07:27,501 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:07:48,868 ==================== TRACER ====================== +2021-12-30 09:07:48,871 Channel (server worker num[20]): +2021-12-30 09:07:48,872 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:07:48,873 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:08:06,615 ==================== TRACER ====================== +2021-12-30 09:08:06,618 Channel (server worker num[20]): +2021-12-30 09:08:06,620 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:08:06,620 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:08:36,651 ==================== TRACER ====================== +2021-12-30 09:08:36,652 DAGExecutor: +2021-12-30 09:08:36,652 Query count[1] +2021-12-30 09:08:36,652 QPS[0.03333333333333333 q/s] +2021-12-30 09:08:36,653 Succ[0.0] +2021-12-30 09:08:36,653 Error req[0] +2021-12-30 09:08:36,653 Latency: +2021-12-30 09:08:36,653 ave[1661.09 ms] +2021-12-30 09:08:36,653 .50[1661.09 ms] +2021-12-30 09:08:36,654 .60[1661.09 ms] +2021-12-30 09:08:36,654 .70[1661.09 ms] +2021-12-30 09:08:36,654 .80[1661.09 ms] +2021-12-30 09:08:36,654 .90[1661.09 ms] +2021-12-30 09:08:36,654 .95[1661.09 ms] +2021-12-30 09:08:36,654 .99[1661.09 ms] +2021-12-30 09:08:36,655 Channel (server worker num[20]): +2021-12-30 09:08:36,655 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:08:36,656 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:09:06,686 ==================== TRACER ====================== +2021-12-30 09:09:06,687 Channel (server worker num[20]): +2021-12-30 09:09:06,688 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:09:06,689 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:09:36,719 ==================== TRACER ====================== +2021-12-30 09:09:36,720 Channel (server worker num[20]): +2021-12-30 09:09:36,721 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:09:36,722 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:10:06,752 ==================== TRACER ====================== +2021-12-30 09:10:06,753 Channel (server worker num[20]): +2021-12-30 09:10:06,754 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:10:06,755 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:10:36,785 ==================== TRACER ====================== +2021-12-30 09:10:36,786 Channel (server worker num[20]): +2021-12-30 09:10:36,787 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:10:36,787 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:11:06,818 ==================== TRACER ====================== +2021-12-30 09:11:06,819 Channel (server worker num[20]): +2021-12-30 09:11:06,819 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:11:06,820 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:11:36,851 ==================== TRACER ====================== +2021-12-30 09:11:36,851 Channel (server worker num[20]): +2021-12-30 09:11:36,852 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:11:36,853 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:12:06,883 ==================== TRACER ====================== +2021-12-30 09:12:06,884 Channel (server worker num[20]): +2021-12-30 09:12:06,885 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:12:06,886 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:12:36,900 ==================== TRACER ====================== +2021-12-30 09:12:36,901 Channel (server worker num[20]): +2021-12-30 09:12:36,902 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:12:36,903 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:13:06,933 ==================== TRACER ====================== +2021-12-30 09:13:06,934 Channel (server worker num[20]): +2021-12-30 09:13:06,935 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:13:06,935 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:13:36,966 ==================== TRACER ====================== +2021-12-30 09:13:36,967 Channel (server worker num[20]): +2021-12-30 09:13:36,967 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:13:36,968 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:14:06,998 ==================== TRACER ====================== +2021-12-30 09:14:06,999 Channel (server worker num[20]): +2021-12-30 09:14:07,000 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:14:07,001 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:14:37,031 ==================== TRACER ====================== +2021-12-30 09:14:37,032 Channel (server worker num[20]): +2021-12-30 09:14:37,033 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:14:37,034 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:15:07,064 ==================== TRACER ====================== +2021-12-30 09:15:07,065 Channel (server worker num[20]): +2021-12-30 09:15:07,068 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:15:07,068 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:15:37,096 ==================== TRACER ====================== +2021-12-30 09:15:37,097 Channel (server worker num[20]): +2021-12-30 09:15:37,098 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:15:37,099 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:16:07,129 ==================== TRACER ====================== +2021-12-30 09:16:07,130 Channel (server worker num[20]): +2021-12-30 09:16:07,131 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:16:07,131 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:16:37,162 ==================== TRACER ====================== +2021-12-30 09:16:37,163 Channel (server worker num[20]): +2021-12-30 09:16:37,163 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:16:37,164 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:17:07,195 ==================== TRACER ====================== +2021-12-30 09:17:07,195 Channel (server worker num[20]): +2021-12-30 09:17:07,196 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:17:07,197 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:17:37,227 ==================== TRACER ====================== +2021-12-30 09:17:37,228 Channel (server worker num[20]): +2021-12-30 09:17:37,229 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:17:37,230 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:18:07,260 ==================== TRACER ====================== +2021-12-30 09:18:07,261 Channel (server worker num[20]): +2021-12-30 09:18:07,262 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:18:07,263 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:18:37,293 ==================== TRACER ====================== +2021-12-30 09:18:37,294 Channel (server worker num[20]): +2021-12-30 09:18:37,295 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:18:37,296 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:19:07,326 ==================== TRACER ====================== +2021-12-30 09:19:07,327 Channel (server worker num[20]): +2021-12-30 09:19:07,328 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:19:07,328 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:19:39,524 ==================== TRACER ====================== +2021-12-30 09:19:39,525 Channel (server worker num[20]): +2021-12-30 09:19:39,527 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:19:39,528 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:20:11,284 ==================== TRACER ====================== +2021-12-30 09:20:11,286 Channel (server worker num[20]): +2021-12-30 09:20:11,289 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:20:11,289 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:20:41,320 ==================== TRACER ====================== +2021-12-30 09:20:41,321 DAGExecutor: +2021-12-30 09:20:41,321 Query count[1] +2021-12-30 09:20:41,321 QPS[0.03333333333333333 q/s] +2021-12-30 09:20:41,322 Succ[0.0] +2021-12-30 09:20:41,322 Error req[0] +2021-12-30 09:20:41,322 Latency: +2021-12-30 09:20:41,322 ave[2327.767 ms] +2021-12-30 09:20:41,322 .50[2327.767 ms] +2021-12-30 09:20:41,323 .60[2327.767 ms] +2021-12-30 09:20:41,323 .70[2327.767 ms] +2021-12-30 09:20:41,323 .80[2327.767 ms] +2021-12-30 09:20:41,323 .90[2327.767 ms] +2021-12-30 09:20:41,323 .95[2327.767 ms] +2021-12-30 09:20:41,323 .99[2327.767 ms] +2021-12-30 09:20:41,324 Channel (server worker num[20]): +2021-12-30 09:20:41,324 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:20:41,325 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:21:11,356 ==================== TRACER ====================== +2021-12-30 09:21:11,356 Channel (server worker num[20]): +2021-12-30 09:21:11,357 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:21:11,358 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:21:19,661 ==================== TRACER ====================== +2021-12-30 09:21:19,663 Channel (server worker num[20]): +2021-12-30 09:21:19,665 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:21:19,666 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:21:49,668 ==================== TRACER ====================== +2021-12-30 09:21:49,669 DAGExecutor: +2021-12-30 09:21:49,669 Query count[1] +2021-12-30 09:21:49,670 QPS[0.03333333333333333 q/s] +2021-12-30 09:21:49,670 Succ[0.0] +2021-12-30 09:21:49,670 Error req[0] +2021-12-30 09:21:49,670 Latency: +2021-12-30 09:21:49,670 ave[1667.456 ms] +2021-12-30 09:21:49,670 .50[1667.456 ms] +2021-12-30 09:21:49,671 .60[1667.456 ms] +2021-12-30 09:21:49,671 .70[1667.456 ms] +2021-12-30 09:21:49,671 .80[1667.456 ms] +2021-12-30 09:21:49,671 .90[1667.456 ms] +2021-12-30 09:21:49,671 .95[1667.456 ms] +2021-12-30 09:21:49,671 .99[1667.456 ms] +2021-12-30 09:21:49,672 Channel (server worker num[20]): +2021-12-30 09:21:49,672 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:21:49,673 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:22:19,704 ==================== TRACER ====================== +2021-12-30 09:22:19,704 Channel (server worker num[20]): +2021-12-30 09:22:19,705 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:22:19,706 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:22:49,736 ==================== TRACER ====================== +2021-12-30 09:22:49,737 Channel (server worker num[20]): +2021-12-30 09:22:49,738 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:22:49,739 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:23:19,769 ==================== TRACER ====================== +2021-12-30 09:23:19,770 Channel (server worker num[20]): +2021-12-30 09:23:19,771 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:23:19,772 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:23:31,946 ==================== TRACER ====================== +2021-12-30 09:23:31,948 Channel (server worker num[20]): +2021-12-30 09:23:31,951 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:23:31,951 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:24:02,960 ==================== TRACER ====================== +2021-12-30 09:24:02,962 Channel (server worker num[20]): +2021-12-30 09:24:02,964 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:24:02,965 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:24:32,995 ==================== TRACER ====================== +2021-12-30 09:24:32,996 DAGExecutor: +2021-12-30 09:24:32,997 Query count[1] +2021-12-30 09:24:32,997 QPS[0.03333333333333333 q/s] +2021-12-30 09:24:32,997 Succ[0.0] +2021-12-30 09:24:32,997 Error req[0] +2021-12-30 09:24:32,997 Latency: +2021-12-30 09:24:32,997 ave[2969.908 ms] +2021-12-30 09:24:32,998 .50[2969.908 ms] +2021-12-30 09:24:32,998 .60[2969.908 ms] +2021-12-30 09:24:32,998 .70[2969.908 ms] +2021-12-30 09:24:32,998 .80[2969.908 ms] +2021-12-30 09:24:32,998 .90[2969.908 ms] +2021-12-30 09:24:32,998 .95[2969.908 ms] +2021-12-30 09:24:32,999 .99[2969.908 ms] +2021-12-30 09:24:32,999 Channel (server worker num[20]): +2021-12-30 09:24:33,000 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:24:33,000 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:25:03,031 ==================== TRACER ====================== +2021-12-30 09:25:03,032 Channel (server worker num[20]): +2021-12-30 09:25:03,032 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:25:03,033 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:25:33,049 ==================== TRACER ====================== +2021-12-30 09:25:33,050 Channel (server worker num[20]): +2021-12-30 09:25:33,051 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:25:33,052 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:25:54,144 ==================== TRACER ====================== +2021-12-30 09:25:54,146 Channel (server worker num[20]): +2021-12-30 09:25:54,149 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:25:54,149 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:26:24,168 ==================== TRACER ====================== +2021-12-30 09:26:24,170 Op(ppyolo_mbv3): +2021-12-30 09:26:24,170 in[11.296 ms] +2021-12-30 09:26:24,170 prep[49.388 ms] +2021-12-30 09:26:24,170 midp[1611.812 ms] +2021-12-30 09:26:24,170 postp[11.047 ms] +2021-12-30 09:26:24,171 out[2.017 ms] +2021-12-30 09:26:24,171 idle[0.007898265264956517] +2021-12-30 09:26:24,171 DAGExecutor: +2021-12-30 09:26:24,171 Query count[1] +2021-12-30 09:26:24,171 QPS[0.03333333333333333 q/s] +2021-12-30 09:26:24,171 Succ[1.0] +2021-12-30 09:26:24,172 Error req[] +2021-12-30 09:26:24,172 Latency: +2021-12-30 09:26:24,172 ave[1682.576 ms] +2021-12-30 09:26:24,172 .50[1682.576 ms] +2021-12-30 09:26:24,172 .60[1682.576 ms] +2021-12-30 09:26:24,173 .70[1682.576 ms] +2021-12-30 09:26:24,173 .80[1682.576 ms] +2021-12-30 09:26:24,173 .90[1682.576 ms] +2021-12-30 09:26:24,173 .95[1682.576 ms] +2021-12-30 09:26:24,173 .99[1682.576 ms] +2021-12-30 09:26:24,173 Channel (server worker num[20]): +2021-12-30 09:26:24,174 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:26:24,175 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:26:54,205 ==================== TRACER ====================== +2021-12-30 09:26:54,206 Channel (server worker num[20]): +2021-12-30 09:26:54,207 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:26:54,208 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:39:07,251 ==================== TRACER ====================== +2021-12-30 09:39:07,254 Channel (server worker num[20]): +2021-12-30 09:39:07,256 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:39:07,256 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:39:37,287 ==================== TRACER ====================== +2021-12-30 09:39:37,288 Op(ppyolo_mbv3): +2021-12-30 09:39:37,289 in[16052.636 ms] +2021-12-30 09:39:37,289 prep[52.574 ms] +2021-12-30 09:39:37,289 midp[1722.923 ms] +2021-12-30 09:39:37,289 postp[11.854 ms] +2021-12-30 09:39:37,289 out[1.654 ms] +2021-12-30 09:39:37,289 idle[0.8998213785379944] +2021-12-30 09:39:37,290 DAGExecutor: +2021-12-30 09:39:37,290 Query count[1] +2021-12-30 09:39:37,290 QPS[0.03333333333333333 q/s] +2021-12-30 09:39:37,290 Succ[1.0] +2021-12-30 09:39:37,290 Error req[] +2021-12-30 09:39:37,291 Latency: +2021-12-30 09:39:37,291 ave[1797.892 ms] +2021-12-30 09:39:37,291 .50[1797.892 ms] +2021-12-30 09:39:37,291 .60[1797.892 ms] +2021-12-30 09:39:37,291 .70[1797.892 ms] +2021-12-30 09:39:37,291 .80[1797.892 ms] +2021-12-30 09:39:37,292 .90[1797.892 ms] +2021-12-30 09:39:37,292 .95[1797.892 ms] +2021-12-30 09:39:37,292 .99[1797.892 ms] +2021-12-30 09:39:37,292 Channel (server worker num[20]): +2021-12-30 09:39:37,293 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:39:37,294 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:40:07,324 ==================== TRACER ====================== +2021-12-30 09:40:07,325 DAGExecutor: +2021-12-30 09:40:07,325 Query count[1] +2021-12-30 09:40:07,325 QPS[0.03333333333333333 q/s] +2021-12-30 09:40:07,326 Succ[0.0] +2021-12-30 09:40:07,326 Error req[1] +2021-12-30 09:40:07,326 Latency: +2021-12-30 09:40:07,326 ave[118.163 ms] +2021-12-30 09:40:07,326 .50[118.163 ms] +2021-12-30 09:40:07,326 .60[118.163 ms] +2021-12-30 09:40:07,327 .70[118.163 ms] +2021-12-30 09:40:07,327 .80[118.163 ms] +2021-12-30 09:40:07,327 .90[118.163 ms] +2021-12-30 09:40:07,327 .95[118.163 ms] +2021-12-30 09:40:07,327 .99[118.163 ms] +2021-12-30 09:40:07,327 Channel (server worker num[20]): +2021-12-30 09:40:07,328 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:40:07,329 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:40:37,359 ==================== TRACER ====================== +2021-12-30 09:40:37,360 Channel (server worker num[20]): +2021-12-30 09:40:37,361 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:40:37,362 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:41:07,392 ==================== TRACER ====================== +2021-12-30 09:41:07,393 DAGExecutor: +2021-12-30 09:41:07,394 Query count[2] +2021-12-30 09:41:07,394 QPS[0.06666666666666667 q/s] +2021-12-30 09:41:07,394 Succ[0.0] +2021-12-30 09:41:07,394 Error req[2, 3] +2021-12-30 09:41:07,394 Latency: +2021-12-30 09:41:07,395 ave[91.7105 ms] +2021-12-30 09:41:07,395 .50[110.376 ms] +2021-12-30 09:41:07,395 .60[110.376 ms] +2021-12-30 09:41:07,395 .70[110.376 ms] +2021-12-30 09:41:07,395 .80[110.376 ms] +2021-12-30 09:41:07,395 .90[110.376 ms] +2021-12-30 09:41:07,396 .95[110.376 ms] +2021-12-30 09:41:07,396 .99[110.376 ms] +2021-12-30 09:41:07,396 Channel (server worker num[20]): +2021-12-30 09:41:07,397 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:41:07,397 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:41:37,428 ==================== TRACER ====================== +2021-12-30 09:41:37,429 Channel (server worker num[20]): +2021-12-30 09:41:37,430 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:41:37,430 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:42:01,898 ==================== TRACER ====================== +2021-12-30 09:42:01,900 Channel (server worker num[20]): +2021-12-30 09:42:01,902 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:42:01,902 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:42:24,079 ==================== TRACER ====================== +2021-12-30 09:42:24,081 Channel (server worker num[20]): +2021-12-30 09:42:24,083 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:42:24,084 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:42:54,114 ==================== TRACER ====================== +2021-12-30 09:42:54,116 Op(ppyolo_mbv3): +2021-12-30 09:42:54,116 in[238.366 ms] +2021-12-30 09:42:54,116 prep[78.16 ms] +2021-12-30 09:42:54,116 midp[1605.08 ms] +2021-12-30 09:42:54,117 postp[10.166 ms] +2021-12-30 09:42:54,117 out[1.317 ms] +2021-12-30 09:42:54,117 idle[0.12398963524183315] +2021-12-30 09:42:54,117 DAGExecutor: +2021-12-30 09:42:54,117 Query count[1] +2021-12-30 09:42:54,117 QPS[0.03333333333333333 q/s] +2021-12-30 09:42:54,118 Succ[1.0] +2021-12-30 09:42:54,118 Error req[] +2021-12-30 09:42:54,118 Latency: +2021-12-30 09:42:54,118 ave[1705.072 ms] +2021-12-30 09:42:54,118 .50[1705.072 ms] +2021-12-30 09:42:54,119 .60[1705.072 ms] +2021-12-30 09:42:54,119 .70[1705.072 ms] +2021-12-30 09:42:54,119 .80[1705.072 ms] +2021-12-30 09:42:54,119 .90[1705.072 ms] +2021-12-30 09:42:54,119 .95[1705.072 ms] +2021-12-30 09:42:54,119 .99[1705.072 ms] +2021-12-30 09:42:54,119 Channel (server worker num[20]): +2021-12-30 09:42:54,120 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:42:54,121 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:43:24,151 ==================== TRACER ====================== +2021-12-30 09:43:24,153 DAGExecutor: +2021-12-30 09:43:24,153 Query count[2] +2021-12-30 09:43:24,153 QPS[0.06666666666666667 q/s] +2021-12-30 09:43:24,153 Succ[0.0] +2021-12-30 09:43:24,153 Error req[1, 2] +2021-12-30 09:43:24,154 Latency: +2021-12-30 09:43:24,154 ave[111.03649999999999 ms] +2021-12-30 09:43:24,154 .50[111.139 ms] +2021-12-30 09:43:24,154 .60[111.139 ms] +2021-12-30 09:43:24,154 .70[111.139 ms] +2021-12-30 09:43:24,154 .80[111.139 ms] +2021-12-30 09:43:24,155 .90[111.139 ms] +2021-12-30 09:43:24,155 .95[111.139 ms] +2021-12-30 09:43:24,155 .99[111.139 ms] +2021-12-30 09:43:24,155 Channel (server worker num[20]): +2021-12-30 09:43:24,156 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:43:24,156 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:43:54,187 ==================== TRACER ====================== +2021-12-30 09:43:54,188 Channel (server worker num[20]): +2021-12-30 09:43:54,188 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:43:54,189 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:44:24,220 ==================== TRACER ====================== +2021-12-30 09:44:24,220 Channel (server worker num[20]): +2021-12-30 09:44:24,221 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:44:24,222 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:44:54,933 ==================== TRACER ====================== +2021-12-30 09:44:54,935 Channel (server worker num[20]): +2021-12-30 09:44:54,937 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:44:54,938 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:45:24,968 ==================== TRACER ====================== +2021-12-30 09:45:24,970 Op(ppyolo_mbv3): +2021-12-30 09:45:24,970 in[1609.535 ms] +2021-12-30 09:45:24,970 prep[67.209 ms] +2021-12-30 09:45:24,970 midp[1766.049 ms] +2021-12-30 09:45:24,970 postp[11.493 ms] +2021-12-30 09:45:24,971 out[1.567 ms] +2021-12-30 09:45:24,971 idle[0.4661951767045647] +2021-12-30 09:45:24,971 DAGExecutor: +2021-12-30 09:45:24,971 Query count[2] +2021-12-30 09:45:24,971 QPS[0.06666666666666667 q/s] +2021-12-30 09:45:24,972 Succ[0.5] +2021-12-30 09:45:24,972 Error req[1] +2021-12-30 09:45:24,972 Latency: +2021-12-30 09:45:24,972 ave[984.508 ms] +2021-12-30 09:45:24,972 .50[1859.535 ms] +2021-12-30 09:45:24,972 .60[1859.535 ms] +2021-12-30 09:45:24,973 .70[1859.535 ms] +2021-12-30 09:45:24,973 .80[1859.535 ms] +2021-12-30 09:45:24,973 .90[1859.535 ms] +2021-12-30 09:45:24,973 .95[1859.535 ms] +2021-12-30 09:45:24,973 .99[1859.535 ms] +2021-12-30 09:45:24,973 Channel (server worker num[20]): +2021-12-30 09:45:24,974 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:45:24,975 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:45:55,005 ==================== TRACER ====================== +2021-12-30 09:45:55,006 Channel (server worker num[20]): +2021-12-30 09:45:55,007 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:45:55,007 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:46:17,802 ==================== TRACER ====================== +2021-12-30 09:46:17,804 Channel (server worker num[20]): +2021-12-30 09:46:17,807 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:46:17,808 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:46:47,838 ==================== TRACER ====================== +2021-12-30 09:46:47,840 Op(ppyolo_mbv3): +2021-12-30 09:46:47,840 in[842.57 ms] +2021-12-30 09:46:47,840 prep[79.406 ms] +2021-12-30 09:46:47,841 midp[1591.784 ms] +2021-12-30 09:46:47,841 postp[10.156 ms] +2021-12-30 09:46:47,841 out[1.137 ms] +2021-12-30 09:46:47,841 idle[0.33413437262504997] +2021-12-30 09:46:47,841 DAGExecutor: +2021-12-30 09:46:47,841 Query count[2] +2021-12-30 09:46:47,842 QPS[0.06666666666666667 q/s] +2021-12-30 09:46:47,842 Succ[0.5] +2021-12-30 09:46:47,842 Error req[1] +2021-12-30 09:46:47,842 Latency: +2021-12-30 09:46:47,842 ave[910.8975 ms] +2021-12-30 09:46:47,842 .50[1695.613 ms] +2021-12-30 09:46:47,843 .60[1695.613 ms] +2021-12-30 09:46:47,843 .70[1695.613 ms] +2021-12-30 09:46:47,843 .80[1695.613 ms] +2021-12-30 09:46:47,843 .90[1695.613 ms] +2021-12-30 09:46:47,843 .95[1695.613 ms] +2021-12-30 09:46:47,843 .99[1695.613 ms] +2021-12-30 09:46:47,844 Channel (server worker num[20]): +2021-12-30 09:46:47,844 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:46:47,845 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:47:17,876 ==================== TRACER ====================== +2021-12-30 09:47:17,877 Channel (server worker num[20]): +2021-12-30 09:47:17,877 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:47:17,878 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:47:47,909 ==================== TRACER ====================== +2021-12-30 09:47:47,910 Channel (server worker num[20]): +2021-12-30 09:47:47,910 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:47:47,911 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:48:17,942 ==================== TRACER ====================== +2021-12-30 09:48:17,943 Channel (server worker num[20]): +2021-12-30 09:48:17,944 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:48:17,944 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:48:47,975 ==================== TRACER ====================== +2021-12-30 09:48:47,976 Channel (server worker num[20]): +2021-12-30 09:48:47,977 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:48:47,977 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:49:18,008 ==================== TRACER ====================== +2021-12-30 09:49:18,009 Channel (server worker num[20]): +2021-12-30 09:49:18,011 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:49:18,012 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:49:48,042 ==================== TRACER ====================== +2021-12-30 09:49:48,043 Channel (server worker num[20]): +2021-12-30 09:49:48,044 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:49:48,045 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:50:18,057 ==================== TRACER ====================== +2021-12-30 09:50:18,058 Channel (server worker num[20]): +2021-12-30 09:50:18,059 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:50:18,060 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:50:48,091 ==================== TRACER ====================== +2021-12-30 09:50:48,092 Channel (server worker num[20]): +2021-12-30 09:50:48,093 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:50:48,093 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:51:18,124 ==================== TRACER ====================== +2021-12-30 09:51:18,125 Channel (server worker num[20]): +2021-12-30 09:51:18,126 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:51:18,127 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:51:48,157 ==================== TRACER ====================== +2021-12-30 09:51:48,158 Channel (server worker num[20]): +2021-12-30 09:51:48,159 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:51:48,159 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:52:18,190 ==================== TRACER ====================== +2021-12-30 09:52:18,191 Channel (server worker num[20]): +2021-12-30 09:52:18,192 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:52:18,193 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:52:48,220 ==================== TRACER ====================== +2021-12-30 09:52:48,221 Channel (server worker num[20]): +2021-12-30 09:52:48,222 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:52:48,223 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:53:18,253 ==================== TRACER ====================== +2021-12-30 09:53:18,254 Channel (server worker num[20]): +2021-12-30 09:53:18,255 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:53:18,256 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:53:48,286 ==================== TRACER ====================== +2021-12-30 09:53:48,287 Channel (server worker num[20]): +2021-12-30 09:53:48,288 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:53:48,289 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:54:18,318 ==================== TRACER ====================== +2021-12-30 09:54:18,319 Channel (server worker num[20]): +2021-12-30 09:54:18,319 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:54:18,320 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:54:48,351 ==================== TRACER ====================== +2021-12-30 09:54:48,352 Channel (server worker num[20]): +2021-12-30 09:54:48,352 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:54:48,353 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:55:18,384 ==================== TRACER ====================== +2021-12-30 09:55:18,385 Channel (server worker num[20]): +2021-12-30 09:55:18,385 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:55:18,386 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:55:48,417 ==================== TRACER ====================== +2021-12-30 09:55:48,418 Channel (server worker num[20]): +2021-12-30 09:55:48,418 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:55:48,419 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:56:18,450 ==================== TRACER ====================== +2021-12-30 09:56:18,451 Channel (server worker num[20]): +2021-12-30 09:56:18,451 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:56:18,452 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:56:48,483 ==================== TRACER ====================== +2021-12-30 09:56:48,484 Channel (server worker num[20]): +2021-12-30 09:56:48,484 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:56:48,485 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:57:18,516 ==================== TRACER ====================== +2021-12-30 09:57:18,517 Channel (server worker num[20]): +2021-12-30 09:57:18,518 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:57:18,518 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:57:48,549 ==================== TRACER ====================== +2021-12-30 09:57:48,550 Channel (server worker num[20]): +2021-12-30 09:57:48,550 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:57:48,551 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:58:18,582 ==================== TRACER ====================== +2021-12-30 09:58:18,583 Channel (server worker num[20]): +2021-12-30 09:58:18,584 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:58:18,584 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:58:48,615 ==================== TRACER ====================== +2021-12-30 09:58:48,616 Channel (server worker num[20]): +2021-12-30 09:58:48,617 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:58:48,617 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:59:18,648 ==================== TRACER ====================== +2021-12-30 09:59:18,649 Channel (server worker num[20]): +2021-12-30 09:59:18,650 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:59:18,650 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 09:59:48,680 ==================== TRACER ====================== +2021-12-30 09:59:48,681 Channel (server worker num[20]): +2021-12-30 09:59:48,682 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 09:59:48,683 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:00:18,713 ==================== TRACER ====================== +2021-12-30 10:00:18,714 Channel (server worker num[20]): +2021-12-30 10:00:18,715 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:00:18,716 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:00:48,731 ==================== TRACER ====================== +2021-12-30 10:00:48,732 Channel (server worker num[20]): +2021-12-30 10:00:48,733 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:00:48,734 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:01:18,748 ==================== TRACER ====================== +2021-12-30 10:01:18,749 Channel (server worker num[20]): +2021-12-30 10:01:18,750 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:01:18,751 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:01:48,781 ==================== TRACER ====================== +2021-12-30 10:01:48,782 Channel (server worker num[20]): +2021-12-30 10:01:48,783 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:01:48,784 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:02:18,814 ==================== TRACER ====================== +2021-12-30 10:02:18,815 Channel (server worker num[20]): +2021-12-30 10:02:18,816 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:02:18,817 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:02:48,847 ==================== TRACER ====================== +2021-12-30 10:02:48,848 Channel (server worker num[20]): +2021-12-30 10:02:48,849 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:02:48,850 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:03:18,880 ==================== TRACER ====================== +2021-12-30 10:03:18,881 Channel (server worker num[20]): +2021-12-30 10:03:18,882 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:03:18,883 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:03:48,892 ==================== TRACER ====================== +2021-12-30 10:03:48,893 Channel (server worker num[20]): +2021-12-30 10:03:48,894 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:03:48,895 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:04:18,925 ==================== TRACER ====================== +2021-12-30 10:04:18,926 Channel (server worker num[20]): +2021-12-30 10:04:18,927 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:04:18,928 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:04:48,958 ==================== TRACER ====================== +2021-12-30 10:04:48,959 Channel (server worker num[20]): +2021-12-30 10:04:48,960 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:04:48,961 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:05:18,991 ==================== TRACER ====================== +2021-12-30 10:05:18,992 Channel (server worker num[20]): +2021-12-30 10:05:18,993 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:05:18,994 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:05:49,000 ==================== TRACER ====================== +2021-12-30 10:05:49,001 Channel (server worker num[20]): +2021-12-30 10:05:49,002 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:05:49,003 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:06:19,033 ==================== TRACER ====================== +2021-12-30 10:06:19,034 Channel (server worker num[20]): +2021-12-30 10:06:19,035 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:06:19,036 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:06:49,066 ==================== TRACER ====================== +2021-12-30 10:06:49,067 Channel (server worker num[20]): +2021-12-30 10:06:49,068 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:06:49,069 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:07:19,099 ==================== TRACER ====================== +2021-12-30 10:07:19,100 Channel (server worker num[20]): +2021-12-30 10:07:19,101 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:07:19,102 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:07:49,132 ==================== TRACER ====================== +2021-12-30 10:07:49,133 Channel (server worker num[20]): +2021-12-30 10:07:49,134 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:07:49,135 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:08:19,165 ==================== TRACER ====================== +2021-12-30 10:08:19,166 Channel (server worker num[20]): +2021-12-30 10:08:19,167 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:08:19,168 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:08:49,198 ==================== TRACER ====================== +2021-12-30 10:08:49,199 Channel (server worker num[20]): +2021-12-30 10:08:49,200 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:08:49,201 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:09:19,231 ==================== TRACER ====================== +2021-12-30 10:09:19,232 Channel (server worker num[20]): +2021-12-30 10:09:19,233 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:09:19,234 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:09:49,265 ==================== TRACER ====================== +2021-12-30 10:09:49,266 Channel (server worker num[20]): +2021-12-30 10:09:49,266 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:09:49,267 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:10:19,294 ==================== TRACER ====================== +2021-12-30 10:10:19,295 Channel (server worker num[20]): +2021-12-30 10:10:19,296 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:10:19,296 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:10:49,327 ==================== TRACER ====================== +2021-12-30 10:10:49,328 Channel (server worker num[20]): +2021-12-30 10:10:49,329 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:10:49,329 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:11:19,360 ==================== TRACER ====================== +2021-12-30 10:11:19,361 Channel (server worker num[20]): +2021-12-30 10:11:19,362 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:11:19,363 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:11:49,368 ==================== TRACER ====================== +2021-12-30 10:11:49,369 Channel (server worker num[20]): +2021-12-30 10:11:49,370 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:11:49,371 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:12:19,401 ==================== TRACER ====================== +2021-12-30 10:12:19,402 Channel (server worker num[20]): +2021-12-30 10:12:19,403 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:12:19,404 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:12:49,434 ==================== TRACER ====================== +2021-12-30 10:12:49,435 Channel (server worker num[20]): +2021-12-30 10:12:49,436 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:12:49,437 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:13:19,468 ==================== TRACER ====================== +2021-12-30 10:13:19,469 Channel (server worker num[20]): +2021-12-30 10:13:19,469 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:13:19,470 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:13:49,500 ==================== TRACER ====================== +2021-12-30 10:13:49,501 Channel (server worker num[20]): +2021-12-30 10:13:49,502 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:13:49,503 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:14:19,533 ==================== TRACER ====================== +2021-12-30 10:14:19,534 Channel (server worker num[20]): +2021-12-30 10:14:19,535 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:14:19,536 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:14:49,566 ==================== TRACER ====================== +2021-12-30 10:14:49,567 Channel (server worker num[20]): +2021-12-30 10:14:49,570 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:14:49,570 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:15:19,592 ==================== TRACER ====================== +2021-12-30 10:15:19,593 Channel (server worker num[20]): +2021-12-30 10:15:19,594 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:15:19,595 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:15:49,625 ==================== TRACER ====================== +2021-12-30 10:15:49,626 Channel (server worker num[20]): +2021-12-30 10:15:49,627 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:15:49,628 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:16:19,658 ==================== TRACER ====================== +2021-12-30 10:16:19,659 Channel (server worker num[20]): +2021-12-30 10:16:19,660 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:16:19,661 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:16:49,692 ==================== TRACER ====================== +2021-12-30 10:16:49,693 Channel (server worker num[20]): +2021-12-30 10:16:49,693 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:16:49,694 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:17:19,725 ==================== TRACER ====================== +2021-12-30 10:17:19,726 Channel (server worker num[20]): +2021-12-30 10:17:19,727 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:17:19,727 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:17:49,732 ==================== TRACER ====================== +2021-12-30 10:17:49,733 Channel (server worker num[20]): +2021-12-30 10:17:49,734 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:17:49,735 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:18:19,765 ==================== TRACER ====================== +2021-12-30 10:18:19,766 Channel (server worker num[20]): +2021-12-30 10:18:19,767 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:18:19,768 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:18:49,798 ==================== TRACER ====================== +2021-12-30 10:18:49,799 Channel (server worker num[20]): +2021-12-30 10:18:49,800 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:18:49,801 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:19:19,811 ==================== TRACER ====================== +2021-12-30 10:19:19,812 Channel (server worker num[20]): +2021-12-30 10:19:19,813 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:19:19,814 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:19:49,844 ==================== TRACER ====================== +2021-12-30 10:19:49,845 Channel (server worker num[20]): +2021-12-30 10:19:49,846 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:19:49,847 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:20:19,877 ==================== TRACER ====================== +2021-12-30 10:20:19,878 Channel (server worker num[20]): +2021-12-30 10:20:19,879 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:20:19,880 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:20:49,910 ==================== TRACER ====================== +2021-12-30 10:20:49,911 Channel (server worker num[20]): +2021-12-30 10:20:49,912 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:20:49,913 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:21:19,943 ==================== TRACER ====================== +2021-12-30 10:21:19,944 Channel (server worker num[20]): +2021-12-30 10:21:19,945 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:21:19,946 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:21:49,976 ==================== TRACER ====================== +2021-12-30 10:21:49,977 Channel (server worker num[20]): +2021-12-30 10:21:49,978 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:21:49,979 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:22:20,009 ==================== TRACER ====================== +2021-12-30 10:22:20,010 Channel (server worker num[20]): +2021-12-30 10:22:20,011 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:22:20,012 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:22:50,042 ==================== TRACER ====================== +2021-12-30 10:22:50,043 Channel (server worker num[20]): +2021-12-30 10:22:50,044 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:22:50,045 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:23:20,075 ==================== TRACER ====================== +2021-12-30 10:23:20,076 Channel (server worker num[20]): +2021-12-30 10:23:20,077 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:23:20,078 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:23:50,108 ==================== TRACER ====================== +2021-12-30 10:23:50,109 Channel (server worker num[20]): +2021-12-30 10:23:50,110 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:23:50,111 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:24:20,141 ==================== TRACER ====================== +2021-12-30 10:24:20,142 Channel (server worker num[20]): +2021-12-30 10:24:20,143 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:24:20,144 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:24:50,165 ==================== TRACER ====================== +2021-12-30 10:24:50,166 Channel (server worker num[20]): +2021-12-30 10:24:50,167 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:24:50,168 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:25:20,198 ==================== TRACER ====================== +2021-12-30 10:25:20,199 Channel (server worker num[20]): +2021-12-30 10:25:20,200 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:25:20,201 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:25:50,231 ==================== TRACER ====================== +2021-12-30 10:25:50,232 Channel (server worker num[20]): +2021-12-30 10:25:50,233 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:25:50,234 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:26:20,264 ==================== TRACER ====================== +2021-12-30 10:26:20,265 Channel (server worker num[20]): +2021-12-30 10:26:20,266 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:26:20,267 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:26:50,297 ==================== TRACER ====================== +2021-12-30 10:26:50,298 Channel (server worker num[20]): +2021-12-30 10:26:50,299 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:26:50,300 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:27:20,330 ==================== TRACER ====================== +2021-12-30 10:27:20,331 Channel (server worker num[20]): +2021-12-30 10:27:20,332 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:27:20,333 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:27:50,363 ==================== TRACER ====================== +2021-12-30 10:27:50,364 Channel (server worker num[20]): +2021-12-30 10:27:50,365 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:27:50,366 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:28:20,396 ==================== TRACER ====================== +2021-12-30 10:28:20,397 Channel (server worker num[20]): +2021-12-30 10:28:20,398 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:28:20,399 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:28:50,429 ==================== TRACER ====================== +2021-12-30 10:28:50,430 Channel (server worker num[20]): +2021-12-30 10:28:50,431 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:28:50,432 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:29:20,462 ==================== TRACER ====================== +2021-12-30 10:29:20,463 Channel (server worker num[20]): +2021-12-30 10:29:20,464 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:29:20,465 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:29:50,495 ==================== TRACER ====================== +2021-12-30 10:29:50,496 Channel (server worker num[20]): +2021-12-30 10:29:50,497 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:29:50,498 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:30:20,528 ==================== TRACER ====================== +2021-12-30 10:30:20,529 Channel (server worker num[20]): +2021-12-30 10:30:20,530 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:30:20,531 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:30:50,561 ==================== TRACER ====================== +2021-12-30 10:30:50,562 Channel (server worker num[20]): +2021-12-30 10:30:50,563 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:30:50,564 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:31:20,594 ==================== TRACER ====================== +2021-12-30 10:31:20,595 Channel (server worker num[20]): +2021-12-30 10:31:20,596 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:31:20,597 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:31:50,627 ==================== TRACER ====================== +2021-12-30 10:31:50,628 Channel (server worker num[20]): +2021-12-30 10:31:50,629 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:31:50,630 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:32:20,660 ==================== TRACER ====================== +2021-12-30 10:32:20,662 Channel (server worker num[20]): +2021-12-30 10:32:20,662 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:32:20,663 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:32:50,694 ==================== TRACER ====================== +2021-12-30 10:32:50,695 Channel (server worker num[20]): +2021-12-30 10:32:50,695 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:32:50,696 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:33:20,727 ==================== TRACER ====================== +2021-12-30 10:33:20,728 Channel (server worker num[20]): +2021-12-30 10:33:20,733 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:33:20,733 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:33:50,764 ==================== TRACER ====================== +2021-12-30 10:33:50,765 Channel (server worker num[20]): +2021-12-30 10:33:50,766 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:33:50,766 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:34:20,797 ==================== TRACER ====================== +2021-12-30 10:34:20,798 Channel (server worker num[20]): +2021-12-30 10:34:20,799 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:34:20,799 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:34:50,830 ==================== TRACER ====================== +2021-12-30 10:34:50,831 Channel (server worker num[20]): +2021-12-30 10:34:50,831 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:34:50,832 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:35:20,863 ==================== TRACER ====================== +2021-12-30 10:35:20,864 Channel (server worker num[20]): +2021-12-30 10:35:20,864 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:35:20,865 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:35:50,896 ==================== TRACER ====================== +2021-12-30 10:35:50,897 Channel (server worker num[20]): +2021-12-30 10:35:50,898 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:35:50,898 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:36:20,929 ==================== TRACER ====================== +2021-12-30 10:36:20,930 Channel (server worker num[20]): +2021-12-30 10:36:20,930 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:36:20,931 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:36:50,962 ==================== TRACER ====================== +2021-12-30 10:36:50,963 Channel (server worker num[20]): +2021-12-30 10:36:50,963 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:36:50,964 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:37:20,995 ==================== TRACER ====================== +2021-12-30 10:37:20,996 Channel (server worker num[20]): +2021-12-30 10:37:20,996 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:37:20,997 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:37:51,028 ==================== TRACER ====================== +2021-12-30 10:37:51,029 Channel (server worker num[20]): +2021-12-30 10:37:51,030 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:37:51,030 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:38:21,061 ==================== TRACER ====================== +2021-12-30 10:38:21,062 Channel (server worker num[20]): +2021-12-30 10:38:21,062 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:38:21,063 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:38:51,094 ==================== TRACER ====================== +2021-12-30 10:38:51,095 Channel (server worker num[20]): +2021-12-30 10:38:51,095 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:38:51,096 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:39:21,127 ==================== TRACER ====================== +2021-12-30 10:39:21,128 Channel (server worker num[20]): +2021-12-30 10:39:21,128 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:39:21,129 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:39:51,160 ==================== TRACER ====================== +2021-12-30 10:39:51,161 Channel (server worker num[20]): +2021-12-30 10:39:51,161 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:39:51,162 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:40:21,185 ==================== TRACER ====================== +2021-12-30 10:40:21,186 Channel (server worker num[20]): +2021-12-30 10:40:21,186 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:40:21,187 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:40:51,218 ==================== TRACER ====================== +2021-12-30 10:40:51,219 Channel (server worker num[20]): +2021-12-30 10:40:51,219 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:40:51,220 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:41:21,244 ==================== TRACER ====================== +2021-12-30 10:41:21,245 Channel (server worker num[20]): +2021-12-30 10:41:21,246 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:41:21,247 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:41:51,277 ==================== TRACER ====================== +2021-12-30 10:41:51,278 Channel (server worker num[20]): +2021-12-30 10:41:51,279 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:41:51,280 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:42:21,310 ==================== TRACER ====================== +2021-12-30 10:42:21,311 Channel (server worker num[20]): +2021-12-30 10:42:21,312 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:42:21,313 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:42:51,343 ==================== TRACER ====================== +2021-12-30 10:42:51,344 Channel (server worker num[20]): +2021-12-30 10:42:51,345 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:42:51,345 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:43:21,369 ==================== TRACER ====================== +2021-12-30 10:43:21,370 Channel (server worker num[20]): +2021-12-30 10:43:21,371 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:43:21,372 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:43:51,402 ==================== TRACER ====================== +2021-12-30 10:43:51,403 Channel (server worker num[20]): +2021-12-30 10:43:51,404 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:43:51,405 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:44:21,436 ==================== TRACER ====================== +2021-12-30 10:44:21,437 Channel (server worker num[20]): +2021-12-30 10:44:21,437 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:44:21,438 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:44:51,469 ==================== TRACER ====================== +2021-12-30 10:44:51,469 Channel (server worker num[20]): +2021-12-30 10:44:51,470 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:44:51,471 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:45:21,501 ==================== TRACER ====================== +2021-12-30 10:45:21,502 Channel (server worker num[20]): +2021-12-30 10:45:21,503 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:45:21,504 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:45:51,534 ==================== TRACER ====================== +2021-12-30 10:45:51,535 Channel (server worker num[20]): +2021-12-30 10:45:51,536 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:45:51,537 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:46:21,568 ==================== TRACER ====================== +2021-12-30 10:46:21,569 Channel (server worker num[20]): +2021-12-30 10:46:21,569 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:46:21,570 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:46:51,584 ==================== TRACER ====================== +2021-12-30 10:46:51,585 Channel (server worker num[20]): +2021-12-30 10:46:51,586 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:46:51,587 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:47:21,617 ==================== TRACER ====================== +2021-12-30 10:47:21,618 Channel (server worker num[20]): +2021-12-30 10:47:21,619 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:47:21,620 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:47:51,624 ==================== TRACER ====================== +2021-12-30 10:47:51,625 Channel (server worker num[20]): +2021-12-30 10:47:51,626 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:47:51,627 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:48:21,656 ==================== TRACER ====================== +2021-12-30 10:48:21,657 Channel (server worker num[20]): +2021-12-30 10:48:21,658 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:48:21,658 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:48:51,689 ==================== TRACER ====================== +2021-12-30 10:48:51,690 Channel (server worker num[20]): +2021-12-30 10:48:51,691 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:48:51,691 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:49:21,722 ==================== TRACER ====================== +2021-12-30 10:49:21,723 Channel (server worker num[20]): +2021-12-30 10:49:21,724 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:49:21,724 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:49:51,755 ==================== TRACER ====================== +2021-12-30 10:49:51,755 Channel (server worker num[20]): +2021-12-30 10:49:51,756 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:49:51,757 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:50:21,787 ==================== TRACER ====================== +2021-12-30 10:50:21,788 Channel (server worker num[20]): +2021-12-30 10:50:21,789 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:50:21,790 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:50:51,820 ==================== TRACER ====================== +2021-12-30 10:50:51,821 Channel (server worker num[20]): +2021-12-30 10:50:51,822 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:50:51,823 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:51:21,853 ==================== TRACER ====================== +2021-12-30 10:51:21,854 Channel (server worker num[20]): +2021-12-30 10:51:21,855 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:51:21,856 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:51:51,886 ==================== TRACER ====================== +2021-12-30 10:51:51,887 Channel (server worker num[20]): +2021-12-30 10:51:51,888 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:51:51,888 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:52:21,896 ==================== TRACER ====================== +2021-12-30 10:52:21,897 Channel (server worker num[20]): +2021-12-30 10:52:21,898 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:52:21,899 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:52:51,929 ==================== TRACER ====================== +2021-12-30 10:52:51,930 Channel (server worker num[20]): +2021-12-30 10:52:51,930 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:52:51,931 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:53:21,961 ==================== TRACER ====================== +2021-12-30 10:53:21,962 Channel (server worker num[20]): +2021-12-30 10:53:21,963 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:53:21,964 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:53:51,994 ==================== TRACER ====================== +2021-12-30 10:53:51,995 Channel (server worker num[20]): +2021-12-30 10:53:51,996 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:53:51,997 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:54:22,027 ==================== TRACER ====================== +2021-12-30 10:54:22,028 Channel (server worker num[20]): +2021-12-30 10:54:22,029 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:54:22,030 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:54:52,060 ==================== TRACER ====================== +2021-12-30 10:54:52,061 Channel (server worker num[20]): +2021-12-30 10:54:52,062 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:54:52,062 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:55:22,072 ==================== TRACER ====================== +2021-12-30 10:55:22,073 Channel (server worker num[20]): +2021-12-30 10:55:22,074 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:55:22,074 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:55:52,105 ==================== TRACER ====================== +2021-12-30 10:55:52,106 Channel (server worker num[20]): +2021-12-30 10:55:52,107 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:55:52,107 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:56:22,138 ==================== TRACER ====================== +2021-12-30 10:56:22,139 Channel (server worker num[20]): +2021-12-30 10:56:22,139 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:56:22,140 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:56:52,170 ==================== TRACER ====================== +2021-12-30 10:56:52,171 Channel (server worker num[20]): +2021-12-30 10:56:52,172 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:56:52,173 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:57:22,203 ==================== TRACER ====================== +2021-12-30 10:57:22,204 Channel (server worker num[20]): +2021-12-30 10:57:22,205 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:57:22,206 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:57:52,236 ==================== TRACER ====================== +2021-12-30 10:57:52,237 Channel (server worker num[20]): +2021-12-30 10:57:52,238 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:57:52,238 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:58:22,269 ==================== TRACER ====================== +2021-12-30 10:58:22,270 Channel (server worker num[20]): +2021-12-30 10:58:22,270 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:58:22,271 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:58:52,302 ==================== TRACER ====================== +2021-12-30 10:58:52,303 Channel (server worker num[20]): +2021-12-30 10:58:52,303 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:58:52,304 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:59:22,335 ==================== TRACER ====================== +2021-12-30 10:59:22,335 Channel (server worker num[20]): +2021-12-30 10:59:22,336 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:59:22,337 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 10:59:52,367 ==================== TRACER ====================== +2021-12-30 10:59:52,368 Channel (server worker num[20]): +2021-12-30 10:59:52,369 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 10:59:52,370 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:00:22,400 ==================== TRACER ====================== +2021-12-30 11:00:22,401 Channel (server worker num[20]): +2021-12-30 11:00:22,402 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:00:22,403 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:00:52,433 ==================== TRACER ====================== +2021-12-30 11:00:52,434 Channel (server worker num[20]): +2021-12-30 11:00:52,435 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:00:52,436 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:01:22,466 ==================== TRACER ====================== +2021-12-30 11:01:22,467 Channel (server worker num[20]): +2021-12-30 11:01:22,468 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:01:22,469 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:01:52,499 ==================== TRACER ====================== +2021-12-30 11:01:52,500 Channel (server worker num[20]): +2021-12-30 11:01:52,501 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:01:52,502 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:02:22,511 ==================== TRACER ====================== +2021-12-30 11:02:22,512 Channel (server worker num[20]): +2021-12-30 11:02:22,512 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:02:22,513 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:02:52,544 ==================== TRACER ====================== +2021-12-30 11:02:52,544 Channel (server worker num[20]): +2021-12-30 11:02:52,545 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:02:52,546 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:03:22,576 ==================== TRACER ====================== +2021-12-30 11:03:22,577 Channel (server worker num[20]): +2021-12-30 11:03:22,578 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:03:22,579 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:03:52,609 ==================== TRACER ====================== +2021-12-30 11:03:52,610 Channel (server worker num[20]): +2021-12-30 11:03:52,611 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:03:52,612 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:04:22,642 ==================== TRACER ====================== +2021-12-30 11:04:22,643 Channel (server worker num[20]): +2021-12-30 11:04:22,644 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:04:22,644 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:04:52,675 ==================== TRACER ====================== +2021-12-30 11:04:52,675 Channel (server worker num[20]): +2021-12-30 11:04:52,676 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:04:52,677 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:05:22,699 ==================== TRACER ====================== +2021-12-30 11:05:22,700 Channel (server worker num[20]): +2021-12-30 11:05:22,701 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:05:22,702 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:05:52,732 ==================== TRACER ====================== +2021-12-30 11:05:52,734 DAGExecutor: +2021-12-30 11:05:52,734 Query count[1] +2021-12-30 11:05:52,734 QPS[0.03333333333333333 q/s] +2021-12-30 11:05:52,734 Succ[0.0] +2021-12-30 11:05:52,734 Error req[2] +2021-12-30 11:05:52,735 Latency: +2021-12-30 11:05:52,735 ave[111.508 ms] +2021-12-30 11:05:52,735 .50[111.508 ms] +2021-12-30 11:05:52,735 .60[111.508 ms] +2021-12-30 11:05:52,735 .70[111.508 ms] +2021-12-30 11:05:52,736 .80[111.508 ms] +2021-12-30 11:05:52,736 .90[111.508 ms] +2021-12-30 11:05:52,736 .95[111.508 ms] +2021-12-30 11:05:52,736 .99[111.508 ms] +2021-12-30 11:05:52,736 Channel (server worker num[20]): +2021-12-30 11:05:52,737 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:05:52,738 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:05:54,677 ==================== TRACER ====================== +2021-12-30 11:05:54,679 Channel (server worker num[20]): +2021-12-30 11:05:54,681 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:05:54,682 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:06:24,712 ==================== TRACER ====================== +2021-12-30 11:06:24,714 Op(ppyolo_mbv3): +2021-12-30 11:06:24,714 in[3.782 ms] +2021-12-30 11:06:24,714 prep[61.954 ms] +2021-12-30 11:06:24,715 midp[1570.405 ms] +2021-12-30 11:06:24,715 postp[10.504 ms] +2021-12-30 11:06:24,715 out[1.288 ms] +2021-12-30 11:06:24,715 idle[0.003076581390141553] +2021-12-30 11:06:24,715 DAGExecutor: +2021-12-30 11:06:24,715 Query count[1] +2021-12-30 11:06:24,716 QPS[0.03333333333333333 q/s] +2021-12-30 11:06:24,716 Succ[1.0] +2021-12-30 11:06:24,716 Error req[] +2021-12-30 11:06:24,716 Latency: +2021-12-30 11:06:24,716 ave[2632.994 ms] +2021-12-30 11:06:24,716 .50[2632.994 ms] +2021-12-30 11:06:24,717 .60[2632.994 ms] +2021-12-30 11:06:24,717 .70[2632.994 ms] +2021-12-30 11:06:24,717 .80[2632.994 ms] +2021-12-30 11:06:24,717 .90[2632.994 ms] +2021-12-30 11:06:24,717 .95[2632.994 ms] +2021-12-30 11:06:24,718 .99[2632.994 ms] +2021-12-30 11:06:24,718 Channel (server worker num[20]): +2021-12-30 11:06:24,718 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:06:24,719 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:06:54,750 ==================== TRACER ====================== +2021-12-30 11:06:54,750 Channel (server worker num[20]): +2021-12-30 11:06:54,751 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:06:54,752 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:07:24,782 ==================== TRACER ====================== +2021-12-30 11:07:24,783 Channel (server worker num[20]): +2021-12-30 11:07:24,784 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:07:24,785 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:07:54,815 ==================== TRACER ====================== +2021-12-30 11:07:54,816 Channel (server worker num[20]): +2021-12-30 11:07:54,817 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:07:54,818 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:08:24,848 ==================== TRACER ====================== +2021-12-30 11:08:24,849 Channel (server worker num[20]): +2021-12-30 11:08:24,850 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:08:24,851 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:08:54,881 ==================== TRACER ====================== +2021-12-30 11:08:54,882 Channel (server worker num[20]): +2021-12-30 11:08:54,883 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:08:54,883 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:09:24,914 ==================== TRACER ====================== +2021-12-30 11:09:24,915 Channel (server worker num[20]): +2021-12-30 11:09:24,915 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:09:24,916 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:09:54,946 ==================== TRACER ====================== +2021-12-30 11:09:54,947 Channel (server worker num[20]): +2021-12-30 11:09:54,948 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:09:54,949 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:10:24,979 ==================== TRACER ====================== +2021-12-30 11:10:24,980 Channel (server worker num[20]): +2021-12-30 11:10:24,981 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:10:24,982 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:10:55,012 ==================== TRACER ====================== +2021-12-30 11:10:55,013 Channel (server worker num[20]): +2021-12-30 11:10:55,014 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:10:55,015 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:11:25,045 ==================== TRACER ====================== +2021-12-30 11:11:25,046 Channel (server worker num[20]): +2021-12-30 11:11:25,047 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:11:25,047 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:11:55,078 ==================== TRACER ====================== +2021-12-30 11:11:55,079 Channel (server worker num[20]): +2021-12-30 11:11:55,080 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:11:55,080 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:12:25,111 ==================== TRACER ====================== +2021-12-30 11:12:25,112 Channel (server worker num[20]): +2021-12-30 11:12:25,112 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:12:25,113 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:12:55,144 ==================== TRACER ====================== +2021-12-30 11:12:55,144 Channel (server worker num[20]): +2021-12-30 11:12:55,145 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:12:55,146 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:13:25,176 ==================== TRACER ====================== +2021-12-30 11:13:25,177 Channel (server worker num[20]): +2021-12-30 11:13:25,178 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:13:25,179 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:13:55,209 ==================== TRACER ====================== +2021-12-30 11:13:55,210 Channel (server worker num[20]): +2021-12-30 11:13:55,211 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:13:55,212 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:14:25,242 ==================== TRACER ====================== +2021-12-30 11:14:25,243 Channel (server worker num[20]): +2021-12-30 11:14:25,244 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:14:25,244 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:14:55,275 ==================== TRACER ====================== +2021-12-30 11:14:55,276 Channel (server worker num[20]): +2021-12-30 11:14:55,276 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:14:55,277 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:15:25,308 ==================== TRACER ====================== +2021-12-30 11:15:25,308 Channel (server worker num[20]): +2021-12-30 11:15:25,309 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:15:25,310 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:15:55,340 ==================== TRACER ====================== +2021-12-30 11:15:55,341 DAGExecutor: +2021-12-30 11:15:55,343 Query count[1] +2021-12-30 11:15:55,343 QPS[0.03333333333333333 q/s] +2021-12-30 11:15:55,343 Succ[0.0] +2021-12-30 11:15:55,344 Error req[1] +2021-12-30 11:15:55,344 Latency: +2021-12-30 11:15:55,344 ave[115.746 ms] +2021-12-30 11:15:55,344 .50[115.746 ms] +2021-12-30 11:15:55,344 .60[115.746 ms] +2021-12-30 11:15:55,344 .70[115.746 ms] +2021-12-30 11:15:55,345 .80[115.746 ms] +2021-12-30 11:15:55,345 .90[115.746 ms] +2021-12-30 11:15:55,345 .95[115.746 ms] +2021-12-30 11:15:55,345 .99[115.746 ms] +2021-12-30 11:15:55,345 Channel (server worker num[20]): +2021-12-30 11:15:55,346 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:15:55,347 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:16:25,377 ==================== TRACER ====================== +2021-12-30 11:16:25,378 Channel (server worker num[20]): +2021-12-30 11:16:25,379 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:16:25,380 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:16:55,410 ==================== TRACER ====================== +2021-12-30 11:16:55,411 Channel (server worker num[20]): +2021-12-30 11:16:55,412 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:16:55,412 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:17:25,443 ==================== TRACER ====================== +2021-12-30 11:17:25,444 Channel (server worker num[20]): +2021-12-30 11:17:25,444 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:17:25,445 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:17:55,476 ==================== TRACER ====================== +2021-12-30 11:17:55,476 Channel (server worker num[20]): +2021-12-30 11:17:55,477 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:17:55,478 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:18:25,508 ==================== TRACER ====================== +2021-12-30 11:18:25,509 Channel (server worker num[20]): +2021-12-30 11:18:25,510 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:18:25,511 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:18:55,513 ==================== TRACER ====================== +2021-12-30 11:18:55,514 Channel (server worker num[20]): +2021-12-30 11:18:55,515 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:18:55,515 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:19:25,546 ==================== TRACER ====================== +2021-12-30 11:19:25,546 Channel (server worker num[20]): +2021-12-30 11:19:25,547 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:19:25,548 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:19:55,578 ==================== TRACER ====================== +2021-12-30 11:19:55,579 Channel (server worker num[20]): +2021-12-30 11:19:55,580 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:19:55,581 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:20:25,611 ==================== TRACER ====================== +2021-12-30 11:20:25,612 Channel (server worker num[20]): +2021-12-30 11:20:25,613 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:20:25,614 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:20:55,644 ==================== TRACER ====================== +2021-12-30 11:20:55,645 Channel (server worker num[20]): +2021-12-30 11:20:55,646 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:20:55,647 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:21:25,664 ==================== TRACER ====================== +2021-12-30 11:21:25,665 Channel (server worker num[20]): +2021-12-30 11:21:25,666 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:21:25,667 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:21:55,697 ==================== TRACER ====================== +2021-12-30 11:21:55,698 Channel (server worker num[20]): +2021-12-30 11:21:55,699 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:21:55,699 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:22:25,730 ==================== TRACER ====================== +2021-12-30 11:22:25,731 Channel (server worker num[20]): +2021-12-30 11:22:25,731 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:22:25,732 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:22:55,762 ==================== TRACER ====================== +2021-12-30 11:22:55,763 Channel (server worker num[20]): +2021-12-30 11:22:55,764 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:22:55,765 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:23:25,768 ==================== TRACER ====================== +2021-12-30 11:23:25,769 Channel (server worker num[20]): +2021-12-30 11:23:25,770 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:23:25,771 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:23:55,801 ==================== TRACER ====================== +2021-12-30 11:23:55,802 Channel (server worker num[20]): +2021-12-30 11:23:55,803 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:23:55,803 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:24:24,537 ==================== TRACER ====================== +2021-12-30 11:24:24,538 Channel (server worker num[20]): +2021-12-30 11:24:24,540 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:24:24,540 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:24:54,571 ==================== TRACER ====================== +2021-12-30 11:24:54,572 Op(ppyolo_mbv3): +2021-12-30 11:24:54,573 in[1233.596 ms] +2021-12-30 11:24:54,573 prep[79.632 ms] +2021-12-30 11:24:54,573 midp[1828.419 ms] +2021-12-30 11:24:54,573 postp[25.206 ms] +2021-12-30 11:24:54,573 out[1.554 ms] +2021-12-30 11:24:54,574 idle[0.38983312434292694] +2021-12-30 11:24:54,574 DAGExecutor: +2021-12-30 11:24:54,574 Query count[2] +2021-12-30 11:24:54,574 QPS[0.06666666666666667 q/s] +2021-12-30 11:24:54,574 Succ[0.5] +2021-12-30 11:24:54,574 Error req[1] +2021-12-30 11:24:54,575 Latency: +2021-12-30 11:24:54,575 ave[1038.919 ms] +2021-12-30 11:24:54,575 .50[1947.202 ms] +2021-12-30 11:24:54,575 .60[1947.202 ms] +2021-12-30 11:24:54,575 .70[1947.202 ms] +2021-12-30 11:24:54,575 .80[1947.202 ms] +2021-12-30 11:24:54,576 .90[1947.202 ms] +2021-12-30 11:24:54,576 .95[1947.202 ms] +2021-12-30 11:24:54,576 .99[1947.202 ms] +2021-12-30 11:24:54,576 Channel (server worker num[20]): +2021-12-30 11:24:54,577 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:24:54,578 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:25:24,608 ==================== TRACER ====================== +2021-12-30 11:25:24,609 Channel (server worker num[20]): +2021-12-30 11:25:24,610 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:25:24,610 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:37:49,731 ==================== TRACER ====================== +2021-12-30 11:37:49,733 Channel (server worker num[20]): +2021-12-30 11:37:49,735 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:37:49,736 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:38:09,368 ==================== TRACER ====================== +2021-12-30 11:38:09,370 Channel (server worker num[20]): +2021-12-30 11:38:09,373 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:38:09,373 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:38:39,404 ==================== TRACER ====================== +2021-12-30 11:38:39,405 DAGExecutor: +2021-12-30 11:38:39,405 Query count[1] +2021-12-30 11:38:39,405 QPS[0.03333333333333333 q/s] +2021-12-30 11:38:39,406 Succ[0.0] +2021-12-30 11:38:39,406 Error req[0] +2021-12-30 11:38:39,406 Latency: +2021-12-30 11:38:39,406 ave[2798.136 ms] +2021-12-30 11:38:39,406 .50[2798.136 ms] +2021-12-30 11:38:39,406 .60[2798.136 ms] +2021-12-30 11:38:39,407 .70[2798.136 ms] +2021-12-30 11:38:39,407 .80[2798.136 ms] +2021-12-30 11:38:39,407 .90[2798.136 ms] +2021-12-30 11:38:39,407 .95[2798.136 ms] +2021-12-30 11:38:39,407 .99[2798.136 ms] +2021-12-30 11:38:39,407 Channel (server worker num[20]): +2021-12-30 11:38:39,408 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:38:39,409 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:39:09,439 ==================== TRACER ====================== +2021-12-30 11:39:09,440 Channel (server worker num[20]): +2021-12-30 11:39:09,441 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:39:09,442 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:39:39,472 ==================== TRACER ====================== +2021-12-30 11:39:39,473 Channel (server worker num[20]): +2021-12-30 11:39:39,474 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:39:39,474 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:40:05,674 ==================== TRACER ====================== +2021-12-30 11:40:05,677 Channel (server worker num[20]): +2021-12-30 11:40:05,679 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:40:05,679 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:40:35,710 ==================== TRACER ====================== +2021-12-30 11:40:35,712 Op(ppyolo_mbv3): +2021-12-30 11:40:35,712 in[1703.2785 ms] +2021-12-30 11:40:35,712 prep[66.81 ms] +2021-12-30 11:40:35,713 midp[792.723 ms] +2021-12-30 11:40:35,713 postp[10.012 ms] +2021-12-30 11:40:35,713 out[1.1055 ms] +2021-12-30 11:40:35,713 idle[0.6621721111965404] +2021-12-30 11:40:35,713 DAGExecutor: +2021-12-30 11:40:35,713 Query count[2] +2021-12-30 11:40:35,714 QPS[0.06666666666666667 q/s] +2021-12-30 11:40:35,714 Succ[1.0] +2021-12-30 11:40:35,714 Error req[] +2021-12-30 11:40:35,714 Latency: +2021-12-30 11:40:35,714 ave[880.886 ms] +2021-12-30 11:40:35,714 .50[1658.566 ms] +2021-12-30 11:40:35,715 .60[1658.566 ms] +2021-12-30 11:40:35,715 .70[1658.566 ms] +2021-12-30 11:40:35,715 .80[1658.566 ms] +2021-12-30 11:40:35,715 .90[1658.566 ms] +2021-12-30 11:40:35,715 .95[1658.566 ms] +2021-12-30 11:40:35,715 .99[1658.566 ms] +2021-12-30 11:40:35,716 Channel (server worker num[20]): +2021-12-30 11:40:35,716 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:40:35,717 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:41:05,746 ==================== TRACER ====================== +2021-12-30 11:41:05,747 Op(ppyolo_mbv3): +2021-12-30 11:41:05,748 in[45842.715 ms] +2021-12-30 11:41:05,748 prep[76.967 ms] +2021-12-30 11:41:05,748 midp[18.287 ms] +2021-12-30 11:41:05,748 postp[9.692 ms] +2021-12-30 11:41:05,748 out[1.296 ms] +2021-12-30 11:41:05,749 idle[0.9977160308557167] +2021-12-30 11:41:05,749 DAGExecutor: +2021-12-30 11:41:05,749 Query count[1] +2021-12-30 11:41:05,749 QPS[0.03333333333333333 q/s] +2021-12-30 11:41:05,749 Succ[1.0] +2021-12-30 11:41:05,750 Error req[] +2021-12-30 11:41:05,750 Latency: +2021-12-30 11:41:05,750 ave[118.137 ms] +2021-12-30 11:41:05,750 .50[118.137 ms] +2021-12-30 11:41:05,750 .60[118.137 ms] +2021-12-30 11:41:05,750 .70[118.137 ms] +2021-12-30 11:41:05,751 .80[118.137 ms] +2021-12-30 11:41:05,751 .90[118.137 ms] +2021-12-30 11:41:05,751 .95[118.137 ms] +2021-12-30 11:41:05,751 .99[118.137 ms] +2021-12-30 11:41:05,751 Channel (server worker num[20]): +2021-12-30 11:41:05,752 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:41:05,753 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:41:27,033 ==================== TRACER ====================== +2021-12-30 11:41:27,035 Channel (server worker num[20]): +2021-12-30 11:41:27,037 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:41:27,038 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:41:57,060 ==================== TRACER ====================== +2021-12-30 11:41:57,062 Op(ppyolo_mbv3): +2021-12-30 11:41:57,062 in[1008.0535 ms] +2021-12-30 11:41:57,063 prep[61.519 ms] +2021-12-30 11:41:57,063 midp[796.019 ms] +2021-12-30 11:41:57,063 postp[10.4375 ms] +2021-12-30 11:41:57,063 out[1.295 ms] +2021-12-30 11:41:57,063 idle[0.537652797279532] +2021-12-30 11:41:57,063 DAGExecutor: +2021-12-30 11:41:57,064 Query count[2] +2021-12-30 11:41:57,064 QPS[0.06666666666666667 q/s] +2021-12-30 11:41:57,064 Succ[1.0] +2021-12-30 11:41:57,064 Error req[] +2021-12-30 11:41:57,064 Latency: +2021-12-30 11:41:57,065 ave[1179.855 ms] +2021-12-30 11:41:57,065 .50[2258.924 ms] +2021-12-30 11:41:57,065 .60[2258.924 ms] +2021-12-30 11:41:57,065 .70[2258.924 ms] +2021-12-30 11:41:57,065 .80[2258.924 ms] +2021-12-30 11:41:57,065 .90[2258.924 ms] +2021-12-30 11:41:57,066 .95[2258.924 ms] +2021-12-30 11:41:57,066 .99[2258.924 ms] +2021-12-30 11:41:57,066 Channel (server worker num[20]): +2021-12-30 11:41:57,067 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:41:57,067 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:42:27,098 ==================== TRACER ====================== +2021-12-30 11:42:27,099 Channel (server worker num[20]): +2021-12-30 11:42:27,099 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:42:27,100 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:42:57,131 ==================== TRACER ====================== +2021-12-30 11:42:57,131 Channel (server worker num[20]): +2021-12-30 11:42:57,132 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:42:57,133 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:43:27,163 ==================== TRACER ====================== +2021-12-30 11:43:27,164 Channel (server worker num[20]): +2021-12-30 11:43:27,165 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:43:27,166 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:43:57,196 ==================== TRACER ====================== +2021-12-30 11:43:57,197 Channel (server worker num[20]): +2021-12-30 11:43:57,198 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:43:57,199 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:44:27,229 ==================== TRACER ====================== +2021-12-30 11:44:27,230 Channel (server worker num[20]): +2021-12-30 11:44:27,231 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:44:27,231 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:44:57,262 ==================== TRACER ====================== +2021-12-30 11:44:57,263 Channel (server worker num[20]): +2021-12-30 11:44:57,264 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:44:57,264 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:45:27,295 ==================== TRACER ====================== +2021-12-30 11:45:27,296 Channel (server worker num[20]): +2021-12-30 11:45:27,296 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:45:27,297 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:45:57,328 ==================== TRACER ====================== +2021-12-30 11:45:57,328 Channel (server worker num[20]): +2021-12-30 11:45:57,329 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:45:57,330 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:46:27,360 ==================== TRACER ====================== +2021-12-30 11:46:27,361 Channel (server worker num[20]): +2021-12-30 11:46:27,362 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:46:27,363 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:46:57,393 ==================== TRACER ====================== +2021-12-30 11:46:57,394 Channel (server worker num[20]): +2021-12-30 11:46:57,395 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:46:57,396 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:47:27,426 ==================== TRACER ====================== +2021-12-30 11:47:27,427 Channel (server worker num[20]): +2021-12-30 11:47:27,428 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:47:27,428 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:47:57,459 ==================== TRACER ====================== +2021-12-30 11:47:57,460 Channel (server worker num[20]): +2021-12-30 11:47:57,460 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:47:57,461 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:48:27,492 ==================== TRACER ====================== +2021-12-30 11:48:27,492 Channel (server worker num[20]): +2021-12-30 11:48:27,493 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:48:27,494 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:48:57,524 ==================== TRACER ====================== +2021-12-30 11:48:57,525 Channel (server worker num[20]): +2021-12-30 11:48:57,526 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:48:57,527 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:49:27,557 ==================== TRACER ====================== +2021-12-30 11:49:27,558 Channel (server worker num[20]): +2021-12-30 11:49:27,559 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:49:27,560 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:49:57,590 ==================== TRACER ====================== +2021-12-30 11:49:57,591 Channel (server worker num[20]): +2021-12-30 11:49:57,592 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:49:57,593 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2021-12-30 11:50:27,623 ==================== TRACER ====================== +2021-12-30 11:50:27,624 Channel (server worker num[20]): +2021-12-30 11:50:27,625 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2021-12-30 11:50:27,625 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-14 09:24:33,177 ==================== TRACER ====================== +2022-02-14 09:24:33,179 Channel (server worker num[20]): +2022-02-14 09:24:33,182 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-14 09:24:33,182 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-14 09:24:42,822 ==================== TRACER ====================== +2022-02-14 09:24:42,824 Channel (server worker num[20]): +2022-02-14 09:24:42,827 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-14 09:24:42,827 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-14 09:26:03,784 ==================== TRACER ====================== +2022-02-14 09:26:03,786 Channel (server worker num[20]): +2022-02-14 09:26:03,789 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-14 09:26:03,789 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 16:56:51,916 ==================== TRACER ====================== +2022-02-16 16:56:51,917 Channel (server worker num[20]): +2022-02-16 16:56:51,918 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 16:56:51,918 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:05:23,231 ==================== TRACER ====================== +2022-02-16 17:05:23,232 Channel (server worker num[20]): +2022-02-16 17:05:23,233 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:05:23,234 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:05:53,260 ==================== TRACER ====================== +2022-02-16 17:05:53,261 Op(ppyolo_mbv3): +2022-02-16 17:05:53,261 in[10785.568 ms] +2022-02-16 17:05:53,261 prep[48.4655 ms] +2022-02-16 17:05:53,261 midp[1298.1825 ms] +2022-02-16 17:05:53,261 postp[9.903 ms] +2022-02-16 17:05:53,261 out[0.8555 ms] +2022-02-16 17:05:53,261 idle[0.888285114985624] +2022-02-16 17:05:53,261 DAGExecutor: +2022-02-16 17:05:53,261 Query count[2] +2022-02-16 17:05:53,262 QPS[0.06666666666666667 q/s] +2022-02-16 17:05:53,262 Succ[1.0] +2022-02-16 17:05:53,262 Error req[] +2022-02-16 17:05:53,262 Latency: +2022-02-16 17:05:53,262 ave[1365.0625 ms] +2022-02-16 17:05:53,262 .50[2649.873 ms] +2022-02-16 17:05:53,262 .60[2649.873 ms] +2022-02-16 17:05:53,262 .70[2649.873 ms] +2022-02-16 17:05:53,262 .80[2649.873 ms] +2022-02-16 17:05:53,262 .90[2649.873 ms] +2022-02-16 17:05:53,262 .95[2649.873 ms] +2022-02-16 17:05:53,262 .99[2649.873 ms] +2022-02-16 17:05:53,262 Channel (server worker num[20]): +2022-02-16 17:05:53,262 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:05:53,263 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:06:23,264 ==================== TRACER ====================== +2022-02-16 17:06:23,265 Channel (server worker num[20]): +2022-02-16 17:06:23,265 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:06:23,265 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:06:53,294 ==================== TRACER ====================== +2022-02-16 17:06:53,294 Channel (server worker num[20]): +2022-02-16 17:06:53,295 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:06:53,295 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:07:23,303 ==================== TRACER ====================== +2022-02-16 17:07:23,304 Op(ppyolo_mbv3): +2022-02-16 17:07:23,304 in[69773.431 ms] +2022-02-16 17:07:23,304 prep[41.191 ms] +2022-02-16 17:07:23,304 midp[11.872 ms] +2022-02-16 17:07:23,305 postp[9.658 ms] +2022-02-16 17:07:23,305 out[0.679 ms] +2022-02-16 17:07:23,305 idle[0.9991018922379225] +2022-02-16 17:07:23,305 DAGExecutor: +2022-02-16 17:07:23,305 Query count[1] +2022-02-16 17:07:23,305 QPS[0.03333333333333333 q/s] +2022-02-16 17:07:23,305 Succ[1.0] +2022-02-16 17:07:23,305 Error req[] +2022-02-16 17:07:23,305 Latency: +2022-02-16 17:07:23,305 ave[69.518 ms] +2022-02-16 17:07:23,305 .50[69.518 ms] +2022-02-16 17:07:23,305 .60[69.518 ms] +2022-02-16 17:07:23,305 .70[69.518 ms] +2022-02-16 17:07:23,305 .80[69.518 ms] +2022-02-16 17:07:23,305 .90[69.518 ms] +2022-02-16 17:07:23,305 .95[69.518 ms] +2022-02-16 17:07:23,305 .99[69.518 ms] +2022-02-16 17:07:23,305 Channel (server worker num[20]): +2022-02-16 17:07:23,306 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:07:23,306 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:09:22,123 ==================== TRACER ====================== +2022-02-16 17:09:22,124 Channel (server worker num[20]): +2022-02-16 17:09:22,125 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:09:22,126 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:09:52,156 ==================== TRACER ====================== +2022-02-16 17:09:52,157 Channel (server worker num[20]): +2022-02-16 17:09:52,157 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:09:52,157 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:10:22,184 ==================== TRACER ====================== +2022-02-16 17:10:22,185 Op(ppyolo_mbv3): +2022-02-16 17:10:22,185 in[35440.544 ms] +2022-02-16 17:10:22,185 prep[42.793 ms] +2022-02-16 17:10:22,185 midp[2504.427 ms] +2022-02-16 17:10:22,185 postp[10.631 ms] +2022-02-16 17:10:22,185 out[0.959 ms] +2022-02-16 17:10:22,186 idle[0.9326869872577308] +2022-02-16 17:10:22,186 DAGExecutor: +2022-02-16 17:10:22,186 Query count[1] +2022-02-16 17:10:22,186 QPS[0.03333333333333333 q/s] +2022-02-16 17:10:22,186 Succ[1.0] +2022-02-16 17:10:22,186 Error req[] +2022-02-16 17:10:22,186 Latency: +2022-02-16 17:10:22,186 ave[2566.559 ms] +2022-02-16 17:10:22,186 .50[2566.559 ms] +2022-02-16 17:10:22,186 .60[2566.559 ms] +2022-02-16 17:10:22,186 .70[2566.559 ms] +2022-02-16 17:10:22,186 .80[2566.559 ms] +2022-02-16 17:10:22,186 .90[2566.559 ms] +2022-02-16 17:10:22,186 .95[2566.559 ms] +2022-02-16 17:10:22,186 .99[2566.559 ms] +2022-02-16 17:10:22,186 Channel (server worker num[20]): +2022-02-16 17:10:22,187 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:10:22,187 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:10:52,217 ==================== TRACER ====================== +2022-02-16 17:10:52,218 Channel (server worker num[20]): +2022-02-16 17:10:52,219 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:10:52,219 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:11:22,249 ==================== TRACER ====================== +2022-02-16 17:11:22,250 Channel (server worker num[20]): +2022-02-16 17:11:22,250 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:11:22,250 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:11:52,264 ==================== TRACER ====================== +2022-02-16 17:11:52,265 Channel (server worker num[20]): +2022-02-16 17:11:52,265 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:11:52,265 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:12:22,292 ==================== TRACER ====================== +2022-02-16 17:12:22,292 Channel (server worker num[20]): +2022-02-16 17:12:22,293 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:12:22,293 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:12:52,319 ==================== TRACER ====================== +2022-02-16 17:12:52,320 Channel (server worker num[20]): +2022-02-16 17:12:52,321 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:12:52,321 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:13:22,323 ==================== TRACER ====================== +2022-02-16 17:13:22,324 Channel (server worker num[20]): +2022-02-16 17:13:22,324 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:13:22,324 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:13:52,354 ==================== TRACER ====================== +2022-02-16 17:13:52,355 Channel (server worker num[20]): +2022-02-16 17:13:52,355 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:13:52,356 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:14:22,373 ==================== TRACER ====================== +2022-02-16 17:14:22,374 Channel (server worker num[20]): +2022-02-16 17:14:22,374 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:14:22,374 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:14:52,404 ==================== TRACER ====================== +2022-02-16 17:14:52,405 Op(ppyolo_mbv3): +2022-02-16 17:14:52,405 in[289118.996 ms] +2022-02-16 17:14:52,405 prep[46.16 ms] +2022-02-16 17:14:52,405 midp[11.854 ms] +2022-02-16 17:14:52,405 postp[9.602 ms] +2022-02-16 17:14:52,405 out[0.799 ms] +2022-02-16 17:14:52,405 idle[0.9997661862258589] +2022-02-16 17:14:52,405 DAGExecutor: +2022-02-16 17:14:52,405 Query count[1] +2022-02-16 17:14:52,406 QPS[0.03333333333333333 q/s] +2022-02-16 17:14:52,406 Succ[1.0] +2022-02-16 17:14:52,406 Error req[] +2022-02-16 17:14:52,406 Latency: +2022-02-16 17:14:52,406 ave[76.35 ms] +2022-02-16 17:14:52,406 .50[76.35 ms] +2022-02-16 17:14:52,406 .60[76.35 ms] +2022-02-16 17:14:52,406 .70[76.35 ms] +2022-02-16 17:14:52,406 .80[76.35 ms] +2022-02-16 17:14:52,406 .90[76.35 ms] +2022-02-16 17:14:52,406 .95[76.35 ms] +2022-02-16 17:14:52,406 .99[76.35 ms] +2022-02-16 17:14:52,406 Channel (server worker num[20]): +2022-02-16 17:14:52,406 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:14:52,407 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:15:22,428 ==================== TRACER ====================== +2022-02-16 17:15:22,429 Op(ppyolo_mbv3): +2022-02-16 17:15:22,429 in[14226.853 ms] +2022-02-16 17:15:22,430 prep[40.298 ms] +2022-02-16 17:15:22,430 midp[11.486 ms] +2022-02-16 17:15:22,430 postp[9.523 ms] +2022-02-16 17:15:22,430 out[0.712 ms] +2022-02-16 17:15:22,430 idle[0.9957094583813194] +2022-02-16 17:15:22,430 DAGExecutor: +2022-02-16 17:15:22,430 Query count[1] +2022-02-16 17:15:22,430 QPS[0.03333333333333333 q/s] +2022-02-16 17:15:22,430 Succ[1.0] +2022-02-16 17:15:22,430 Error req[] +2022-02-16 17:15:22,430 Latency: +2022-02-16 17:15:22,430 ave[68.343 ms] +2022-02-16 17:15:22,430 .50[68.343 ms] +2022-02-16 17:15:22,430 .60[68.343 ms] +2022-02-16 17:15:22,430 .70[68.343 ms] +2022-02-16 17:15:22,430 .80[68.343 ms] +2022-02-16 17:15:22,430 .90[68.343 ms] +2022-02-16 17:15:22,430 .95[68.343 ms] +2022-02-16 17:15:22,430 .99[68.343 ms] +2022-02-16 17:15:22,430 Channel (server worker num[20]): +2022-02-16 17:15:22,431 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:15:22,431 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:15:52,452 ==================== TRACER ====================== +2022-02-16 17:15:52,452 Channel (server worker num[20]): +2022-02-16 17:15:52,453 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:15:52,453 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:16:22,472 ==================== TRACER ====================== +2022-02-16 17:16:22,473 Channel (server worker num[20]): +2022-02-16 17:16:22,473 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:16:22,474 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:16:52,504 ==================== TRACER ====================== +2022-02-16 17:16:52,504 Channel (server worker num[20]): +2022-02-16 17:16:52,505 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:16:52,505 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:17:22,529 ==================== TRACER ====================== +2022-02-16 17:17:22,530 Channel (server worker num[20]): +2022-02-16 17:17:22,530 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:17:22,530 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:17:52,560 ==================== TRACER ====================== +2022-02-16 17:17:52,561 Channel (server worker num[20]): +2022-02-16 17:17:52,562 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:17:52,562 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:18:22,592 ==================== TRACER ====================== +2022-02-16 17:18:22,593 Channel (server worker num[20]): +2022-02-16 17:18:22,593 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:18:22,593 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:18:52,624 ==================== TRACER ====================== +2022-02-16 17:18:52,624 Channel (server worker num[20]): +2022-02-16 17:18:52,625 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:18:52,625 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:19:22,655 ==================== TRACER ====================== +2022-02-16 17:19:22,656 Channel (server worker num[20]): +2022-02-16 17:19:22,656 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:19:22,656 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:19:52,679 ==================== TRACER ====================== +2022-02-16 17:19:52,679 Channel (server worker num[20]): +2022-02-16 17:19:52,680 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:19:52,680 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:20:22,710 ==================== TRACER ====================== +2022-02-16 17:20:22,711 Channel (server worker num[20]): +2022-02-16 17:20:22,711 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:20:22,711 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:20:52,740 ==================== TRACER ====================== +2022-02-16 17:20:52,741 Channel (server worker num[20]): +2022-02-16 17:20:52,741 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:20:52,741 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:21:22,771 ==================== TRACER ====================== +2022-02-16 17:21:22,772 Channel (server worker num[20]): +2022-02-16 17:21:22,772 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:21:22,773 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:21:52,795 ==================== TRACER ====================== +2022-02-16 17:21:52,795 Op(ppyolo_mbv3): +2022-02-16 17:21:52,796 in[378619.8 ms] +2022-02-16 17:21:52,796 prep[40.567 ms] +2022-02-16 17:21:52,796 midp[11.664 ms] +2022-02-16 17:21:52,796 postp[9.171 ms] +2022-02-16 17:21:52,796 out[0.723 ms] +2022-02-16 17:21:52,796 idle[0.9998378533646675] +2022-02-16 17:21:52,796 DAGExecutor: +2022-02-16 17:21:52,796 Query count[1] +2022-02-16 17:21:52,796 QPS[0.03333333333333333 q/s] +2022-02-16 17:21:52,796 Succ[1.0] +2022-02-16 17:21:52,796 Error req[] +2022-02-16 17:21:52,796 Latency: +2022-02-16 17:21:52,796 ave[67.215 ms] +2022-02-16 17:21:52,796 .50[67.215 ms] +2022-02-16 17:21:52,796 .60[67.215 ms] +2022-02-16 17:21:52,796 .70[67.215 ms] +2022-02-16 17:21:52,796 .80[67.215 ms] +2022-02-16 17:21:52,796 .90[67.215 ms] +2022-02-16 17:21:52,796 .95[67.215 ms] +2022-02-16 17:21:52,796 .99[67.215 ms] +2022-02-16 17:21:52,796 Channel (server worker num[20]): +2022-02-16 17:21:52,797 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:21:52,797 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:22:22,827 ==================== TRACER ====================== +2022-02-16 17:22:22,828 Op(ppyolo_mbv3): +2022-02-16 17:22:22,828 in[28828.316 ms] +2022-02-16 17:22:22,828 prep[40.625 ms] +2022-02-16 17:22:22,828 midp[11.704 ms] +2022-02-16 17:22:22,828 postp[9.511 ms] +2022-02-16 17:22:22,828 out[0.838 ms] +2022-02-16 17:22:22,828 idle[0.9978595405890154] +2022-02-16 17:22:22,828 DAGExecutor: +2022-02-16 17:22:22,828 Query count[1] +2022-02-16 17:22:22,828 QPS[0.03333333333333333 q/s] +2022-02-16 17:22:22,828 Succ[1.0] +2022-02-16 17:22:22,828 Error req[] +2022-02-16 17:22:22,828 Latency: +2022-02-16 17:22:22,829 ave[69.462 ms] +2022-02-16 17:22:22,829 .50[69.462 ms] +2022-02-16 17:22:22,829 .60[69.462 ms] +2022-02-16 17:22:22,829 .70[69.462 ms] +2022-02-16 17:22:22,829 .80[69.462 ms] +2022-02-16 17:22:22,829 .90[69.462 ms] +2022-02-16 17:22:22,829 .95[69.462 ms] +2022-02-16 17:22:22,829 .99[69.462 ms] +2022-02-16 17:22:22,829 Channel (server worker num[20]): +2022-02-16 17:22:22,829 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:22:22,829 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:22:52,860 ==================== TRACER ====================== +2022-02-16 17:22:52,860 Op(ppyolo_mbv3): +2022-02-16 17:22:52,860 in[53789.073 ms] +2022-02-16 17:22:52,861 prep[41.018 ms] +2022-02-16 17:22:52,861 midp[11.874 ms] +2022-02-16 17:22:52,861 postp[9.638 ms] +2022-02-16 17:22:52,861 out[0.767 ms] +2022-02-16 17:22:52,861 idle[0.9988388626164456] +2022-02-16 17:22:52,861 DAGExecutor: +2022-02-16 17:22:52,861 Query count[1] +2022-02-16 17:22:52,861 QPS[0.03333333333333333 q/s] +2022-02-16 17:22:52,861 Succ[1.0] +2022-02-16 17:22:52,861 Error req[] +2022-02-16 17:22:52,861 Latency: +2022-02-16 17:22:52,861 ave[69.775 ms] +2022-02-16 17:22:52,861 .50[69.775 ms] +2022-02-16 17:22:52,861 .60[69.775 ms] +2022-02-16 17:22:52,861 .70[69.775 ms] +2022-02-16 17:22:52,861 .80[69.775 ms] +2022-02-16 17:22:52,861 .90[69.775 ms] +2022-02-16 17:22:52,861 .95[69.775 ms] +2022-02-16 17:22:52,861 .99[69.775 ms] +2022-02-16 17:22:52,861 Channel (server worker num[20]): +2022-02-16 17:22:52,862 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:22:52,862 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:23:22,869 ==================== TRACER ====================== +2022-02-16 17:23:22,870 Op(ppyolo_mbv3): +2022-02-16 17:23:22,870 in[16002.513 ms] +2022-02-16 17:23:22,870 prep[40.564 ms] +2022-02-16 17:23:22,870 midp[11.608 ms] +2022-02-16 17:23:22,870 postp[9.549 ms] +2022-02-16 17:23:22,870 out[0.796 ms] +2022-02-16 17:23:22,870 idle[0.9961580526149033] +2022-02-16 17:23:22,870 DAGExecutor: +2022-02-16 17:23:22,870 Query count[1] +2022-02-16 17:23:22,871 QPS[0.03333333333333333 q/s] +2022-02-16 17:23:22,871 Succ[1.0] +2022-02-16 17:23:22,871 Error req[] +2022-02-16 17:23:22,871 Latency: +2022-02-16 17:23:22,871 ave[68.978 ms] +2022-02-16 17:23:22,871 .50[68.978 ms] +2022-02-16 17:23:22,871 .60[68.978 ms] +2022-02-16 17:23:22,871 .70[68.978 ms] +2022-02-16 17:23:22,871 .80[68.978 ms] +2022-02-16 17:23:22,871 .90[68.978 ms] +2022-02-16 17:23:22,871 .95[68.978 ms] +2022-02-16 17:23:22,871 .99[68.978 ms] +2022-02-16 17:23:22,871 Channel (server worker num[20]): +2022-02-16 17:23:22,871 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:23:22,872 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:23:52,895 ==================== TRACER ====================== +2022-02-16 17:23:52,895 Op(ppyolo_mbv3): +2022-02-16 17:23:52,896 in[26647.47 ms] +2022-02-16 17:23:52,896 prep[41.165 ms] +2022-02-16 17:23:52,896 midp[11.559 ms] +2022-02-16 17:23:52,896 postp[9.357 ms] +2022-02-16 17:23:52,896 out[0.736 ms] +2022-02-16 17:23:52,896 idle[0.9976757643974399] +2022-02-16 17:23:52,896 DAGExecutor: +2022-02-16 17:23:52,896 Query count[1] +2022-02-16 17:23:52,896 QPS[0.03333333333333333 q/s] +2022-02-16 17:23:52,896 Succ[1.0] +2022-02-16 17:23:52,896 Error req[] +2022-02-16 17:23:52,896 Latency: +2022-02-16 17:23:52,896 ave[69.043 ms] +2022-02-16 17:23:52,896 .50[69.043 ms] +2022-02-16 17:23:52,896 .60[69.043 ms] +2022-02-16 17:23:52,896 .70[69.043 ms] +2022-02-16 17:23:52,896 .80[69.043 ms] +2022-02-16 17:23:52,896 .90[69.043 ms] +2022-02-16 17:23:52,896 .95[69.043 ms] +2022-02-16 17:23:52,896 .99[69.043 ms] +2022-02-16 17:23:52,896 Channel (server worker num[20]): +2022-02-16 17:23:52,897 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:23:52,897 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:24:22,925 ==================== TRACER ====================== +2022-02-16 17:24:22,926 Op(ppyolo_mbv3): +2022-02-16 17:24:22,926 in[42124.841 ms] +2022-02-16 17:24:22,926 prep[42.925 ms] +2022-02-16 17:24:22,926 midp[12.13 ms] +2022-02-16 17:24:22,926 postp[9.767 ms] +2022-02-16 17:24:22,926 out[0.895 ms] +2022-02-16 17:24:22,926 idle[0.9984635898866282] +2022-02-16 17:24:22,926 DAGExecutor: +2022-02-16 17:24:22,926 Query count[1] +2022-02-16 17:24:22,926 QPS[0.03333333333333333 q/s] +2022-02-16 17:24:22,926 Succ[1.0] +2022-02-16 17:24:22,926 Error req[] +2022-02-16 17:24:22,926 Latency: +2022-02-16 17:24:22,927 ave[73.248 ms] +2022-02-16 17:24:22,927 .50[73.248 ms] +2022-02-16 17:24:22,927 .60[73.248 ms] +2022-02-16 17:24:22,927 .70[73.248 ms] +2022-02-16 17:24:22,927 .80[73.248 ms] +2022-02-16 17:24:22,927 .90[73.248 ms] +2022-02-16 17:24:22,927 .95[73.248 ms] +2022-02-16 17:24:22,927 .99[73.248 ms] +2022-02-16 17:24:22,927 Channel (server worker num[20]): +2022-02-16 17:24:22,927 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:24:22,927 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:24:52,956 ==================== TRACER ====================== +2022-02-16 17:24:52,957 Op(ppyolo_mbv3): +2022-02-16 17:24:52,957 in[15533.448 ms] +2022-02-16 17:24:52,957 prep[42.1205 ms] +2022-02-16 17:24:52,957 midp[11.651 ms] +2022-02-16 17:24:52,957 postp[9.449 ms] +2022-02-16 17:24:52,957 out[0.808 ms] +2022-02-16 17:24:52,957 idle[0.9959467481807073] +2022-02-16 17:24:52,957 DAGExecutor: +2022-02-16 17:24:52,957 Query count[2] +2022-02-16 17:24:52,957 QPS[0.06666666666666667 q/s] +2022-02-16 17:24:52,957 Succ[1.0] +2022-02-16 17:24:52,957 Error req[] +2022-02-16 17:24:52,957 Latency: +2022-02-16 17:24:52,957 ave[69.7875 ms] +2022-02-16 17:24:52,957 .50[69.917 ms] +2022-02-16 17:24:52,957 .60[69.917 ms] +2022-02-16 17:24:52,958 .70[69.917 ms] +2022-02-16 17:24:52,958 .80[69.917 ms] +2022-02-16 17:24:52,958 .90[69.917 ms] +2022-02-16 17:24:52,958 .95[69.917 ms] +2022-02-16 17:24:52,958 .99[69.917 ms] +2022-02-16 17:24:52,958 Channel (server worker num[20]): +2022-02-16 17:24:52,958 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:24:52,958 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:25:22,988 ==================== TRACER ====================== +2022-02-16 17:25:22,989 Op(ppyolo_mbv3): +2022-02-16 17:25:22,989 in[14139.5015 ms] +2022-02-16 17:25:22,990 prep[40.245 ms] +2022-02-16 17:25:22,990 midp[11.6935 ms] +2022-02-16 17:25:22,990 postp[9.9075 ms] +2022-02-16 17:25:22,990 out[0.7675 ms] +2022-02-16 17:25:22,990 idle[0.9956452964928111] +2022-02-16 17:25:22,990 DAGExecutor: +2022-02-16 17:25:22,990 Query count[2] +2022-02-16 17:25:22,990 QPS[0.06666666666666667 q/s] +2022-02-16 17:25:22,990 Succ[1.0] +2022-02-16 17:25:22,990 Error req[] +2022-02-16 17:25:22,990 Latency: +2022-02-16 17:25:22,990 ave[68.187 ms] +2022-02-16 17:25:22,990 .50[68.877 ms] +2022-02-16 17:25:22,990 .60[68.877 ms] +2022-02-16 17:25:22,990 .70[68.877 ms] +2022-02-16 17:25:22,990 .80[68.877 ms] +2022-02-16 17:25:22,990 .90[68.877 ms] +2022-02-16 17:25:22,990 .95[68.877 ms] +2022-02-16 17:25:22,990 .99[68.877 ms] +2022-02-16 17:25:22,990 Channel (server worker num[20]): +2022-02-16 17:25:22,991 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:25:22,991 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:25:53,021 ==================== TRACER ====================== +2022-02-16 17:25:53,022 Channel (server worker num[20]): +2022-02-16 17:25:53,022 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:25:53,022 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:26:23,052 ==================== TRACER ====================== +2022-02-16 17:26:23,053 Channel (server worker num[20]): +2022-02-16 17:26:23,054 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:26:23,054 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:26:53,066 ==================== TRACER ====================== +2022-02-16 17:26:53,067 Channel (server worker num[20]): +2022-02-16 17:26:53,068 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:26:53,068 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:27:23,098 ==================== TRACER ====================== +2022-02-16 17:27:23,099 Channel (server worker num[20]): +2022-02-16 17:27:23,099 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:27:23,099 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:27:53,128 ==================== TRACER ====================== +2022-02-16 17:27:53,129 Channel (server worker num[20]): +2022-02-16 17:27:53,129 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:27:53,129 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:28:23,147 ==================== TRACER ====================== +2022-02-16 17:28:23,147 Channel (server worker num[20]): +2022-02-16 17:28:23,148 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:28:23,148 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:28:53,154 ==================== TRACER ====================== +2022-02-16 17:28:53,155 Channel (server worker num[20]): +2022-02-16 17:28:53,156 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:28:53,156 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:29:23,157 ==================== TRACER ====================== +2022-02-16 17:29:23,157 Channel (server worker num[20]): +2022-02-16 17:29:23,158 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:29:23,158 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:29:53,180 ==================== TRACER ====================== +2022-02-16 17:29:53,180 Channel (server worker num[20]): +2022-02-16 17:29:53,181 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:29:53,181 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:30:23,211 ==================== TRACER ====================== +2022-02-16 17:30:23,212 Channel (server worker num[20]): +2022-02-16 17:30:23,212 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:30:23,213 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:30:53,221 ==================== TRACER ====================== +2022-02-16 17:30:53,222 Channel (server worker num[20]): +2022-02-16 17:30:53,222 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:30:53,222 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:31:23,252 ==================== TRACER ====================== +2022-02-16 17:31:23,253 Channel (server worker num[20]): +2022-02-16 17:31:23,253 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:31:23,254 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:31:53,284 ==================== TRACER ====================== +2022-02-16 17:31:53,285 Channel (server worker num[20]): +2022-02-16 17:31:53,285 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:31:53,285 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:32:23,291 ==================== TRACER ====================== +2022-02-16 17:32:23,291 Channel (server worker num[20]): +2022-02-16 17:32:23,292 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:32:23,292 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:32:53,322 ==================== TRACER ====================== +2022-02-16 17:32:53,323 Channel (server worker num[20]): +2022-02-16 17:32:53,327 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:32:53,327 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:33:23,341 ==================== TRACER ====================== +2022-02-16 17:33:23,342 Channel (server worker num[20]): +2022-02-16 17:33:23,342 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:33:23,342 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:33:53,361 ==================== TRACER ====================== +2022-02-16 17:33:53,361 Channel (server worker num[20]): +2022-02-16 17:33:53,362 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:33:53,362 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:34:23,365 ==================== TRACER ====================== +2022-02-16 17:34:23,365 Channel (server worker num[20]): +2022-02-16 17:34:23,366 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:34:23,366 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] +2022-02-16 17:34:53,368 ==================== TRACER ====================== +2022-02-16 17:34:53,369 Channel (server worker num[20]): +2022-02-16 17:34:53,369 chl0(In: ['@DAGExecutor'], Out: ['ppyolo_mbv3']) size[0/0] +2022-02-16 17:34:53,370 chl1(In: ['ppyolo_mbv3'], Out: ['@DAGExecutor']) size[0/0] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/ProcessInfo.json b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/ProcessInfo.json new file mode 100644 index 000000000..300ca0031 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/ProcessInfo.json @@ -0,0 +1 @@ +[{"pid": 8611, "port": [9999, 2009], "model": "pipline", "start_time": 1645002562.0576003}] \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/picodet_postprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/picodet_postprocess.cpython-37.pyc new file mode 100644 index 000000000..668e81847 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/picodet_postprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/preprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/preprocess.cpython-37.pyc new file mode 100644 index 000000000..d6993735d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/__pycache__/preprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/config.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/config.yml new file mode 100644 index 000000000..2eea1e1d0 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/config.yml @@ -0,0 +1,25 @@ +dag: + is_thread_op: false + tracer: + interval_s: 30 +http_port: 2009 +op: + ppyolo_mbv3: + concurrency: 1 + + local_service_conf: + client_type: local_predictor + device_type: 2 + devices: '0' + fetch_list: + - save_infer_model/scale_0.tmp_1 + - save_infer_model/scale_1.tmp_1 + - save_infer_model/scale_2.tmp_1 + - save_infer_model/scale_3.tmp_1 + - save_infer_model/scale_4.tmp_1 + - save_infer_model/scale_5.tmp_1 + - save_infer_model/scale_6.tmp_1 + - save_infer_model/scale_7.tmp_1 + model_config: serving_server/ +rpc_port: 9999 +worker_num: 20 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/infer_cfg.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/infer_cfg.yml new file mode 100644 index 000000000..e29f9298f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/infer_cfg.yml @@ -0,0 +1,118 @@ +mode: fluid +draw_threshold: 0.5 +metric: COCO +use_dynamic_shape: false +arch: PicoDet +min_subgraph_size: 3 +Preprocess: +- interp: 2 + keep_ratio: false + target_size: + - 640 + - 640 + type: Resize +- is_scale: true + mean: + - 0.485 + - 0.456 + - 0.406 + std: + - 0.229 + - 0.224 + - 0.225 + type: NormalizeImage +- type: Permute +- stride: 32 + type: PadStride +label_list: +- person +- bicycle +- car +- motorcycle +- airplane +- bus +- train +- truck +- boat +- traffic light +- fire hydrant +- stop sign +- parking meter +- bench +- bird +- cat +- dog +- horse +- sheep +- cow +- elephant +- bear +- zebra +- giraffe +- backpack +- umbrella +- handbag +- tie +- suitcase +- frisbee +- skis +- snowboard +- sports ball +- kite +- baseball bat +- baseball glove +- skateboard +- surfboard +- tennis racket +- bottle +- wine glass +- cup +- fork +- knife +- spoon +- bowl +- banana +- apple +- sandwich +- orange +- broccoli +- carrot +- hot dog +- pizza +- donut +- cake +- chair +- couch +- potted plant +- bed +- dining table +- toilet +- tv +- laptop +- mouse +- remote +- keyboard +- cell phone +- microwave +- oven +- toaster +- sink +- refrigerator +- book +- clock +- vase +- scissors +- teddy bear +- hair drier +- toothbrush +NMS: + keep_top_k: 100 + name: MultiClassNMS + nms_threshold: 0.5 + nms_top_k: 1000 + score_threshold: 0.3 +fpn_stride: +- 8 +- 16 +- 32 +- 64 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/picodet_postprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/picodet_postprocess.py new file mode 100644 index 000000000..7b8159f4e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/picodet_postprocess.py @@ -0,0 +1,228 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +from scipy.special import softmax + + +def hard_nms(box_scores, iou_threshold, top_k=-1, candidate_size=200): + """ + Args: + box_scores (N, 5): boxes in corner-form and probabilities. + iou_threshold: intersection over union threshold. + top_k: keep top_k results. If k <= 0, keep all the results. + candidate_size: only consider the candidates with the highest scores. + Returns: + picked: a list of indexes of the kept boxes + """ + scores = box_scores[:, -1] + boxes = box_scores[:, :-1] + picked = [] + indexes = np.argsort(scores) + indexes = indexes[-candidate_size:] + while len(indexes) > 0: + current = indexes[-1] + picked.append(current) + if 0 < top_k == len(picked) or len(indexes) == 1: + break + current_box = boxes[current, :] + indexes = indexes[:-1] + rest_boxes = boxes[indexes, :] + iou = iou_of( + rest_boxes, + np.expand_dims( + current_box, axis=0), ) + indexes = indexes[iou <= iou_threshold] + + return box_scores[picked, :] + + +def iou_of(boxes0, boxes1, eps=1e-5): + """Return intersection-over-union (Jaccard index) of boxes. + Args: + boxes0 (N, 4): ground truth boxes. + boxes1 (N or 1, 4): predicted boxes. + eps: a small number to avoid 0 as denominator. + Returns: + iou (N): IoU values. + """ + overlap_left_top = np.maximum(boxes0[..., :2], boxes1[..., :2]) + overlap_right_bottom = np.minimum(boxes0[..., 2:], boxes1[..., 2:]) + + overlap_area = area_of(overlap_left_top, overlap_right_bottom) + area0 = area_of(boxes0[..., :2], boxes0[..., 2:]) + area1 = area_of(boxes1[..., :2], boxes1[..., 2:]) + return overlap_area / (area0 + area1 - overlap_area + eps) + + +def area_of(left_top, right_bottom): + """Compute the areas of rectangles given two corners. + Args: + left_top (N, 2): left top corner. + right_bottom (N, 2): right bottom corner. + Returns: + area (N): return the area. + """ + hw = np.clip(right_bottom - left_top, 0.0, None) + return hw[..., 0] * hw[..., 1] + + +class PicoDetPostProcess(object): + """ + Args: + input_shape (int): network input image size + ori_shape (int): ori image shape of before padding + scale_factor (float): scale factor of ori image + enable_mkldnn (bool): whether to open MKLDNN + """ + + def __init__(self, + input_shape, + ori_shape, + scale_factor, + strides=[8, 16, 32, 64], + score_threshold=0.4, + nms_threshold=0.5, + nms_top_k=1000, + keep_top_k=100): + self.ori_shape = ori_shape + self.input_shape = input_shape + self.scale_factor = scale_factor + self.strides = strides + self.score_threshold = score_threshold + self.nms_threshold = nms_threshold + self.nms_top_k = nms_top_k + self.keep_top_k = keep_top_k + + def warp_boxes(self, boxes, ori_shape): + """Apply transform to boxes + """ + width, height = ori_shape[1], ori_shape[0] + n = len(boxes) + if n: + # warp points + xy = np.ones((n * 4, 3)) + xy[:, :2] = boxes[:, [0, 1, 2, 3, 0, 3, 2, 1]].reshape( + n * 4, 2) # x1y1, x2y2, x1y2, x2y1 + # xy = xy @ M.T # transform + xy = (xy[:, :2] / xy[:, 2:3]).reshape(n, 8) # rescale + # create new boxes + x = xy[:, [0, 2, 4, 6]] + y = xy[:, [1, 3, 5, 7]] + xy = np.concatenate( + (x.min(1), y.min(1), x.max(1), y.max(1))).reshape(4, n).T + # clip boxes + xy[:, [0, 2]] = xy[:, [0, 2]].clip(0, width) + xy[:, [1, 3]] = xy[:, [1, 3]].clip(0, height) + return xy.astype(np.float32) + else: + return boxes + + def __call__(self, scores, raw_boxes): + batch_size = raw_boxes[0].shape[0] + reg_max = int(raw_boxes[0].shape[-1] / 4 - 1) + out_boxes_num = [] + out_boxes_list = [] + for batch_id in range(batch_size): + # generate centers + decode_boxes = [] + select_scores = [] + for stride, box_distribute, score in zip(self.strides, raw_boxes, + scores): + box_distribute = box_distribute[batch_id] + score = score[batch_id] + # centers + fm_h = self.input_shape[0] / stride + fm_w = self.input_shape[1] / stride + h_range = np.arange(fm_h) + w_range = np.arange(fm_w) + ww, hh = np.meshgrid(w_range, h_range) + ct_row = (hh.flatten() + 0.5) * stride + ct_col = (ww.flatten() + 0.5) * stride + center = np.stack((ct_col, ct_row, ct_col, ct_row), axis=1) + + # box distribution to distance + reg_range = np.arange(reg_max + 1) + box_distance = box_distribute.reshape((-1, reg_max + 1)) + box_distance = softmax(box_distance, axis=1) + box_distance = box_distance * np.expand_dims(reg_range, axis=0) + box_distance = np.sum(box_distance, axis=1).reshape((-1, 4)) + box_distance = box_distance * stride + + # top K candidate + topk_idx = np.argsort(score.max(axis=1))[::-1] + topk_idx = topk_idx[:self.nms_top_k] + center = center[topk_idx] + score = score[topk_idx] + box_distance = box_distance[topk_idx] + + # decode box + decode_box = center + [-1, -1, 1, 1] * box_distance + + select_scores.append(score) + decode_boxes.append(decode_box) + + # nms + bboxes = np.concatenate(decode_boxes, axis=0) + confidences = np.concatenate(select_scores, axis=0) + picked_box_probs = [] + picked_labels = [] + for class_index in range(0, confidences.shape[1]): + probs = confidences[:, class_index] + mask = probs > self.score_threshold + probs = probs[mask] + if probs.shape[0] == 0: + continue + subset_boxes = bboxes[mask, :] + box_probs = np.concatenate( + [subset_boxes, probs.reshape(-1, 1)], axis=1) + box_probs = hard_nms( + box_probs, + iou_threshold=self.nms_threshold, + top_k=self.keep_top_k, ) + picked_box_probs.append(box_probs) + picked_labels.extend([class_index] * box_probs.shape[0]) + + if len(picked_box_probs) == 0: + out_boxes_list.append(np.empty((0, 4))) + out_boxes_num.append(0) + + else: + picked_box_probs = np.concatenate(picked_box_probs) + + # resize output boxes + picked_box_probs[:, :4] = self.warp_boxes( + picked_box_probs[:, :4], self.ori_shape[batch_id]) + im_scale = np.concatenate([ + self.scale_factor[batch_id][::-1], + self.scale_factor[batch_id][::-1] + ]) + picked_box_probs[:, :4] /= im_scale + # clas score box + out_boxes_list.append( + np.concatenate( + [ + np.expand_dims( + np.array(picked_labels), + axis=-1), np.expand_dims( + picked_box_probs[:, 4], axis=-1), + picked_box_probs[:, :4] + ], + axis=1)) + out_boxes_num.append(len(picked_labels)) + + out_boxes_list = np.concatenate(out_boxes_list, axis=0) + out_boxes_num = np.asarray(out_boxes_num).astype(np.int32) + + return out_boxes_list, out_boxes_num diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/pipeline_http_client.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/pipeline_http_client.py new file mode 100644 index 000000000..186e29590 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/pipeline_http_client.py @@ -0,0 +1,58 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# from paddle_serving_server.pipeline import PipelineClient +import numpy as np +import requests +import json +import cv2 +import base64 +import os +from time import * +import threading + + +def demo(url,data,i): + begin_time = time() + r = requests.post(url=url, data=json.dumps(data)) + end_time = time() + run_time = end_time-begin_time + print ('绾跨▼ %d 鏃堕棿 %f '%(i,run_time)) + print(r.json()) + + +def cv2_to_base64(image): + return base64.b64encode(image).decode('utf8') + +url = "http://127.0.0.1:2009/ppyolo_mbv3/prediction" +with open(os.path.join(".", "test.jpg"), 'rb') as file: + image_data1 = file.read() +image = cv2_to_base64(image_data1) +category_dict={0.0:"person",1.0:"bicycle",2.0:"motorcycle"} +data = {"key": ["image"], "value": [image]} +r = requests.post(url=url, data=json.dumps(data)) +print(r.json()) +results = eval(r.json()['value'][0]) +img = cv2.imread("test.jpg") +for result in results: + if result["score"] > 0.5: + left, right, top, bottom= int(result['bbox'][0]), int(result['bbox'][2]), int(result['bbox'][1]), int(result['bbox'][3]) + cv2.rectangle(img,(left ,top),(right,bottom), (0, 0, 255), 2) + cv2.putText(img,str(round(result["score"],2)),(left,top-10), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) + print(category_dict[result["category_id"]]) + cv2.putText(img,category_dict[result["category_id"]],(left,top+20), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) +cv2.imwrite("./result.jpg",img) + + + + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/preprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/preprocess.py new file mode 100644 index 000000000..644c8ce3f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/preprocess.py @@ -0,0 +1,395 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import cv2 +import numpy as np + + +def decode_image(im_file, im_info): + """read rgb image + Args: + im_file (str|np.ndarray): input can be image path or np.ndarray + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + if isinstance(im_file, str): + with open(im_file, 'rb') as f: + im_read = f.read() + data = np.frombuffer(im_read, dtype='uint8') + im = cv2.imdecode(data, 1) # BGR mode, but need RGB mode + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + else: + im = im_file + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + im_info['scale_factor'] = np.array([1., 1.], dtype=np.float32) + return im, im_info + + +class Resize(object): + """resize image by target_size and max_size + Args: + target_size (int): the target size of image + keep_ratio (bool): whether keep_ratio or not, default true + interp (int): method of resize + """ + + def __init__(self, target_size, keep_ratio=True, interp=cv2.INTER_LINEAR): + if isinstance(target_size, int): + target_size = [target_size, target_size] + self.target_size = target_size + self.keep_ratio = keep_ratio + self.interp = interp + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + assert len(self.target_size) == 2 + assert self.target_size[0] > 0 and self.target_size[1] > 0 + im_channel = im.shape[2] + im_scale_y, im_scale_x = self.generate_scale(im) + im = cv2.resize( + im, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + im_info['im_shape'] = np.array(im.shape[:2]).astype('float32') + im_info['scale_factor'] = np.array( + [im_scale_y, im_scale_x]).astype('float32') + return im, im_info + + def generate_scale(self, im): + """ + Args: + im (np.ndarray): image (np.ndarray) + Returns: + im_scale_x: the resize ratio of X + im_scale_y: the resize ratio of Y + """ + origin_shape = im.shape[:2] + im_c = im.shape[2] + if self.keep_ratio: + im_size_min = np.min(origin_shape) + im_size_max = np.max(origin_shape) + target_size_min = np.min(self.target_size) + target_size_max = np.max(self.target_size) + im_scale = float(target_size_min) / float(im_size_min) + if np.round(im_scale * im_size_max) > target_size_max: + im_scale = float(target_size_max) / float(im_size_max) + im_scale_x = im_scale + im_scale_y = im_scale + else: + resize_h, resize_w = self.target_size + im_scale_y = resize_h / float(origin_shape[0]) + im_scale_x = resize_w / float(origin_shape[1]) + return im_scale_y, im_scale_x + + +class NormalizeImage(object): + """normalize image + Args: + mean (list): im - mean + std (list): im / std + is_scale (bool): whether need im / 255 + is_channel_first (bool): if True: image shape is CHW, else: HWC + """ + + def __init__(self, mean, std, is_scale=True): + self.mean = mean + self.std = std + self.is_scale = is_scale + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + im = im.astype(np.float32, copy=False) + mean = np.array(self.mean)[np.newaxis, np.newaxis, :] + std = np.array(self.std)[np.newaxis, np.newaxis, :] + + if self.is_scale: + im = im / 255.0 + im -= mean + im /= std + return im, im_info + + +class Permute(object): + """permute image + Args: + to_bgr (bool): whether convert RGB to BGR + channel_first (bool): whether convert HWC to CHW + """ + + def __init__(self, ): + super(Permute, self).__init__() + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + im = im.transpose((2, 0, 1)).copy() + return im, im_info + + +class PadStride(object): + """ padding image for model with FPN, instead PadBatch(pad_to_stride) in original config + Args: + stride (bool): model with FPN need image shape % stride == 0 + """ + + def __init__(self, stride=0): + self.coarsest_stride = stride + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + coarsest_stride = self.coarsest_stride + if coarsest_stride <= 0: + return im, im_info + im_c, im_h, im_w = im.shape + pad_h = int(np.ceil(float(im_h) / coarsest_stride) * coarsest_stride) + pad_w = int(np.ceil(float(im_w) / coarsest_stride) * coarsest_stride) + padding_im = np.zeros((im_c, pad_h, pad_w), dtype=np.float32) + padding_im[:, :im_h, :im_w] = im + return padding_im, im_info + + +class LetterBoxResize(object): + def __init__(self, target_size): + """ + Resize image to target size, convert normalized xywh to pixel xyxy + format ([x_center, y_center, width, height] -> [x0, y0, x1, y1]). + Args: + target_size (int|list): image target size. + """ + super(LetterBoxResize, self).__init__() + if isinstance(target_size, int): + target_size = [target_size, target_size] + self.target_size = target_size + + def letterbox(self, img, height, width, color=(127.5, 127.5, 127.5)): + # letterbox: resize a rectangular image to a padded rectangular + shape = img.shape[:2] # [height, width] + ratio_h = float(height) / shape[0] + ratio_w = float(width) / shape[1] + ratio = min(ratio_h, ratio_w) + new_shape = (round(shape[1] * ratio), + round(shape[0] * ratio)) # [width, height] + padw = (width - new_shape[0]) / 2 + padh = (height - new_shape[1]) / 2 + top, bottom = round(padh - 0.1), round(padh + 0.1) + left, right = round(padw - 0.1), round(padw + 0.1) + + img = cv2.resize( + img, new_shape, interpolation=cv2.INTER_AREA) # resized, no border + img = cv2.copyMakeBorder( + img, top, bottom, left, right, cv2.BORDER_CONSTANT, + value=color) # padded rectangular + return img, ratio, padw, padh + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + assert len(self.target_size) == 2 + assert self.target_size[0] > 0 and self.target_size[1] > 0 + height, width = self.target_size + h, w = im.shape[:2] + im, ratio, padw, padh = self.letterbox(im, height=height, width=width) + + new_shape = [round(h * ratio), round(w * ratio)] + im_info['im_shape'] = np.array(new_shape, dtype=np.float32) + im_info['scale_factor'] = np.array([ratio, ratio], dtype=np.float32) + return im, im_info + + +class WarpAffine(object): + """Warp affine the image + """ + + def __init__(self, + keep_res=False, + pad=31, + input_h=512, + input_w=512, + scale=0.4, + shift=0.1): + self.keep_res = keep_res + self.pad = pad + self.input_h = input_h + self.input_w = input_w + self.scale = scale + self.shift = shift + + def _get_3rd_point(self, a, b): + assert len( + a) == 2, 'input of _get_3rd_point should be point with length of 2' + assert len( + b) == 2, 'input of _get_3rd_point should be point with length of 2' + direction = a - b + third_pt = b + np.array([-direction[1], direction[0]], dtype=np.float32) + return third_pt + + def rotate_point(self, pt, angle_rad): + """Rotate a point by an angle. + + Args: + pt (list[float]): 2 dimensional point to be rotated + angle_rad (float): rotation angle by radian + + Returns: + list[float]: Rotated point. + """ + assert len(pt) == 2 + sn, cs = np.sin(angle_rad), np.cos(angle_rad) + new_x = pt[0] * cs - pt[1] * sn + new_y = pt[0] * sn + pt[1] * cs + rotated_pt = [new_x, new_y] + + return rotated_pt + + def get_affine_transform(self, + center, + input_size, + rot, + output_size, + shift=(0., 0.), + inv=False): + """Get the affine transform matrix, given the center/scale/rot/output_size. + + Args: + center (np.ndarray[2, ]): Center of the bounding box (x, y). + input_size (np.ndarray[2, ]): Size of input feature (width, height). + rot (float): Rotation angle (degree). + output_size (np.ndarray[2, ]): Size of the destination heatmaps. + shift (0-100%): Shift translation ratio wrt the width/height. + Default (0., 0.). + inv (bool): Option to inverse the affine transform direction. + (inv=False: src->dst or inv=True: dst->src) + + Returns: + np.ndarray: The transform matrix. + """ + assert len(center) == 2 + assert len(output_size) == 2 + assert len(shift) == 2 + + if not isinstance(input_size, (np.ndarray, list)): + input_size = np.array([input_size, input_size], dtype=np.float32) + scale_tmp = input_size + + shift = np.array(shift) + src_w = scale_tmp[0] + dst_w = output_size[0] + dst_h = output_size[1] + + rot_rad = np.pi * rot / 180 + src_dir = self.rotate_point([0., src_w * -0.5], rot_rad) + dst_dir = np.array([0., dst_w * -0.5]) + + src = np.zeros((3, 2), dtype=np.float32) + + src[0, :] = center + scale_tmp * shift + src[1, :] = center + src_dir + scale_tmp * shift + src[2, :] = self._get_3rd_point(src[0, :], src[1, :]) + + dst = np.zeros((3, 2), dtype=np.float32) + dst[0, :] = [dst_w * 0.5, dst_h * 0.5] + dst[1, :] = np.array([dst_w * 0.5, dst_h * 0.5]) + dst_dir + dst[2, :] = self._get_3rd_point(dst[0, :], dst[1, :]) + + if inv: + trans = cv2.getAffineTransform(np.float32(dst), np.float32(src)) + else: + trans = cv2.getAffineTransform(np.float32(src), np.float32(dst)) + + return trans + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + img = cv2.cvtColor(im, cv2.COLOR_RGB2BGR) + + h, w = img.shape[:2] + + if self.keep_res: + input_h = (h | self.pad) + 1 + input_w = (w | self.pad) + 1 + s = np.array([input_w, input_h], dtype=np.float32) + c = np.array([w // 2, h // 2], dtype=np.float32) + + else: + s = max(h, w) * 1.0 + input_h, input_w = self.input_h, self.input_w + c = np.array([w / 2., h / 2.], dtype=np.float32) + + trans_input = self.get_affine_transform(c, s, 0, [input_w, input_h]) + img = cv2.resize(img, (w, h)) + inp = cv2.warpAffine( + img, trans_input, (input_w, input_h), flags=cv2.INTER_LINEAR) + return inp, im_info + + +def preprocess(im, preprocess_ops): + # process image by preprocess_ops + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': None, + } + im, im_info = decode_image(im, im_info) + #print(im) + #print(im_info) + #print(preprocess_ops) + #print("preprocess(im, preprocess_ops):................") + for operator in preprocess_ops: + im, im_info = operator(im, im_info) + return im, im_info diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/result.jpg b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/result.jpg new file mode 100644 index 000000000..b20b3f2f6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/result.jpg differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/test.jpg b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/test.jpg new file mode 100644 index 000000000..4f18d55c4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/test.jpg differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/web_service.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/web_service.py new file mode 100644 index 000000000..527cce22c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/picodet_lcnet_1_5x_416_coco/web_service.py @@ -0,0 +1,116 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +from paddle_serving_server.web_service import WebService, Op +import logging +import numpy as np +import sys +import cv2 +from paddle_serving_app.reader import * +import base64 +import os +import yaml +import glob +from picodet_postprocess import PicoDetPostProcess +from preprocess import preprocess, Resize, NormalizeImage, Permute, PadStride, LetterBoxResize, WarpAffine + +class PPYoloMbvOp(Op): + def init_op(self): + self.feed_dict={} + deploy_file = 'infer_cfg.yml' + with open(deploy_file) as f: + yml_conf = yaml.safe_load(f) + preprocess_infos = yml_conf['Preprocess'] + self.preprocess_ops = [] + for op_info in preprocess_infos: + new_op_info = op_info.copy() + op_type = new_op_info.pop('type') + self.preprocess_ops.append(eval(op_type)(**new_op_info)) + #print(self.preprocess_ops) + + def preprocess(self, input_dicts, data_id, log_id): + (_, input_dict), = input_dicts.items() + imgs = [] + for key in input_dict.keys(): + data = base64.b64decode(input_dict[key].encode('utf8')) + data = np.fromstring(data, np.uint8) + im = cv2.imdecode(data, 1) + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': None, + } + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + im_info['scale_factor'] = np.array([1., 1.], dtype=np.float32) + for operator in self.preprocess_ops: + im, im_info = operator(im, im_info) + imgs.append({ + "image": im[np.newaxis,:], + "im_shape": [im_info['im_shape']],#np.array(list(im.shape[1:])).reshape(-1)[np.newaxis,:], + "scale_factor": [im_info['scale_factor']],#np.array([im_scale_y, im_scale_x]).astype('float32'), + }) + self.feed_dict = { + "image": np.concatenate([x["image"] for x in imgs], axis=0), + "im_shape": np.concatenate([x["im_shape"] for x in imgs], axis=0), + "scale_factor": np.concatenate([x["scale_factor"] for x in imgs], axis=0) + } + #print(self.feed_dict) + #for key in self.feed_dict.keys(): + # print(key, self.feed_dict[key].shape) + + return self.feed_dict, False, None, "" + + def postprocess(self, input_dicts, fetch_dict, log_id,data_id =0): + #print(fetch_dict) + np_score_list = [] + np_boxes_list = [] + i = 0 + for value in fetch_dict.values():#range(4): + if i<4: + np_score_list.append(value) + else: + np_boxes_list.append(value) + i=i+1 + + post_process = PicoDetPostProcess( + (640,640), + self.feed_dict['im_shape'], + self.feed_dict['scale_factor'], + [8, 16, 32, 64], + 0.5) + np_boxes, np_boxes_num = post_process(np_score_list, np_boxes_list) + res_dict = {} + d = [] + for b in range(np_boxes.shape[0]): + c = {} + #print(b) + c["category_id"] = np_boxes[b][0] + c["bbox"] = [np_boxes[b][2],np_boxes[b][3],np_boxes[b][4],np_boxes[b][5]] + c["score"] = np_boxes[b][1] + d.append(c) + res_dict["bbox_result"] = str(d) + #fetch_dict["image"] = "234.png" + #res_dict = {"bbox_result": str(self.img_postprocess(fetch_dict, visualize=False))} + return res_dict, None, "" + + +class PPYoloMbv(WebService): + def get_pipeline_response(self, read_op): + ppyolo_mbv3_op = PPYoloMbvOp(name="ppyolo_mbv3", input_ops=[read_op]) + return ppyolo_mbv3_op + + +ppyolo_mbv3_service = PPYoloMbv(name="ppyolo_mbv3") +ppyolo_mbv3_service.prepare_pipeline_config("config.yml") +ppyolo_mbv3_service.run_service() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__init__.py new file mode 100644 index 000000000..6fcc982fb --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__init__.py @@ -0,0 +1,26 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import (core, data, engine, modeling, model_zoo, optimizer, metrics, + utils, slim) + + +try: + from .version import full_version as __version__ + from .version import commit as __git_commit__ +except ImportError: + import sys + sys.stderr.write("Warning: import ppdet from source directory " \ + "without installing, run 'python setup.py install' to " \ + "install ppdet firstly\n") diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..2798aa67c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/optimizer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/optimizer.cpython-37.pyc new file mode 100644 index 000000000..f313e3454 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/__pycache__/optimizer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__init__.py new file mode 100644 index 000000000..d04277177 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__init__.py @@ -0,0 +1,15 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import config diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..e9d9642f5 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/workspace.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/workspace.cpython-37.pyc new file mode 100644 index 000000000..9b731d98d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/__pycache__/workspace.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__init__.py new file mode 100644 index 000000000..d0c32e260 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__init__.py @@ -0,0 +1,13 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..c6c49d37f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/schema.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/schema.cpython-37.pyc new file mode 100644 index 000000000..b29e3e2bb Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/schema.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/yaml_helpers.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/yaml_helpers.cpython-37.pyc new file mode 100644 index 000000000..d92cfca35 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/__pycache__/yaml_helpers.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/schema.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/schema.py new file mode 100644 index 000000000..2e41b5c34 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/schema.py @@ -0,0 +1,248 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import print_function +from __future__ import division + +import inspect +import importlib +import re + +try: + from docstring_parser import parse as doc_parse +except Exception: + + def doc_parse(*args): + pass + + +try: + from typeguard import check_type +except Exception: + + def check_type(*args): + pass + + +__all__ = ['SchemaValue', 'SchemaDict', 'SharedConfig', 'extract_schema'] + + +class SchemaValue(object): + def __init__(self, name, doc='', type=None): + super(SchemaValue, self).__init__() + self.name = name + self.doc = doc + self.type = type + + def set_default(self, value): + self.default = value + + def has_default(self): + return hasattr(self, 'default') + + +class SchemaDict(dict): + def __init__(self, **kwargs): + super(SchemaDict, self).__init__() + self.schema = {} + self.strict = False + self.doc = "" + self.update(kwargs) + + def __setitem__(self, key, value): + # XXX also update regular dict to SchemaDict?? + if isinstance(value, dict) and key in self and isinstance(self[key], + SchemaDict): + self[key].update(value) + else: + super(SchemaDict, self).__setitem__(key, value) + + def __missing__(self, key): + if self.has_default(key): + return self.schema[key].default + elif key in self.schema: + return self.schema[key] + else: + raise KeyError(key) + + def copy(self): + newone = SchemaDict() + newone.__dict__.update(self.__dict__) + newone.update(self) + return newone + + def set_schema(self, key, value): + assert isinstance(value, SchemaValue) + self.schema[key] = value + + def set_strict(self, strict): + self.strict = strict + + def has_default(self, key): + return key in self.schema and self.schema[key].has_default() + + def is_default(self, key): + if not self.has_default(key): + return False + if hasattr(self[key], '__dict__'): + return True + else: + return key not in self or self[key] == self.schema[key].default + + def find_default_keys(self): + return [ + k for k in list(self.keys()) + list(self.schema.keys()) + if self.is_default(k) + ] + + def mandatory(self): + return any([k for k in self.schema.keys() if not self.has_default(k)]) + + def find_missing_keys(self): + missing = [ + k for k in self.schema.keys() + if k not in self and not self.has_default(k) + ] + placeholders = [k for k in self if self[k] in ('', '')] + return missing + placeholders + + def find_extra_keys(self): + return list(set(self.keys()) - set(self.schema.keys())) + + def find_mismatch_keys(self): + mismatch_keys = [] + for arg in self.schema.values(): + if arg.type is not None: + try: + check_type("{}.{}".format(self.name, arg.name), + self[arg.name], arg.type) + except Exception: + mismatch_keys.append(arg.name) + return mismatch_keys + + def validate(self): + missing_keys = self.find_missing_keys() + if missing_keys: + raise ValueError("Missing param for class<{}>: {}".format( + self.name, ", ".join(missing_keys))) + extra_keys = self.find_extra_keys() + if extra_keys and self.strict: + raise ValueError("Extraneous param for class<{}>: {}".format( + self.name, ", ".join(extra_keys))) + mismatch_keys = self.find_mismatch_keys() + if mismatch_keys: + raise TypeError("Wrong param type for class<{}>: {}".format( + self.name, ", ".join(mismatch_keys))) + + +class SharedConfig(object): + """ + Representation class for `__shared__` annotations, which work as follows: + + - if `key` is set for the module in config file, its value will take + precedence + - if `key` is not set for the module but present in the config file, its + value will be used + - otherwise, use the provided `default_value` as fallback + + Args: + key: config[key] will be injected + default_value: fallback value + """ + + def __init__(self, key, default_value=None): + super(SharedConfig, self).__init__() + self.key = key + self.default_value = default_value + + +def extract_schema(cls): + """ + Extract schema from a given class + + Args: + cls (type): Class from which to extract. + + Returns: + schema (SchemaDict): Extracted schema. + """ + ctor = cls.__init__ + # python 2 compatibility + if hasattr(inspect, 'getfullargspec'): + argspec = inspect.getfullargspec(ctor) + annotations = argspec.annotations + has_kwargs = argspec.varkw is not None + else: + argspec = inspect.getfullargspec(ctor) + # python 2 type hinting workaround, see pep-3107 + # however, since `typeguard` does not support python 2, type checking + # is still python 3 only for now + annotations = getattr(ctor, '__annotations__', {}) + has_kwargs = argspec.varkw is not None + + names = [arg for arg in argspec.args if arg != 'self'] + defaults = argspec.defaults + num_defaults = argspec.defaults is not None and len(argspec.defaults) or 0 + num_required = len(names) - num_defaults + + docs = cls.__doc__ + if docs is None and getattr(cls, '__category__', None) == 'op': + docs = cls.__call__.__doc__ + try: + docstring = doc_parse(docs) + except Exception: + docstring = None + + if docstring is None: + comments = {} + else: + comments = {} + for p in docstring.params: + match_obj = re.match('^([a-zA-Z_]+[a-zA-Z_0-9]*).*', p.arg_name) + if match_obj is not None: + comments[match_obj.group(1)] = p.description + + schema = SchemaDict() + schema.name = cls.__name__ + schema.doc = "" + if docs is not None: + start_pos = docs[0] == '\n' and 1 or 0 + schema.doc = docs[start_pos:].split("\n")[0].strip() + # XXX handle paddle's weird doc convention + if '**' == schema.doc[:2] and '**' == schema.doc[-2:]: + schema.doc = schema.doc[2:-2].strip() + schema.category = hasattr(cls, '__category__') and getattr( + cls, '__category__') or 'module' + schema.strict = not has_kwargs + schema.pymodule = importlib.import_module(cls.__module__) + schema.inject = getattr(cls, '__inject__', []) + schema.shared = getattr(cls, '__shared__', []) + for idx, name in enumerate(names): + comment = name in comments and comments[name] or name + if name in schema.inject: + type_ = None + else: + type_ = name in annotations and annotations[name] or None + value_schema = SchemaValue(name, comment, type_) + if name in schema.shared: + assert idx >= num_required, "shared config must have default value" + default = defaults[idx - num_required] + value_schema.set_default(SharedConfig(name, default)) + elif idx >= num_required: + default = defaults[idx - num_required] + value_schema.set_default(default) + schema.set_schema(name, value_schema) + + return schema diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/yaml_helpers.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/yaml_helpers.py new file mode 100644 index 000000000..181cfe6fc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/config/yaml_helpers.py @@ -0,0 +1,118 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import importlib +import inspect + +import yaml +from .schema import SharedConfig + +__all__ = ['serializable', 'Callable'] + + +def represent_dictionary_order(self, dict_data): + return self.represent_mapping('tag:yaml.org,2002:map', dict_data.items()) + + +def setup_orderdict(): + from collections import OrderedDict + yaml.add_representer(OrderedDict, represent_dictionary_order) + + +def _make_python_constructor(cls): + def python_constructor(loader, node): + if isinstance(node, yaml.SequenceNode): + args = loader.construct_sequence(node, deep=True) + return cls(*args) + else: + kwargs = loader.construct_mapping(node, deep=True) + try: + return cls(**kwargs) + except Exception as ex: + print("Error when construct {} instance from yaml config". + format(cls.__name__)) + raise ex + + return python_constructor + + +def _make_python_representer(cls): + # python 2 compatibility + if hasattr(inspect, 'getfullargspec'): + argspec = inspect.getfullargspec(cls) + else: + argspec = inspect.getfullargspec(cls.__init__) + argnames = [arg for arg in argspec.args if arg != 'self'] + + def python_representer(dumper, obj): + if argnames: + data = {name: getattr(obj, name) for name in argnames} + else: + data = obj.__dict__ + if '_id' in data: + del data['_id'] + return dumper.represent_mapping(u'!{}'.format(cls.__name__), data) + + return python_representer + + +def serializable(cls): + """ + Add loader and dumper for given class, which must be + "trivially serializable" + + Args: + cls: class to be serialized + + Returns: cls + """ + yaml.add_constructor(u'!{}'.format(cls.__name__), + _make_python_constructor(cls)) + yaml.add_representer(cls, _make_python_representer(cls)) + return cls + + +yaml.add_representer(SharedConfig, + lambda d, o: d.represent_data(o.default_value)) + + +@serializable +class Callable(object): + """ + Helper to be used in Yaml for creating arbitrary class objects + + Args: + full_type (str): the full module path to target function + """ + + def __init__(self, full_type, args=[], kwargs={}): + super(Callable, self).__init__() + self.full_type = full_type + self.args = args + self.kwargs = kwargs + + def __call__(self): + if '.' in self.full_type: + idx = self.full_type.rfind('.') + module = importlib.import_module(self.full_type[:idx]) + func_name = self.full_type[idx + 1:] + else: + try: + module = importlib.import_module('builtins') + except Exception: + module = importlib.import_module('__builtin__') + func_name = self.full_type + + func = getattr(module, func_name) + return func(*self.args, **self.kwargs) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/workspace.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/workspace.py new file mode 100644 index 000000000..e633746ed --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/core/workspace.py @@ -0,0 +1,275 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import print_function +from __future__ import division + +import importlib +import os +import sys + +import yaml +import collections + +try: + collectionsAbc = collections.abc +except AttributeError: + collectionsAbc = collections + +from .config.schema import SchemaDict, SharedConfig, extract_schema +from .config.yaml_helpers import serializable + +__all__ = [ + 'global_config', + 'load_config', + 'merge_config', + 'get_registered_modules', + 'create', + 'register', + 'serializable', + 'dump_value', +] + + +def dump_value(value): + # XXX this is hackish, but collections.abc is not available in python 2 + if hasattr(value, '__dict__') or isinstance(value, (dict, tuple, list)): + value = yaml.dump(value, default_flow_style=True) + value = value.replace('\n', '') + value = value.replace('...', '') + return "'{}'".format(value) + else: + # primitive types + return str(value) + + +class AttrDict(dict): + """Single level attribute dict, NOT recursive""" + + def __init__(self, **kwargs): + super(AttrDict, self).__init__() + super(AttrDict, self).update(kwargs) + + def __getattr__(self, key): + if key in self: + return self[key] + raise AttributeError("object has no attribute '{}'".format(key)) + + +global_config = AttrDict() + +BASE_KEY = '_BASE_' + + +# parse and load _BASE_ recursively +def _load_config_with_base(file_path): + with open(file_path) as f: + file_cfg = yaml.load(f, Loader=yaml.Loader) + + # NOTE: cfgs outside have higher priority than cfgs in _BASE_ + if BASE_KEY in file_cfg: + all_base_cfg = AttrDict() + base_ymls = list(file_cfg[BASE_KEY]) + for base_yml in base_ymls: + if base_yml.startswith("~"): + base_yml = os.path.expanduser(base_yml) + if not base_yml.startswith('/'): + base_yml = os.path.join(os.path.dirname(file_path), base_yml) + + with open(base_yml) as f: + base_cfg = _load_config_with_base(base_yml) + all_base_cfg = merge_config(base_cfg, all_base_cfg) + + del file_cfg[BASE_KEY] + return merge_config(file_cfg, all_base_cfg) + + return file_cfg + + +def load_config(file_path): + """ + Load config from file. + + Args: + file_path (str): Path of the config file to be loaded. + + Returns: global config + """ + _, ext = os.path.splitext(file_path) + assert ext in ['.yml', '.yaml'], "only support yaml files for now" + + # load config from file and merge into global config + cfg = _load_config_with_base(file_path) + cfg['filename'] = os.path.splitext(os.path.split(file_path)[-1])[0] + merge_config(cfg) + + return global_config + + +def dict_merge(dct, merge_dct): + """ Recursive dict merge. Inspired by :meth:``dict.update()``, instead of + updating only top-level keys, dict_merge recurses down into dicts nested + to an arbitrary depth, updating keys. The ``merge_dct`` is merged into + ``dct``. + + Args: + dct: dict onto which the merge is executed + merge_dct: dct merged into dct + + Returns: dct + """ + for k, v in merge_dct.items(): + if (k in dct and isinstance(dct[k], dict) and + isinstance(merge_dct[k], collectionsAbc.Mapping)): + dict_merge(dct[k], merge_dct[k]) + else: + dct[k] = merge_dct[k] + return dct + + +def merge_config(config, another_cfg=None): + """ + Merge config into global config or another_cfg. + + Args: + config (dict): Config to be merged. + + Returns: global config + """ + global global_config + dct = another_cfg or global_config + return dict_merge(dct, config) + + +def get_registered_modules(): + return {k: v for k, v in global_config.items() if isinstance(v, SchemaDict)} + + +def make_partial(cls): + op_module = importlib.import_module(cls.__op__.__module__) + op = getattr(op_module, cls.__op__.__name__) + cls.__category__ = getattr(cls, '__category__', None) or 'op' + + def partial_apply(self, *args, **kwargs): + kwargs_ = self.__dict__.copy() + kwargs_.update(kwargs) + return op(*args, **kwargs_) + + if getattr(cls, '__append_doc__', True): # XXX should default to True? + if sys.version_info[0] > 2: + cls.__doc__ = "Wrapper for `{}` OP".format(op.__name__) + cls.__init__.__doc__ = op.__doc__ + cls.__call__ = partial_apply + cls.__call__.__doc__ = op.__doc__ + else: + # XXX work around for python 2 + partial_apply.__doc__ = op.__doc__ + cls.__call__ = partial_apply + return cls + + +def register(cls): + """ + Register a given module class. + + Args: + cls (type): Module class to be registered. + + Returns: cls + """ + if cls.__name__ in global_config: + raise ValueError("Module class already registered: {}".format( + cls.__name__)) + if hasattr(cls, '__op__'): + cls = make_partial(cls) + global_config[cls.__name__] = extract_schema(cls) + return cls + + +def create(cls_or_name, **kwargs): + """ + Create an instance of given module class. + + Args: + cls_or_name (type or str): Class of which to create instance. + + Returns: instance of type `cls_or_name` + """ + assert type(cls_or_name) in [type, str + ], "should be a class or name of a class" + name = type(cls_or_name) == str and cls_or_name or cls_or_name.__name__ + assert name in global_config and \ + isinstance(global_config[name], SchemaDict), \ + "the module {} is not registered".format(name) + config = global_config[name] + cls = getattr(config.pymodule, name) + cls_kwargs = {} + cls_kwargs.update(global_config[name]) + + # parse `shared` annoation of registered modules + if getattr(config, 'shared', None): + for k in config.shared: + target_key = config[k] + shared_conf = config.schema[k].default + assert isinstance(shared_conf, SharedConfig) + if target_key is not None and not isinstance(target_key, + SharedConfig): + continue # value is given for the module + elif shared_conf.key in global_config: + # `key` is present in config + cls_kwargs[k] = global_config[shared_conf.key] + else: + cls_kwargs[k] = shared_conf.default_value + + # parse `inject` annoation of registered modules + if getattr(cls, 'from_config', None): + cls_kwargs.update(cls.from_config(config, **kwargs)) + + if getattr(config, 'inject', None): + for k in config.inject: + target_key = config[k] + # optional dependency + if target_key is None: + continue + + if isinstance(target_key, dict) or hasattr(target_key, '__dict__'): + if 'name' not in target_key.keys(): + continue + inject_name = str(target_key['name']) + if inject_name not in global_config: + raise ValueError( + "Missing injection name {} and check it's name in cfg file". + format(k)) + target = global_config[inject_name] + for i, v in target_key.items(): + if i == 'name': + continue + target[i] = v + if isinstance(target, SchemaDict): + cls_kwargs[k] = create(inject_name) + elif isinstance(target_key, str): + if target_key not in global_config: + raise ValueError("Missing injection config:", target_key) + target = global_config[target_key] + if isinstance(target, SchemaDict): + cls_kwargs[k] = create(target_key) + elif hasattr(target, '__dict__'): # serialized object + cls_kwargs[k] = target + else: + raise ValueError("Unsupported injection type:", target_key) + # prevent modification of global config values of reference types + # (e.g., list, dict) from within the created module instances + #kwargs = copy.deepcopy(kwargs) + return cls(**cls_kwargs) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__init__.py new file mode 100644 index 000000000..a12aa323e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__init__.py @@ -0,0 +1,21 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import source +from . import transform +from . import reader + +from .source import * +from .transform import * +from .reader import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..b366ebbaa Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/reader.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/reader.cpython-37.pyc new file mode 100644 index 000000000..c7c369faf Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/reader.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/shm_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/shm_utils.cpython-37.pyc new file mode 100644 index 000000000..63a13890f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/__pycache__/shm_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__init__.py new file mode 100644 index 000000000..61d5aa213 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__init__.py @@ -0,0 +1,13 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..8c07ff43d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/annotation_cropper.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/annotation_cropper.cpython-37.pyc new file mode 100644 index 000000000..31df2fcf5 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/annotation_cropper.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/chip_box_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/chip_box_utils.cpython-37.pyc new file mode 100644 index 000000000..f24cf29d4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/__pycache__/chip_box_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/annotation_cropper.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/annotation_cropper.py new file mode 100644 index 000000000..93a9a1f75 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/annotation_cropper.py @@ -0,0 +1,542 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import copy +import math +import random +import numpy as np +from copy import deepcopy +from typing import List, Tuple +from collections import defaultdict + +from .chip_box_utils import nms, transform_chip_boxes2image_boxes +from .chip_box_utils import find_chips_to_cover_overlaped_boxes +from .chip_box_utils import transform_chip_box +from .chip_box_utils import intersection_over_box + + +class AnnoCropper(object): + def __init__(self, image_target_sizes: List[int], + valid_box_ratio_ranges: List[List[float]], + chip_target_size: int, chip_target_stride: int, + use_neg_chip: bool = False, + max_neg_num_per_im: int = 8, + max_per_img: int = -1, + nms_thresh: int = 0.5 + ): + """ + Generate chips by chip_target_size and chip_target_stride. + These two parameters just like kernel_size and stride in cnn. + + Each image has its raw size. After resizing, then get its target size. + The resizing scale = target_size / raw_size. + So are chips of the image. + box_ratio = box_raw_size / image_raw_size = box_target_size / image_target_size + The 'size' above mentioned is the size of long-side of image, box or chip. + + :param image_target_sizes: [2000, 1000] + :param valid_box_ratio_ranges: [[-1, 0.1],[0.08, -1]] + :param chip_target_size: 500 + :param chip_target_stride: 200 + """ + self.target_sizes = image_target_sizes + self.valid_box_ratio_ranges = valid_box_ratio_ranges + assert len(self.target_sizes) == len(self.valid_box_ratio_ranges) + self.scale_num = len(self.target_sizes) + self.chip_target_size = chip_target_size # is target size + self.chip_target_stride = chip_target_stride # is target stride + self.use_neg_chip = use_neg_chip + self.max_neg_num_per_im = max_neg_num_per_im + self.max_per_img = max_per_img + self.nms_thresh = nms_thresh + + def crop_anno_records(self, records: List[dict]): + """ + The main logic: + # foreach record(image): + # foreach scale: + # 1 generate chips by chip size and stride for each scale + # 2 get pos chips + # - validate boxes: current scale; h,w >= 1 + # - find pos chips greedily by valid gt boxes in each scale + # - for every valid gt box, find its corresponding pos chips in each scale + # 3 get neg chips + # - If given proposals, find neg boxes in them which are not in pos chips + # - If got neg boxes in last step, we find neg chips and assign neg boxes to neg chips such as 2. + # 4 sample neg chips if too much each image + # transform this image-scale annotations to chips(pos chips&neg chips) annotations + + :param records, standard coco_record but with extra key `proposals`(Px4), which are predicted by stage1 + model and maybe have neg boxes in them. + :return: new_records, list of dict like + { + 'im_file': 'fake_image1.jpg', + 'im_id': np.array([1]), # new _global_chip_id as im_id + 'h': h, # chip height + 'w': w, # chip width + 'is_crowd': is_crowd, # Nx1 -> Mx1 + 'gt_class': gt_class, # Nx1 -> Mx1 + 'gt_bbox': gt_bbox, # Nx4 -> Mx4, 4 represents [x1,y1,x2,y2] + 'gt_poly': gt_poly, # [None]xN -> [None]xM + 'chip': [x1, y1, x2, y2] # added + } + + Attention: + ------------------------------>x + | + | (x1,y1)------ + | | | + | | | + | | | + | | | + | | | + | ---------- + | (x2,y2) + | + 鈫 + y + + If we use [x1, y1, x2, y2] to represent boxes or chips, + (x1,y1) is the left-top point which is in the box, + but (x2,y2) is the right-bottom point which is not in the box. + So x1 in [0, w-1], x2 in [1, w], y1 in [0, h-1], y2 in [1,h]. + And you can use x2-x1 to get width, and you can use image[y1:y2, x1:x2] to get the box area. + """ + + self.chip_records = [] + self._global_chip_id = 1 + for r in records: + self._cur_im_pos_chips = [] # element: (chip, boxes_idx), chip is [x1, y1, x2, y2], boxes_ids is List[int] + self._cur_im_neg_chips = [] # element: (chip, neg_box_num) + for scale_i in range(self.scale_num): + self._get_current_scale_parameters(scale_i, r) + + # Cx4 + chips = self._create_chips(r['h'], r['w'], self._cur_scale) + + # # dict: chipid->[box_id, ...] + pos_chip2boxes_idx = self._get_valid_boxes_and_pos_chips(r['gt_bbox'], chips) + + # dict: chipid->neg_box_num + neg_chip2box_num = self._get_neg_boxes_and_chips(chips, list(pos_chip2boxes_idx.keys()), r.get('proposals', None)) + + self._add_to_cur_im_chips(chips, pos_chip2boxes_idx, neg_chip2box_num) + + cur_image_records = self._trans_all_chips2annotations(r) + self.chip_records.extend(cur_image_records) + return self.chip_records + + def _add_to_cur_im_chips(self, chips, pos_chip2boxes_idx, neg_chip2box_num): + for pos_chipid, boxes_idx in pos_chip2boxes_idx.items(): + chip = np.array(chips[pos_chipid]) # copy chips slice + self._cur_im_pos_chips.append((chip, boxes_idx)) + + if neg_chip2box_num is None: + return + + for neg_chipid, neg_box_num in neg_chip2box_num.items(): + chip = np.array(chips[neg_chipid]) + self._cur_im_neg_chips.append((chip, neg_box_num)) + + def _trans_all_chips2annotations(self, r): + gt_bbox = r['gt_bbox'] + im_file = r['im_file'] + is_crowd = r['is_crowd'] + gt_class = r['gt_class'] + # gt_poly = r['gt_poly'] # [None]xN + # remaining keys: im_id, h, w + chip_records = self._trans_pos_chips2annotations(im_file, gt_bbox, is_crowd, gt_class) + + if not self.use_neg_chip: + return chip_records + + sampled_neg_chips = self._sample_neg_chips() + neg_chip_records = self._trans_neg_chips2annotations(im_file, sampled_neg_chips) + chip_records.extend(neg_chip_records) + return chip_records + + def _trans_pos_chips2annotations(self, im_file, gt_bbox, is_crowd, gt_class): + chip_records = [] + for chip, boxes_idx in self._cur_im_pos_chips: + chip_bbox, final_boxes_idx = transform_chip_box(gt_bbox, boxes_idx, chip) + x1, y1, x2, y2 = chip + chip_h = y2 - y1 + chip_w = x2 - x1 + rec = { + 'im_file': im_file, + 'im_id': np.array([self._global_chip_id]), + 'h': chip_h, + 'w': chip_w, + 'gt_bbox': chip_bbox, + 'is_crowd': is_crowd[final_boxes_idx].copy(), + 'gt_class': gt_class[final_boxes_idx].copy(), + # 'gt_poly': [None] * len(final_boxes_idx), + 'chip': chip + } + self._global_chip_id += 1 + chip_records.append(rec) + return chip_records + + def _sample_neg_chips(self): + pos_num = len(self._cur_im_pos_chips) + neg_num = len(self._cur_im_neg_chips) + sample_num = min(pos_num + 2, self.max_neg_num_per_im) + assert sample_num >= 1 + if neg_num <= sample_num: + return self._cur_im_neg_chips + + candidate_num = int(sample_num * 1.5) + candidate_neg_chips = sorted(self._cur_im_neg_chips, key=lambda x: -x[1])[:candidate_num] + random.shuffle(candidate_neg_chips) + sampled_neg_chips = candidate_neg_chips[:sample_num] + return sampled_neg_chips + + def _trans_neg_chips2annotations(self, im_file: str, sampled_neg_chips: List[Tuple]): + chip_records = [] + for chip, neg_box_num in sampled_neg_chips: + x1, y1, x2, y2 = chip + chip_h = y2 - y1 + chip_w = x2 - x1 + rec = { + 'im_file': im_file, + 'im_id': np.array([self._global_chip_id]), + 'h': chip_h, + 'w': chip_w, + 'gt_bbox': np.zeros((0, 4), dtype=np.float32), + 'is_crowd': np.zeros((0, 1), dtype=np.int32), + 'gt_class': np.zeros((0, 1), dtype=np.int32), + # 'gt_poly': [], + 'chip': chip + } + self._global_chip_id += 1 + chip_records.append(rec) + return chip_records + + def _get_current_scale_parameters(self, scale_i, r): + im_size = max(r['h'], r['w']) + im_target_size = self.target_sizes[scale_i] + self._cur_im_size, self._cur_im_target_size = im_size, im_target_size + self._cur_scale = self._get_current_scale(im_target_size, im_size) + self._cur_valid_ratio_range = self.valid_box_ratio_ranges[scale_i] + + def _get_current_scale(self, im_target_size, im_size): + return im_target_size / im_size + + def _create_chips(self, h: int, w: int, scale: float): + """ + Generate chips by chip_target_size and chip_target_stride. + These two parameters just like kernel_size and stride in cnn. + :return: chips, Cx4, xy in raw size dimension + """ + chip_size = self.chip_target_size # omit target for simplicity + stride = self.chip_target_stride + width = int(scale * w) + height = int(scale * h) + min_chip_location_diff = 20 # in target size + + assert chip_size >= stride + chip_overlap = chip_size - stride + if (width - chip_overlap) % stride > min_chip_location_diff: # 涓嶈兘琚玸tride鏁撮櫎鐨勯儴鍒嗘瘮杈冨ぇ锛屽垯淇濈暀 + w_steps = max(1, int(math.ceil((width - chip_overlap) / stride))) + else: # 涓嶈兘琚玸tride鏁撮櫎鐨勯儴鍒嗘瘮杈冨皬锛屽垯涓㈠純 + w_steps = max(1, int(math.floor((width - chip_overlap) / stride))) + if (height - chip_overlap) % stride > min_chip_location_diff: + h_steps = max(1, int(math.ceil((height - chip_overlap) / stride))) + else: + h_steps = max(1, int(math.floor((height - chip_overlap) / stride))) + + chips = list() + for j in range(h_steps): + for i in range(w_steps): + x1 = i * stride + y1 = j * stride + x2 = min(x1 + chip_size, width) + y2 = min(y1 + chip_size, height) + chips.append([x1, y1, x2, y2]) + + # check chip size + for item in chips: + if item[2] - item[0] > chip_size * 1.1 or item[3] - item[1] > chip_size * 1.1: + raise ValueError(item) + chips = np.array(chips, dtype=np.float) + + raw_size_chips = chips / scale + return raw_size_chips + + def _get_valid_boxes_and_pos_chips(self, gt_bbox, chips): + valid_ratio_range = self._cur_valid_ratio_range + im_size = self._cur_im_size + scale = self._cur_scale + # Nx4 N + valid_boxes, valid_boxes_idx = self._validate_boxes(valid_ratio_range, im_size, gt_bbox, scale) + # dict: chipid->[box_id, ...] + pos_chip2boxes_idx = self._find_pos_chips(chips, valid_boxes, valid_boxes_idx) + return pos_chip2boxes_idx + + def _validate_boxes(self, valid_ratio_range: List[float], + im_size: int, + gt_boxes: 'np.array of Nx4', + scale: float): + """ + :return: valid_boxes: Nx4, valid_boxes_idx: N + """ + ws = (gt_boxes[:, 2] - gt_boxes[:, 0]).astype(np.int32) + hs = (gt_boxes[:, 3] - gt_boxes[:, 1]).astype(np.int32) + maxs = np.maximum(ws, hs) + box_ratio = maxs / im_size + mins = np.minimum(ws, hs) + target_mins = mins * scale + + low = valid_ratio_range[0] if valid_ratio_range[0] > 0 else 0 + high = valid_ratio_range[1] if valid_ratio_range[1] > 0 else np.finfo(np.float).max + + valid_boxes_idx = np.nonzero((low <= box_ratio) & (box_ratio < high) & (target_mins >= 2))[0] + valid_boxes = gt_boxes[valid_boxes_idx] + return valid_boxes, valid_boxes_idx + + def _find_pos_chips(self, chips: 'Cx4', valid_boxes: 'Bx4', valid_boxes_idx: 'B'): + """ + :return: pos_chip2boxes_idx, dict: chipid->[box_id, ...] + """ + iob = intersection_over_box(chips, valid_boxes) # overlap, CxB + + iob_threshold_to_find_chips = 1. + pos_chip_ids, _ = self._find_chips_to_cover_overlaped_boxes(iob, iob_threshold_to_find_chips) + pos_chip_ids = set(pos_chip_ids) + + iob_threshold_to_assign_box = 0.5 + pos_chip2boxes_idx = self._assign_boxes_to_pos_chips( + iob, iob_threshold_to_assign_box, pos_chip_ids, valid_boxes_idx) + return pos_chip2boxes_idx + + def _find_chips_to_cover_overlaped_boxes(self, iob, overlap_threshold): + return find_chips_to_cover_overlaped_boxes(iob, overlap_threshold) + + def _assign_boxes_to_pos_chips(self, iob, overlap_threshold, pos_chip_ids, valid_boxes_idx): + chip_ids, box_ids = np.nonzero(iob >= overlap_threshold) + pos_chip2boxes_idx = defaultdict(list) + for chip_id, box_id in zip(chip_ids, box_ids): + if chip_id not in pos_chip_ids: + continue + raw_gt_box_idx = valid_boxes_idx[box_id] + pos_chip2boxes_idx[chip_id].append(raw_gt_box_idx) + return pos_chip2boxes_idx + + def _get_neg_boxes_and_chips(self, chips: 'Cx4', pos_chip_ids: 'D', proposals: 'Px4'): + """ + :param chips: + :param pos_chip_ids: + :param proposals: + :return: neg_chip2box_num, None or dict: chipid->neg_box_num + """ + if not self.use_neg_chip: + return None + + # train proposals maybe None + if proposals is None or len(proposals) < 1: + return None + + valid_ratio_range = self._cur_valid_ratio_range + im_size = self._cur_im_size + scale = self._cur_scale + + valid_props, _ = self._validate_boxes(valid_ratio_range, im_size, proposals, scale) + neg_boxes = self._find_neg_boxes(chips, pos_chip_ids, valid_props) + neg_chip2box_num = self._find_neg_chips(chips, pos_chip_ids, neg_boxes) + return neg_chip2box_num + + def _find_neg_boxes(self, chips: 'Cx4', pos_chip_ids: 'D', valid_props: 'Px4'): + """ + :return: neg_boxes: Nx4 + """ + if len(pos_chip_ids) == 0: + return valid_props + + pos_chips = chips[pos_chip_ids] + iob = intersection_over_box(pos_chips, valid_props) + overlap_per_prop = np.max(iob, axis=0) + non_overlap_props_idx = overlap_per_prop < 0.5 + neg_boxes = valid_props[non_overlap_props_idx] + return neg_boxes + + def _find_neg_chips(self, chips: 'Cx4', pos_chip_ids: 'D', neg_boxes: 'Nx4'): + """ + :return: neg_chip2box_num, dict: chipid->neg_box_num + """ + neg_chip_ids = np.setdiff1d(np.arange(len(chips)), pos_chip_ids) + neg_chips = chips[neg_chip_ids] + + iob = intersection_over_box(neg_chips, neg_boxes) + iob_threshold_to_find_chips = 0.7 + chosen_neg_chip_ids, chip_id2overlap_box_num = \ + self._find_chips_to_cover_overlaped_boxes(iob, iob_threshold_to_find_chips) + + neg_chipid2box_num = {} + for cid in chosen_neg_chip_ids: + box_num = chip_id2overlap_box_num[cid] + raw_chip_id = neg_chip_ids[cid] + neg_chipid2box_num[raw_chip_id] = box_num + return neg_chipid2box_num + + def crop_infer_anno_records(self, records: List[dict]): + """ + transform image record to chips record + :param records: + :return: new_records, list of dict like + { + 'im_file': 'fake_image1.jpg', + 'im_id': np.array([1]), # new _global_chip_id as im_id + 'h': h, # chip height + 'w': w, # chip width + 'chip': [x1, y1, x2, y2] # added + 'ori_im_h': ori_im_h # added, origin image height + 'ori_im_w': ori_im_w # added, origin image width + 'scale_i': 0 # added, + } + """ + self.chip_records = [] + self._global_chip_id = 1 # im_id start from 1 + self._global_chip_id2img_id = {} + + for r in records: + for scale_i in range(self.scale_num): + self._get_current_scale_parameters(scale_i, r) + # Cx4 + chips = self._create_chips(r['h'], r['w'], self._cur_scale) + cur_img_chip_record = self._get_chips_records(r, chips, scale_i) + self.chip_records.extend(cur_img_chip_record) + + return self.chip_records + + def _get_chips_records(self, rec, chips, scale_i): + cur_img_chip_records = [] + ori_im_h = rec["h"] + ori_im_w = rec["w"] + im_file = rec["im_file"] + ori_im_id = rec["im_id"] + for id, chip in enumerate(chips): + chip_rec = {} + x1, y1, x2, y2 = chip + chip_h = y2 - y1 + chip_w = x2 - x1 + chip_rec["im_file"] = im_file + chip_rec["im_id"] = self._global_chip_id + chip_rec["h"] = chip_h + chip_rec["w"] = chip_w + chip_rec["chip"] = chip + chip_rec["ori_im_h"] = ori_im_h + chip_rec["ori_im_w"] = ori_im_w + chip_rec["scale_i"] = scale_i + + self._global_chip_id2img_id[self._global_chip_id] = int(ori_im_id) + self._global_chip_id += 1 + cur_img_chip_records.append(chip_rec) + + return cur_img_chip_records + + def aggregate_chips_detections(self, results, records=None): + """ + # 1. transform chip dets to image dets + # 2. nms boxes per image; + # 3. format output results + :param results: + :param roidb: + :return: + """ + results = deepcopy(results) + records = records if records else self.chip_records + img_id2bbox = self._transform_chip2image_bboxes(results, records) + nms_img_id2bbox = self._nms_dets(img_id2bbox) + aggregate_results = self._reformat_results(nms_img_id2bbox) + return aggregate_results + + def _transform_chip2image_bboxes(self, results, records): + # 1. Transform chip dets to image dets; + # 2. Filter valid range; + # 3. Reformat and Aggregate chip dets to Get scale_cls_dets + img_id2bbox = defaultdict(list) + for result in results: + bbox_locs = result['bbox'] + bbox_nums = result['bbox_num'] + if len(bbox_locs) == 1 and bbox_locs[0][0] == -1: # current batch has no detections + # bbox_locs = array([[-1.]], dtype=float32); bbox_nums = [[1]] + # MultiClassNMS output: If there is no detected boxes for all images, lod will be set to {1} and Out only contains one value which is -1. + continue + im_ids = result['im_id'] # replace with range(len(bbox_nums)) + + last_bbox_num = 0 + for idx, im_id in enumerate(im_ids): + + cur_bbox_len = bbox_nums[idx] + bboxes = bbox_locs[last_bbox_num: last_bbox_num + cur_bbox_len] + last_bbox_num += cur_bbox_len + # box: [num_id, score, xmin, ymin, xmax, ymax] + if len(bboxes) == 0: # current image has no detections + continue + + chip_rec = records[int(im_id) - 1] # im_id starts from 1, type is np.int64 + image_size = max(chip_rec["ori_im_h"], chip_rec["ori_im_w"]) + + bboxes = transform_chip_boxes2image_boxes(bboxes, chip_rec["chip"], chip_rec["ori_im_h"], chip_rec["ori_im_w"]) + + scale_i = chip_rec["scale_i"] + cur_scale = self._get_current_scale(self.target_sizes[scale_i], image_size) + _, valid_boxes_idx = self._validate_boxes(self.valid_box_ratio_ranges[scale_i], image_size, + bboxes[:, 2:], cur_scale) + ori_img_id = self._global_chip_id2img_id[int(im_id)] + + img_id2bbox[ori_img_id].append(bboxes[valid_boxes_idx]) + + return img_id2bbox + + def _nms_dets(self, img_id2bbox): + # 1. NMS on each image-class + # 2. Limit number of detections to MAX_PER_IMAGE if requested + max_per_img = self.max_per_img + nms_thresh = self.nms_thresh + + for img_id in img_id2bbox: + box = img_id2bbox[img_id] # list of np.array of shape [N, 6], 6 is [label, score, x1, y1, x2, y2] + box = np.concatenate(box, axis=0) + nms_dets = nms(box, nms_thresh) + if max_per_img > 0: + if len(nms_dets) > max_per_img: + keep = np.argsort(-nms_dets[:, 1])[:max_per_img] + nms_dets = nms_dets[keep] + + img_id2bbox[img_id] = nms_dets + + return img_id2bbox + + def _reformat_results(self, img_id2bbox): + """reformat results""" + im_ids = img_id2bbox.keys() + results = [] + for img_id in im_ids: # output by original im_id order + if len(img_id2bbox[img_id]) == 0: + bbox = np.array([[-1., 0., 0., 0., 0., 0.]]) # edge case: no detections + bbox_num = np.array([0]) + else: + # np.array of shape [N, 6], 6 is [label, score, x1, y1, x2, y2] + bbox = img_id2bbox[img_id] + bbox_num = np.array([len(bbox)]) + res = dict( + im_id=np.array([[img_id]]), + bbox=bbox, + bbox_num=bbox_num + ) + results.append(res) + return results + + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/chip_box_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/chip_box_utils.py new file mode 100644 index 000000000..d6e81a165 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/crop_utils/chip_box_utils.py @@ -0,0 +1,166 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np + + +def bbox_area(boxes): + return (boxes[:, 2] - boxes[:, 0]) * (boxes[:, 3] - boxes[:, 1]) + + +def intersection_over_box(chips, boxes): + """ + intersection area over box area + :param chips: C + :param boxes: B + :return: iob, CxB + """ + M = chips.shape[0] + N = boxes.shape[0] + if M * N == 0: + return np.zeros([M, N], dtype='float32') + + box_area = bbox_area(boxes) # B + + inter_x2y2 = np.minimum(np.expand_dims(chips, 1)[:, :, 2:], boxes[:, 2:]) # CxBX2 + inter_x1y1 = np.maximum(np.expand_dims(chips, 1)[:, :, :2], boxes[:, :2]) # CxBx2 + inter_wh = inter_x2y2 - inter_x1y1 + inter_wh = np.clip(inter_wh, a_min=0, a_max=None) + inter_area = inter_wh[:, :, 0] * inter_wh[:, :, 1] # CxB + + iob = inter_area / np.expand_dims(box_area, 0) + return iob + + +def clip_boxes(boxes, im_shape): + """ + Clip boxes to image boundaries. + :param boxes: [N, 4] + :param im_shape: tuple of 2, [h, w] + :return: [N, 4] + """ + # x1 >= 0 + boxes[:, 0] = np.clip(boxes[:, 0], 0, im_shape[1] - 1) + # y1 >= 0 + boxes[:, 1] = np.clip(boxes[:, 1], 0, im_shape[0] - 1) + # x2 < im_shape[1] + boxes[:, 2] = np.clip(boxes[:, 2], 1, im_shape[1]) + # y2 < im_shape[0] + boxes[:, 3] = np.clip(boxes[:, 3], 1, im_shape[0]) + return boxes + + +def transform_chip_box(gt_bbox: 'Gx4', boxes_idx: 'B', chip: '4'): + boxes_idx = np.array(boxes_idx) + cur_gt_bbox = gt_bbox[boxes_idx].copy() # Bx4 + x1, y1, x2, y2 = chip + cur_gt_bbox[:, 0] -= x1 + cur_gt_bbox[:, 1] -= y1 + cur_gt_bbox[:, 2] -= x1 + cur_gt_bbox[:, 3] -= y1 + h = y2 - y1 + w = x2 - x1 + cur_gt_bbox = clip_boxes(cur_gt_bbox, (h, w)) + ws = (cur_gt_bbox[:, 2] - cur_gt_bbox[:, 0]).astype(np.int32) + hs = (cur_gt_bbox[:, 3] - cur_gt_bbox[:, 1]).astype(np.int32) + valid_idx = (ws >= 2) & (hs >= 2) + return cur_gt_bbox[valid_idx], boxes_idx[valid_idx] + + +def find_chips_to_cover_overlaped_boxes(iob, overlap_threshold): + chip_ids, box_ids = np.nonzero(iob >= overlap_threshold) + chip_id2overlap_box_num = np.bincount(chip_ids) # 1d array + chip_id2overlap_box_num = np.pad(chip_id2overlap_box_num, (0, len(iob) - len(chip_id2overlap_box_num)), + constant_values=0) + + chosen_chip_ids = [] + while len(box_ids) > 0: + value_counts = np.bincount(chip_ids) # 1d array + max_count_chip_id = np.argmax(value_counts) + assert max_count_chip_id not in chosen_chip_ids + chosen_chip_ids.append(max_count_chip_id) + + box_ids_in_cur_chip = box_ids[chip_ids == max_count_chip_id] + ids_not_in_cur_boxes_mask = np.logical_not(np.isin(box_ids, box_ids_in_cur_chip)) + chip_ids = chip_ids[ids_not_in_cur_boxes_mask] + box_ids = box_ids[ids_not_in_cur_boxes_mask] + return chosen_chip_ids, chip_id2overlap_box_num + + +def transform_chip_boxes2image_boxes(chip_boxes, chip, img_h, img_w): + chip_boxes = np.array(sorted(chip_boxes, key=lambda item: -item[1])) + xmin, ymin, _, _ = chip + # Transform to origin image loc + chip_boxes[:, 2] += xmin + chip_boxes[:, 4] += xmin + chip_boxes[:, 3] += ymin + chip_boxes[:, 5] += ymin + chip_boxes = clip_boxes(chip_boxes, (img_h, img_w)) + return chip_boxes + + +def nms(dets, thresh): + """Apply classic DPM-style greedy NMS.""" + if dets.shape[0] == 0: + return dets[[], :] + scores = dets[:, 1] + x1 = dets[:, 2] + y1 = dets[:, 3] + x2 = dets[:, 4] + y2 = dets[:, 5] + + areas = (x2 - x1 + 1) * (y2 - y1 + 1) + order = scores.argsort()[::-1] + + ndets = dets.shape[0] + suppressed = np.zeros((ndets), dtype=np.int) + + # nominal indices + # _i, _j + # sorted indices + # i, j + # temp variables for box i's (the box currently under consideration) + # ix1, iy1, ix2, iy2, iarea + + # variables for computing overlap with box j (lower scoring box) + # xx1, yy1, xx2, yy2 + # w, h + # inter, ovr + + for _i in range(ndets): + i = order[_i] + if suppressed[i] == 1: + continue + ix1 = x1[i] + iy1 = y1[i] + ix2 = x2[i] + iy2 = y2[i] + iarea = areas[i] + for _j in range(_i + 1, ndets): + j = order[_j] + if suppressed[j] == 1: + continue + xx1 = max(ix1, x1[j]) + yy1 = max(iy1, y1[j]) + xx2 = min(ix2, x2[j]) + yy2 = min(iy2, y2[j]) + w = max(0.0, xx2 - xx1 + 1) + h = max(0.0, yy2 - yy1 + 1) + inter = w * h + ovr = inter / (iarea + areas[j] - inter) + if ovr >= thresh: + suppressed[j] = 1 + keep = np.where(suppressed == 0)[0] + dets = dets[keep, :] + return dets diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/reader.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/reader.py new file mode 100644 index 000000000..c9ea09af2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/reader.py @@ -0,0 +1,302 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import traceback +import six +import sys +if sys.version_info >= (3, 0): + pass +else: + pass +import numpy as np + +from paddle.io import DataLoader, DistributedBatchSampler +from paddle.fluid.dataloader.collate import default_collate_fn + +from ppdet.core.workspace import register +from . import transform +from .shm_utils import _get_shared_memory_size_in_M + +from ppdet.utils.logger import setup_logger +logger = setup_logger('reader') + +MAIN_PID = os.getpid() + + +class Compose(object): + def __init__(self, transforms, num_classes=80): + self.transforms = transforms + self.transforms_cls = [] + for t in self.transforms: + for k, v in t.items(): + op_cls = getattr(transform, k) + f = op_cls(**v) + if hasattr(f, 'num_classes'): + f.num_classes = num_classes + + self.transforms_cls.append(f) + + def __call__(self, data): + for f in self.transforms_cls: + try: + data = f(data) + except Exception as e: + stack_info = traceback.format_exc() + logger.warning("fail to map sample transform [{}] " + "with error: {} and stack:\n{}".format( + f, e, str(stack_info))) + raise e + + return data + + +class BatchCompose(Compose): + def __init__(self, transforms, num_classes=80, collate_batch=True): + super(BatchCompose, self).__init__(transforms, num_classes) + self.collate_batch = collate_batch + + def __call__(self, data): + for f in self.transforms_cls: + try: + data = f(data) + except Exception as e: + stack_info = traceback.format_exc() + logger.warning("fail to map batch transform [{}] " + "with error: {} and stack:\n{}".format( + f, e, str(stack_info))) + raise e + + # remove keys which is not needed by model + extra_key = ['h', 'w', 'flipped'] + for k in extra_key: + for sample in data: + if k in sample: + sample.pop(k) + + # batch data, if user-define batch function needed + # use user-defined here + if self.collate_batch: + batch_data = default_collate_fn(data) + else: + batch_data = {} + for k in data[0].keys(): + tmp_data = [] + for i in range(len(data)): + tmp_data.append(data[i][k]) + if not 'gt_' in k and not 'is_crowd' in k and not 'difficult' in k: + tmp_data = np.stack(tmp_data, axis=0) + batch_data[k] = tmp_data + return batch_data + + +class BaseDataLoader(object): + """ + Base DataLoader implementation for detection models + + Args: + sample_transforms (list): a list of transforms to perform + on each sample + batch_transforms (list): a list of transforms to perform + on batch + batch_size (int): batch size for batch collating, default 1. + shuffle (bool): whether to shuffle samples + drop_last (bool): whether to drop the last incomplete, + default False + num_classes (int): class number of dataset, default 80 + collate_batch (bool): whether to collate batch in dataloader. + If set to True, the samples will collate into batch according + to the batch size. Otherwise, the ground-truth will not collate, + which is used when the number of ground-truch is different in + samples. + use_shared_memory (bool): whether to use shared memory to + accelerate data loading, enable this only if you + are sure that the shared memory size of your OS + is larger than memory cost of input datas of model. + Note that shared memory will be automatically + disabled if the shared memory of OS is less than + 1G, which is not enough for detection models. + Default False. + """ + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=False, + drop_last=False, + num_classes=80, + collate_batch=True, + use_shared_memory=False, + **kwargs): + # sample transform + self._sample_transforms = Compose( + sample_transforms, num_classes=num_classes) + + # batch transfrom + self._batch_transforms = BatchCompose(batch_transforms, num_classes, + collate_batch) + self.batch_size = batch_size + self.shuffle = shuffle + self.drop_last = drop_last + self.use_shared_memory = use_shared_memory + self.kwargs = kwargs + + def __call__(self, + dataset, + worker_num, + batch_sampler=None, + return_list=False): + self.dataset = dataset + self.dataset.check_or_download_dataset() + self.dataset.parse_dataset() + # get data + self.dataset.set_transform(self._sample_transforms) + # set kwargs + self.dataset.set_kwargs(**self.kwargs) + # batch sampler + if batch_sampler is None: + self._batch_sampler = DistributedBatchSampler( + self.dataset, + batch_size=self.batch_size, + shuffle=self.shuffle, + drop_last=self.drop_last) + else: + self._batch_sampler = batch_sampler + + # DataLoader do not start sub-process in Windows and Mac + # system, do not need to use shared memory + use_shared_memory = self.use_shared_memory and \ + sys.platform not in ['win32', 'darwin'] + # check whether shared memory size is bigger than 1G(1024M) + if use_shared_memory: + shm_size = _get_shared_memory_size_in_M() + if shm_size is not None and shm_size < 1024.: + logger.warning("Shared memory size is less than 1G, " + "disable shared_memory in DataLoader") + use_shared_memory = False + + self.dataloader = DataLoader( + dataset=self.dataset, + batch_sampler=self._batch_sampler, + collate_fn=self._batch_transforms, + num_workers=worker_num, + return_list=return_list, + use_shared_memory=use_shared_memory) + self.loader = iter(self.dataloader) + + return self + + def __len__(self): + return len(self._batch_sampler) + + def __iter__(self): + return self + + def __next__(self): + try: + return next(self.loader) + except StopIteration: + self.loader = iter(self.dataloader) + six.reraise(*sys.exc_info()) + + def next(self): + # python2 compatibility + return self.__next__() + + +@register +class TrainReader(BaseDataLoader): + __shared__ = ['num_classes'] + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=True, + drop_last=True, + num_classes=80, + collate_batch=True, + **kwargs): + super(TrainReader, self).__init__(sample_transforms, batch_transforms, + batch_size, shuffle, drop_last, + num_classes, collate_batch, **kwargs) + + +@register +class EvalReader(BaseDataLoader): + __shared__ = ['num_classes'] + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=False, + drop_last=True, + num_classes=80, + **kwargs): + super(EvalReader, self).__init__(sample_transforms, batch_transforms, + batch_size, shuffle, drop_last, + num_classes, **kwargs) + + +@register +class TestReader(BaseDataLoader): + __shared__ = ['num_classes'] + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=False, + drop_last=False, + num_classes=80, + **kwargs): + super(TestReader, self).__init__(sample_transforms, batch_transforms, + batch_size, shuffle, drop_last, + num_classes, **kwargs) + + +@register +class EvalMOTReader(BaseDataLoader): + __shared__ = ['num_classes'] + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=False, + drop_last=False, + num_classes=1, + **kwargs): + super(EvalMOTReader, self).__init__(sample_transforms, batch_transforms, + batch_size, shuffle, drop_last, + num_classes, **kwargs) + + +@register +class TestMOTReader(BaseDataLoader): + __shared__ = ['num_classes'] + + def __init__(self, + sample_transforms=[], + batch_transforms=[], + batch_size=1, + shuffle=False, + drop_last=False, + num_classes=1, + **kwargs): + super(TestMOTReader, self).__init__(sample_transforms, batch_transforms, + batch_size, shuffle, drop_last, + num_classes, **kwargs) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/shm_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/shm_utils.py new file mode 100644 index 000000000..38d8ba66c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/shm_utils.py @@ -0,0 +1,67 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os + +SIZE_UNIT = ['K', 'M', 'G', 'T'] +SHM_QUERY_CMD = 'df -h' +SHM_KEY = 'shm' +SHM_DEFAULT_MOUNT = '/dev/shm' + +# [ shared memory size check ] +# In detection models, image/target data occupies a lot of memory, and +# will occupy lots of shared memory in multi-process DataLoader, we use +# following code to get shared memory size and perform a size check to +# disable shared memory use if shared memory size is not enough. +# Shared memory getting process as follows: +# 1. use `df -h` get all mount info +# 2. pick up spaces whose mount info contains 'shm' +# 3. if 'shm' space number is only 1, return its size +# 4. if there are multiple 'shm' space, try to find the default mount +# directory '/dev/shm' is Linux-like system, otherwise return the +# biggest space size. + + +def _parse_size_in_M(size_str): + num, unit = size_str[:-1], size_str[-1] + assert unit in SIZE_UNIT, \ + "unknown shm size unit {}".format(unit) + return float(num) * \ + (1024 ** (SIZE_UNIT.index(unit) - 1)) + + +def _get_shared_memory_size_in_M(): + try: + df_infos = os.popen(SHM_QUERY_CMD).readlines() + except: + return None + else: + shm_infos = [] + for df_info in df_infos: + info = df_info.strip() + if info.find(SHM_KEY) >= 0: + shm_infos.append(info.split()) + + if len(shm_infos) == 0: + return None + elif len(shm_infos) == 1: + return _parse_size_in_M(shm_infos[0][3]) + else: + default_mount_infos = [ + si for si in shm_infos if si[-1] == SHM_DEFAULT_MOUNT + ] + if default_mount_infos: + return _parse_size_in_M(default_mount_infos[0][3]) + else: + return max([_parse_size_in_M(si[3]) for si in shm_infos]) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__init__.py new file mode 100644 index 000000000..3854d3d25 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__init__.py @@ -0,0 +1,29 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import coco +from . import voc +from . import widerface +from . import category +from . import keypoint_coco +from . import mot +from . import sniper_coco + +from .coco import * +from .voc import * +from .widerface import * +from .category import * +from .keypoint_coco import * +from .mot import * +from .sniper_coco import SniperCOCODataSet diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..ba53aeca6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/category.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/category.cpython-37.pyc new file mode 100644 index 000000000..b482ceca7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/category.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/coco.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/coco.cpython-37.pyc new file mode 100644 index 000000000..41c30d1af Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/coco.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/dataset.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/dataset.cpython-37.pyc new file mode 100644 index 000000000..5111b072a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/dataset.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/keypoint_coco.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/keypoint_coco.cpython-37.pyc new file mode 100644 index 000000000..86a391729 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/keypoint_coco.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/mot.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/mot.cpython-37.pyc new file mode 100644 index 000000000..6a3f03ffc Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/mot.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/sniper_coco.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/sniper_coco.cpython-37.pyc new file mode 100644 index 000000000..aecd87e2f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/sniper_coco.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/voc.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/voc.cpython-37.pyc new file mode 100644 index 000000000..f028fb109 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/voc.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/widerface.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/widerface.cpython-37.pyc new file mode 100644 index 000000000..b402cc0cf Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/__pycache__/widerface.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/category.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/category.py new file mode 100644 index 000000000..9390e54c4 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/category.py @@ -0,0 +1,904 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os + +from ppdet.data.source.voc import pascalvoc_label +from ppdet.data.source.widerface import widerface_label +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['get_categories'] + + +def get_categories(metric_type, anno_file=None, arch=None): + """ + Get class id to category id map and category id + to category name map from annotation file. + + Args: + metric_type (str): metric type, currently support 'coco', 'voc', 'oid' + and 'widerface'. + anno_file (str): annotation file path + """ + if arch == 'keypoint_arch': + return (None, {'id': 'keypoint'}) + + if metric_type.lower() == 'coco' or metric_type.lower( + ) == 'rbox' or metric_type.lower() == 'snipercoco': + if anno_file and os.path.isfile(anno_file): + # lazy import pycocotools here + from pycocotools.coco import COCO + + coco = COCO(anno_file) + cats = coco.loadCats(coco.getCatIds()) + + clsid2catid = {i: cat['id'] for i, cat in enumerate(cats)} + catid2name = {cat['id']: cat['name'] for cat in cats} + return clsid2catid, catid2name + + # anno file not exist, load default categories of COCO17 + else: + if metric_type.lower() == 'rbox': + return _dota_category() + + return _coco17_category() + + elif metric_type.lower() == 'voc': + if anno_file and os.path.isfile(anno_file): + cats = [] + with open(anno_file) as f: + for line in f.readlines(): + cats.append(line.strip()) + + if cats[0] == 'background': + cats = cats[1:] + + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + + return clsid2catid, catid2name + + # anno file not exist, load default categories of + # VOC all 20 categories + else: + return _vocall_category() + + elif metric_type.lower() == 'oid': + if anno_file and os.path.isfile(anno_file): + logger.warning("only default categories support for OID19") + return _oid19_category() + + elif metric_type.lower() == 'widerface': + return _widerface_category() + + elif metric_type.lower() == 'keypointtopdowncocoeval' or metric_type.lower( + ) == 'keypointtopdownmpiieval': + return (None, {'id': 'keypoint'}) + + elif metric_type.lower() in ['mot', 'motdet', 'reid']: + if anno_file and os.path.isfile(anno_file): + cats = [] + with open(anno_file) as f: + for line in f.readlines(): + cats.append(line.strip()) + if cats[0] == 'background': + cats = cats[1:] + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + return clsid2catid, catid2name + # anno file not exist, load default category 'pedestrian'. + else: + return _mot_category(category='pedestrian') + + elif metric_type.lower() in ['kitti', 'bdd100kmot']: + return _mot_category(category='vehicle') + + elif metric_type.lower() in ['mcmot']: + if anno_file and os.path.isfile(anno_file): + cats = [] + with open(anno_file) as f: + for line in f.readlines(): + cats.append(line.strip()) + if cats[0] == 'background': + cats = cats[1:] + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + return clsid2catid, catid2name + # anno file not exist, load default categories of visdrone all 10 categories + else: + return _visdrone_category() + + else: + raise ValueError("unknown metric type {}".format(metric_type)) + + +def _mot_category(category='pedestrian'): + """ + Get class id to category id map and category id + to category name map of mot dataset + """ + label_map = {category: 0} + label_map = sorted(label_map.items(), key=lambda x: x[1]) + cats = [l[0] for l in label_map] + + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + + return clsid2catid, catid2name + + +def _coco17_category(): + """ + Get class id to category id map and category id + to category name map of COCO2017 dataset + + """ + clsid2catid = { + 1: 1, + 2: 2, + 3: 3, + 4: 4, + 5: 5, + 6: 6, + 7: 7, + 8: 8, + 9: 9, + 10: 10, + 11: 11, + 12: 13, + 13: 14, + 14: 15, + 15: 16, + 16: 17, + 17: 18, + 18: 19, + 19: 20, + 20: 21, + 21: 22, + 22: 23, + 23: 24, + 24: 25, + 25: 27, + 26: 28, + 27: 31, + 28: 32, + 29: 33, + 30: 34, + 31: 35, + 32: 36, + 33: 37, + 34: 38, + 35: 39, + 36: 40, + 37: 41, + 38: 42, + 39: 43, + 40: 44, + 41: 46, + 42: 47, + 43: 48, + 44: 49, + 45: 50, + 46: 51, + 47: 52, + 48: 53, + 49: 54, + 50: 55, + 51: 56, + 52: 57, + 53: 58, + 54: 59, + 55: 60, + 56: 61, + 57: 62, + 58: 63, + 59: 64, + 60: 65, + 61: 67, + 62: 70, + 63: 72, + 64: 73, + 65: 74, + 66: 75, + 67: 76, + 68: 77, + 69: 78, + 70: 79, + 71: 80, + 72: 81, + 73: 82, + 74: 84, + 75: 85, + 76: 86, + 77: 87, + 78: 88, + 79: 89, + 80: 90 + } + + catid2name = { + 0: 'background', + 1: 'person', + 2: 'bicycle', + 3: 'car', + 4: 'motorcycle', + 5: 'airplane', + 6: 'bus', + 7: 'train', + 8: 'truck', + 9: 'boat', + 10: 'traffic light', + 11: 'fire hydrant', + 13: 'stop sign', + 14: 'parking meter', + 15: 'bench', + 16: 'bird', + 17: 'cat', + 18: 'dog', + 19: 'horse', + 20: 'sheep', + 21: 'cow', + 22: 'elephant', + 23: 'bear', + 24: 'zebra', + 25: 'giraffe', + 27: 'backpack', + 28: 'umbrella', + 31: 'handbag', + 32: 'tie', + 33: 'suitcase', + 34: 'frisbee', + 35: 'skis', + 36: 'snowboard', + 37: 'sports ball', + 38: 'kite', + 39: 'baseball bat', + 40: 'baseball glove', + 41: 'skateboard', + 42: 'surfboard', + 43: 'tennis racket', + 44: 'bottle', + 46: 'wine glass', + 47: 'cup', + 48: 'fork', + 49: 'knife', + 50: 'spoon', + 51: 'bowl', + 52: 'banana', + 53: 'apple', + 54: 'sandwich', + 55: 'orange', + 56: 'broccoli', + 57: 'carrot', + 58: 'hot dog', + 59: 'pizza', + 60: 'donut', + 61: 'cake', + 62: 'chair', + 63: 'couch', + 64: 'potted plant', + 65: 'bed', + 67: 'dining table', + 70: 'toilet', + 72: 'tv', + 73: 'laptop', + 74: 'mouse', + 75: 'remote', + 76: 'keyboard', + 77: 'cell phone', + 78: 'microwave', + 79: 'oven', + 80: 'toaster', + 81: 'sink', + 82: 'refrigerator', + 84: 'book', + 85: 'clock', + 86: 'vase', + 87: 'scissors', + 88: 'teddy bear', + 89: 'hair drier', + 90: 'toothbrush' + } + + clsid2catid = {k - 1: v for k, v in clsid2catid.items()} + catid2name.pop(0) + + return clsid2catid, catid2name + + +def _dota_category(): + """ + Get class id to category id map and category id + to category name map of dota dataset + """ + catid2name = { + 0: 'background', + 1: 'plane', + 2: 'baseball-diamond', + 3: 'bridge', + 4: 'ground-track-field', + 5: 'small-vehicle', + 6: 'large-vehicle', + 7: 'ship', + 8: 'tennis-court', + 9: 'basketball-court', + 10: 'storage-tank', + 11: 'soccer-ball-field', + 12: 'roundabout', + 13: 'harbor', + 14: 'swimming-pool', + 15: 'helicopter' + } + catid2name.pop(0) + clsid2catid = {i: i + 1 for i in range(len(catid2name))} + return clsid2catid, catid2name + + +def _vocall_category(): + """ + Get class id to category id map and category id + to category name map of mixup voc dataset + + """ + label_map = pascalvoc_label() + label_map = sorted(label_map.items(), key=lambda x: x[1]) + cats = [l[0] for l in label_map] + + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + + return clsid2catid, catid2name + + +def _widerface_category(): + label_map = widerface_label() + label_map = sorted(label_map.items(), key=lambda x: x[1]) + cats = [l[0] for l in label_map] + clsid2catid = {i: i for i in range(len(cats))} + catid2name = {i: name for i, name in enumerate(cats)} + + return clsid2catid, catid2name + + +def _oid19_category(): + clsid2catid = {k: k + 1 for k in range(500)} + + catid2name = { + 0: "background", + 1: "Infant bed", + 2: "Rose", + 3: "Flag", + 4: "Flashlight", + 5: "Sea turtle", + 6: "Camera", + 7: "Animal", + 8: "Glove", + 9: "Crocodile", + 10: "Cattle", + 11: "House", + 12: "Guacamole", + 13: "Penguin", + 14: "Vehicle registration plate", + 15: "Bench", + 16: "Ladybug", + 17: "Human nose", + 18: "Watermelon", + 19: "Flute", + 20: "Butterfly", + 21: "Washing machine", + 22: "Raccoon", + 23: "Segway", + 24: "Taco", + 25: "Jellyfish", + 26: "Cake", + 27: "Pen", + 28: "Cannon", + 29: "Bread", + 30: "Tree", + 31: "Shellfish", + 32: "Bed", + 33: "Hamster", + 34: "Hat", + 35: "Toaster", + 36: "Sombrero", + 37: "Tiara", + 38: "Bowl", + 39: "Dragonfly", + 40: "Moths and butterflies", + 41: "Antelope", + 42: "Vegetable", + 43: "Torch", + 44: "Building", + 45: "Power plugs and sockets", + 46: "Blender", + 47: "Billiard table", + 48: "Cutting board", + 49: "Bronze sculpture", + 50: "Turtle", + 51: "Broccoli", + 52: "Tiger", + 53: "Mirror", + 54: "Bear", + 55: "Zucchini", + 56: "Dress", + 57: "Volleyball", + 58: "Guitar", + 59: "Reptile", + 60: "Golf cart", + 61: "Tart", + 62: "Fedora", + 63: "Carnivore", + 64: "Car", + 65: "Lighthouse", + 66: "Coffeemaker", + 67: "Food processor", + 68: "Truck", + 69: "Bookcase", + 70: "Surfboard", + 71: "Footwear", + 72: "Bench", + 73: "Necklace", + 74: "Flower", + 75: "Radish", + 76: "Marine mammal", + 77: "Frying pan", + 78: "Tap", + 79: "Peach", + 80: "Knife", + 81: "Handbag", + 82: "Laptop", + 83: "Tent", + 84: "Ambulance", + 85: "Christmas tree", + 86: "Eagle", + 87: "Limousine", + 88: "Kitchen & dining room table", + 89: "Polar bear", + 90: "Tower", + 91: "Football", + 92: "Willow", + 93: "Human head", + 94: "Stop sign", + 95: "Banana", + 96: "Mixer", + 97: "Binoculars", + 98: "Dessert", + 99: "Bee", + 100: "Chair", + 101: "Wood-burning stove", + 102: "Flowerpot", + 103: "Beaker", + 104: "Oyster", + 105: "Woodpecker", + 106: "Harp", + 107: "Bathtub", + 108: "Wall clock", + 109: "Sports uniform", + 110: "Rhinoceros", + 111: "Beehive", + 112: "Cupboard", + 113: "Chicken", + 114: "Man", + 115: "Blue jay", + 116: "Cucumber", + 117: "Balloon", + 118: "Kite", + 119: "Fireplace", + 120: "Lantern", + 121: "Missile", + 122: "Book", + 123: "Spoon", + 124: "Grapefruit", + 125: "Squirrel", + 126: "Orange", + 127: "Coat", + 128: "Punching bag", + 129: "Zebra", + 130: "Billboard", + 131: "Bicycle", + 132: "Door handle", + 133: "Mechanical fan", + 134: "Ring binder", + 135: "Table", + 136: "Parrot", + 137: "Sock", + 138: "Vase", + 139: "Weapon", + 140: "Shotgun", + 141: "Glasses", + 142: "Seahorse", + 143: "Belt", + 144: "Watercraft", + 145: "Window", + 146: "Giraffe", + 147: "Lion", + 148: "Tire", + 149: "Vehicle", + 150: "Canoe", + 151: "Tie", + 152: "Shelf", + 153: "Picture frame", + 154: "Printer", + 155: "Human leg", + 156: "Boat", + 157: "Slow cooker", + 158: "Croissant", + 159: "Candle", + 160: "Pancake", + 161: "Pillow", + 162: "Coin", + 163: "Stretcher", + 164: "Sandal", + 165: "Woman", + 166: "Stairs", + 167: "Harpsichord", + 168: "Stool", + 169: "Bus", + 170: "Suitcase", + 171: "Human mouth", + 172: "Juice", + 173: "Skull", + 174: "Door", + 175: "Violin", + 176: "Chopsticks", + 177: "Digital clock", + 178: "Sunflower", + 179: "Leopard", + 180: "Bell pepper", + 181: "Harbor seal", + 182: "Snake", + 183: "Sewing machine", + 184: "Goose", + 185: "Helicopter", + 186: "Seat belt", + 187: "Coffee cup", + 188: "Microwave oven", + 189: "Hot dog", + 190: "Countertop", + 191: "Serving tray", + 192: "Dog bed", + 193: "Beer", + 194: "Sunglasses", + 195: "Golf ball", + 196: "Waffle", + 197: "Palm tree", + 198: "Trumpet", + 199: "Ruler", + 200: "Helmet", + 201: "Ladder", + 202: "Office building", + 203: "Tablet computer", + 204: "Toilet paper", + 205: "Pomegranate", + 206: "Skirt", + 207: "Gas stove", + 208: "Cookie", + 209: "Cart", + 210: "Raven", + 211: "Egg", + 212: "Burrito", + 213: "Goat", + 214: "Kitchen knife", + 215: "Skateboard", + 216: "Salt and pepper shakers", + 217: "Lynx", + 218: "Boot", + 219: "Platter", + 220: "Ski", + 221: "Swimwear", + 222: "Swimming pool", + 223: "Drinking straw", + 224: "Wrench", + 225: "Drum", + 226: "Ant", + 227: "Human ear", + 228: "Headphones", + 229: "Fountain", + 230: "Bird", + 231: "Jeans", + 232: "Television", + 233: "Crab", + 234: "Microphone", + 235: "Home appliance", + 236: "Snowplow", + 237: "Beetle", + 238: "Artichoke", + 239: "Jet ski", + 240: "Stationary bicycle", + 241: "Human hair", + 242: "Brown bear", + 243: "Starfish", + 244: "Fork", + 245: "Lobster", + 246: "Corded phone", + 247: "Drink", + 248: "Saucer", + 249: "Carrot", + 250: "Insect", + 251: "Clock", + 252: "Castle", + 253: "Tennis racket", + 254: "Ceiling fan", + 255: "Asparagus", + 256: "Jaguar", + 257: "Musical instrument", + 258: "Train", + 259: "Cat", + 260: "Rifle", + 261: "Dumbbell", + 262: "Mobile phone", + 263: "Taxi", + 264: "Shower", + 265: "Pitcher", + 266: "Lemon", + 267: "Invertebrate", + 268: "Turkey", + 269: "High heels", + 270: "Bust", + 271: "Elephant", + 272: "Scarf", + 273: "Barrel", + 274: "Trombone", + 275: "Pumpkin", + 276: "Box", + 277: "Tomato", + 278: "Frog", + 279: "Bidet", + 280: "Human face", + 281: "Houseplant", + 282: "Van", + 283: "Shark", + 284: "Ice cream", + 285: "Swim cap", + 286: "Falcon", + 287: "Ostrich", + 288: "Handgun", + 289: "Whiteboard", + 290: "Lizard", + 291: "Pasta", + 292: "Snowmobile", + 293: "Light bulb", + 294: "Window blind", + 295: "Muffin", + 296: "Pretzel", + 297: "Computer monitor", + 298: "Horn", + 299: "Furniture", + 300: "Sandwich", + 301: "Fox", + 302: "Convenience store", + 303: "Fish", + 304: "Fruit", + 305: "Earrings", + 306: "Curtain", + 307: "Grape", + 308: "Sofa bed", + 309: "Horse", + 310: "Luggage and bags", + 311: "Desk", + 312: "Crutch", + 313: "Bicycle helmet", + 314: "Tick", + 315: "Airplane", + 316: "Canary", + 317: "Spatula", + 318: "Watch", + 319: "Lily", + 320: "Kitchen appliance", + 321: "Filing cabinet", + 322: "Aircraft", + 323: "Cake stand", + 324: "Candy", + 325: "Sink", + 326: "Mouse", + 327: "Wine", + 328: "Wheelchair", + 329: "Goldfish", + 330: "Refrigerator", + 331: "French fries", + 332: "Drawer", + 333: "Treadmill", + 334: "Picnic basket", + 335: "Dice", + 336: "Cabbage", + 337: "Football helmet", + 338: "Pig", + 339: "Person", + 340: "Shorts", + 341: "Gondola", + 342: "Honeycomb", + 343: "Doughnut", + 344: "Chest of drawers", + 345: "Land vehicle", + 346: "Bat", + 347: "Monkey", + 348: "Dagger", + 349: "Tableware", + 350: "Human foot", + 351: "Mug", + 352: "Alarm clock", + 353: "Pressure cooker", + 354: "Human hand", + 355: "Tortoise", + 356: "Baseball glove", + 357: "Sword", + 358: "Pear", + 359: "Miniskirt", + 360: "Traffic sign", + 361: "Girl", + 362: "Roller skates", + 363: "Dinosaur", + 364: "Porch", + 365: "Human beard", + 366: "Submarine sandwich", + 367: "Screwdriver", + 368: "Strawberry", + 369: "Wine glass", + 370: "Seafood", + 371: "Racket", + 372: "Wheel", + 373: "Sea lion", + 374: "Toy", + 375: "Tea", + 376: "Tennis ball", + 377: "Waste container", + 378: "Mule", + 379: "Cricket ball", + 380: "Pineapple", + 381: "Coconut", + 382: "Doll", + 383: "Coffee table", + 384: "Snowman", + 385: "Lavender", + 386: "Shrimp", + 387: "Maple", + 388: "Cowboy hat", + 389: "Goggles", + 390: "Rugby ball", + 391: "Caterpillar", + 392: "Poster", + 393: "Rocket", + 394: "Organ", + 395: "Saxophone", + 396: "Traffic light", + 397: "Cocktail", + 398: "Plastic bag", + 399: "Squash", + 400: "Mushroom", + 401: "Hamburger", + 402: "Light switch", + 403: "Parachute", + 404: "Teddy bear", + 405: "Winter melon", + 406: "Deer", + 407: "Musical keyboard", + 408: "Plumbing fixture", + 409: "Scoreboard", + 410: "Baseball bat", + 411: "Envelope", + 412: "Adhesive tape", + 413: "Briefcase", + 414: "Paddle", + 415: "Bow and arrow", + 416: "Telephone", + 417: "Sheep", + 418: "Jacket", + 419: "Boy", + 420: "Pizza", + 421: "Otter", + 422: "Office supplies", + 423: "Couch", + 424: "Cello", + 425: "Bull", + 426: "Camel", + 427: "Ball", + 428: "Duck", + 429: "Whale", + 430: "Shirt", + 431: "Tank", + 432: "Motorcycle", + 433: "Accordion", + 434: "Owl", + 435: "Porcupine", + 436: "Sun hat", + 437: "Nail", + 438: "Scissors", + 439: "Swan", + 440: "Lamp", + 441: "Crown", + 442: "Piano", + 443: "Sculpture", + 444: "Cheetah", + 445: "Oboe", + 446: "Tin can", + 447: "Mango", + 448: "Tripod", + 449: "Oven", + 450: "Mouse", + 451: "Barge", + 452: "Coffee", + 453: "Snowboard", + 454: "Common fig", + 455: "Salad", + 456: "Marine invertebrates", + 457: "Umbrella", + 458: "Kangaroo", + 459: "Human arm", + 460: "Measuring cup", + 461: "Snail", + 462: "Loveseat", + 463: "Suit", + 464: "Teapot", + 465: "Bottle", + 466: "Alpaca", + 467: "Kettle", + 468: "Trousers", + 469: "Popcorn", + 470: "Centipede", + 471: "Spider", + 472: "Sparrow", + 473: "Plate", + 474: "Bagel", + 475: "Personal care", + 476: "Apple", + 477: "Brassiere", + 478: "Bathroom cabinet", + 479: "studio couch", + 480: "Computer keyboard", + 481: "Table tennis racket", + 482: "Sushi", + 483: "Cabinetry", + 484: "Street light", + 485: "Towel", + 486: "Nightstand", + 487: "Rabbit", + 488: "Dolphin", + 489: "Dog", + 490: "Jug", + 491: "Wok", + 492: "Fire hydrant", + 493: "Human eye", + 494: "Skyscraper", + 495: "Backpack", + 496: "Potato", + 497: "Paper towel", + 498: "Lifejacket", + 499: "Bicycle wheel", + 500: "Toilet", + } + + return clsid2catid, catid2name + + +def _visdrone_category(): + clsid2catid = {i: i for i in range(10)} + + catid2name = { + 0: 'pedestrian', + 1: 'people', + 2: 'bicycle', + 3: 'car', + 4: 'van', + 5: 'truck', + 6: 'tricycle', + 7: 'awning-tricycle', + 8: 'bus', + 9: 'motor' + } + return clsid2catid, catid2name diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/coco.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/coco.py new file mode 100644 index 000000000..0efc9ae0e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/coco.py @@ -0,0 +1,251 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import numpy as np +from ppdet.core.workspace import register, serializable +from .dataset import DetDataset + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class COCODataSet(DetDataset): + """ + Load dataset with COCO format. + + Args: + dataset_dir (str): root directory for dataset. + image_dir (str): directory for images. + anno_path (str): coco annotation file path. + data_fields (list): key name of data dictionary, at least have 'image'. + sample_num (int): number of samples to load, -1 means all. + load_crowd (bool): whether to load crowded ground-truth. + False as default + allow_empty (bool): whether to load empty entry. False as default + empty_ratio (float): the ratio of empty record number to total + record's, if empty_ratio is out of [0. ,1.), do not sample the + records and use all the empty entries. 1. as default + """ + + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + data_fields=['image'], + sample_num=-1, + load_crowd=False, + allow_empty=False, + empty_ratio=1.): + super(COCODataSet, self).__init__(dataset_dir, image_dir, anno_path, + data_fields, sample_num) + self.load_image_only = False + self.load_semantic = False + self.load_crowd = load_crowd + self.allow_empty = allow_empty + self.empty_ratio = empty_ratio + + def _sample_empty(self, records, num): + # if empty_ratio is out of [0. ,1.), do not sample the records + if self.empty_ratio < 0. or self.empty_ratio >= 1.: + return records + import random + sample_num = min( + int(num * self.empty_ratio / (1 - self.empty_ratio)), len(records)) + records = random.sample(records, sample_num) + return records + + def parse_dataset(self): + anno_path = os.path.join(self.dataset_dir, self.anno_path) + image_dir = os.path.join(self.dataset_dir, self.image_dir) + + assert anno_path.endswith('.json'), \ + 'invalid coco annotation file: ' + anno_path + from pycocotools.coco import COCO + coco = COCO(anno_path) + img_ids = coco.getImgIds() + img_ids.sort() + cat_ids = coco.getCatIds() + records = [] + empty_records = [] + ct = 0 + + self.catid2clsid = dict({catid: i for i, catid in enumerate(cat_ids)}) + self.cname2cid = dict({ + coco.loadCats(catid)[0]['name']: clsid + for catid, clsid in self.catid2clsid.items() + }) + + if 'annotations' not in coco.dataset: + self.load_image_only = True + logger.warning('Annotation file: {} does not contains ground truth ' + 'and load image information only.'.format(anno_path)) + + for img_id in img_ids: + img_anno = coco.loadImgs([img_id])[0] + im_fname = img_anno['file_name'] + im_w = float(img_anno['width']) + im_h = float(img_anno['height']) + + im_path = os.path.join(image_dir, + im_fname) if image_dir else im_fname + im_path = im_path.rstrip() + is_empty = False + + if not os.path.exists(im_path): + logger.warning('Illegal image file: {}, and it will be ' + 'ignored'.format(im_path)) + continue + + if im_w < 0 or im_h < 0: + logger.warning('Illegal width: {} or height: {} in annotation, ' + 'and im_id: {} will be ignored'.format( + im_w, im_h, img_id)) + continue + + coco_rec = { + 'im_file': im_path, + 'im_id': np.array([img_id]), + 'h': im_h, + 'w': im_w, + } if 'image' in self.data_fields else {} + + if not self.load_image_only: + ins_anno_ids = coco.getAnnIds( + imgIds=[img_id], iscrowd=None if self.load_crowd else False) + instances = coco.loadAnns(ins_anno_ids) + + bboxes = [] + is_rbox_anno = False + for inst in instances: + # check gt bbox + if inst.get('ignore', False): + continue + if 'bbox' not in inst.keys(): + continue + else: + if not any(np.array(inst['bbox'])): + continue + + # read rbox anno or not + is_rbox_anno = True if len(inst['bbox']) == 5 else False + if is_rbox_anno: + xc, yc, box_w, box_h, angle = inst['bbox'] + x1 = xc - box_w / 2.0 + y1 = yc - box_h / 2.0 + x2 = x1 + box_w + y2 = y1 + box_h + else: + x1, y1, box_w, box_h = inst['bbox'] + x2 = x1 + box_w + y2 = y1 + box_h + eps = 1e-5 + if inst['area'] > 0 and x2 - x1 > eps and y2 - y1 > eps: + inst['clean_bbox'] = [ + round(float(x), 3) for x in [x1, y1, x2, y2] + ] + if is_rbox_anno: + inst['clean_rbox'] = [xc, yc, box_w, box_h, angle] + bboxes.append(inst) + else: + logger.warning( + 'Found an invalid bbox in annotations: im_id: {}, ' + 'area: {} x1: {}, y1: {}, x2: {}, y2: {}.'.format( + img_id, float(inst['area']), x1, y1, x2, y2)) + + num_bbox = len(bboxes) + if num_bbox <= 0 and not self.allow_empty: + continue + elif num_bbox <= 0: + is_empty = True + + gt_bbox = np.zeros((num_bbox, 4), dtype=np.float32) + if is_rbox_anno: + gt_rbox = np.zeros((num_bbox, 5), dtype=np.float32) + gt_theta = np.zeros((num_bbox, 1), dtype=np.int32) + gt_class = np.zeros((num_bbox, 1), dtype=np.int32) + is_crowd = np.zeros((num_bbox, 1), dtype=np.int32) + gt_poly = [None] * num_bbox + + has_segmentation = False + for i, box in enumerate(bboxes): + catid = box['category_id'] + gt_class[i][0] = self.catid2clsid[catid] + gt_bbox[i, :] = box['clean_bbox'] + # xc, yc, w, h, theta + if is_rbox_anno: + gt_rbox[i, :] = box['clean_rbox'] + is_crowd[i][0] = box['iscrowd'] + # check RLE format + if 'segmentation' in box and box['iscrowd'] == 1: + gt_poly[i] = [[0.0, 0.0, 0.0, 0.0, 0.0, 0.0]] + elif 'segmentation' in box and box['segmentation']: + if not np.array(box['segmentation'] + ).size > 0 and not self.allow_empty: + bboxes.pop(i) + gt_poly.pop(i) + np.delete(is_crowd, i) + np.delete(gt_class, i) + np.delete(gt_bbox, i) + else: + gt_poly[i] = box['segmentation'] + has_segmentation = True + + if has_segmentation and not any( + gt_poly) and not self.allow_empty: + continue + + if is_rbox_anno: + gt_rec = { + 'is_crowd': is_crowd, + 'gt_class': gt_class, + 'gt_bbox': gt_bbox, + 'gt_rbox': gt_rbox, + 'gt_poly': gt_poly, + } + else: + gt_rec = { + 'is_crowd': is_crowd, + 'gt_class': gt_class, + 'gt_bbox': gt_bbox, + 'gt_poly': gt_poly, + } + + for k, v in gt_rec.items(): + if k in self.data_fields: + coco_rec[k] = v + + # TODO: remove load_semantic + if self.load_semantic and 'semantic' in self.data_fields: + seg_path = os.path.join(self.dataset_dir, 'stuffthingmaps', + 'train2017', im_fname[:-3] + 'png') + coco_rec.update({'semantic': seg_path}) + + logger.debug('Load file: {}, im_id: {}, h: {}, w: {}.'.format( + im_path, img_id, im_h, im_w)) + if is_empty: + empty_records.append(coco_rec) + else: + records.append(coco_rec) + ct += 1 + if self.sample_num > 0 and ct >= self.sample_num: + break + assert ct > 0, 'not found any coco record in %s' % (anno_path) + logger.debug('{} samples in file {}'.format(ct, anno_path)) + if self.allow_empty and len(empty_records) > 0: + empty_records = self._sample_empty(empty_records, len(records)) + records += empty_records + self.roidbs = records diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/dataset.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/dataset.py new file mode 100644 index 000000000..1bef548e6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/dataset.py @@ -0,0 +1,197 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import numpy as np + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence +from paddle.io import Dataset +from ppdet.core.workspace import register, serializable +from ppdet.utils.download import get_dataset_path +import copy + + +@serializable +class DetDataset(Dataset): + """ + Load detection dataset. + + Args: + dataset_dir (str): root directory for dataset. + image_dir (str): directory for images. + anno_path (str): annotation file path. + data_fields (list): key name of data dictionary, at least have 'image'. + sample_num (int): number of samples to load, -1 means all. + use_default_label (bool): whether to load default label list. + """ + + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + data_fields=['image'], + sample_num=-1, + use_default_label=None, + **kwargs): + super(DetDataset, self).__init__() + self.dataset_dir = dataset_dir if dataset_dir is not None else '' + self.anno_path = anno_path + self.image_dir = image_dir if image_dir is not None else '' + self.data_fields = data_fields + self.sample_num = sample_num + self.use_default_label = use_default_label + self._epoch = 0 + self._curr_iter = 0 + + def __len__(self, ): + return len(self.roidbs) + + def __getitem__(self, idx): + # data batch + roidb = copy.deepcopy(self.roidbs[idx]) + if self.mixup_epoch == 0 or self._epoch < self.mixup_epoch: + n = len(self.roidbs) + idx = np.random.randint(n) + roidb = [roidb, copy.deepcopy(self.roidbs[idx])] + elif self.cutmix_epoch == 0 or self._epoch < self.cutmix_epoch: + n = len(self.roidbs) + idx = np.random.randint(n) + roidb = [roidb, copy.deepcopy(self.roidbs[idx])] + elif self.mosaic_epoch == 0 or self._epoch < self.mosaic_epoch: + n = len(self.roidbs) + roidb = [roidb, ] + [ + copy.deepcopy(self.roidbs[np.random.randint(n)]) + for _ in range(3) + ] + if isinstance(roidb, Sequence): + for r in roidb: + r['curr_iter'] = self._curr_iter + else: + roidb['curr_iter'] = self._curr_iter + self._curr_iter += 1 + + return self.transform(roidb) + + def check_or_download_dataset(self): + self.dataset_dir = get_dataset_path(self.dataset_dir, self.anno_path, + self.image_dir) + + def set_kwargs(self, **kwargs): + self.mixup_epoch = kwargs.get('mixup_epoch', -1) + self.cutmix_epoch = kwargs.get('cutmix_epoch', -1) + self.mosaic_epoch = kwargs.get('mosaic_epoch', -1) + + def set_transform(self, transform): + self.transform = transform + + def set_epoch(self, epoch_id): + self._epoch = epoch_id + + def parse_dataset(self, ): + raise NotImplementedError( + "Need to implement parse_dataset method of Dataset") + + def get_anno(self): + if self.anno_path is None: + return + return os.path.join(self.dataset_dir, self.anno_path) + + +def _is_valid_file(f, extensions=('.jpg', '.jpeg', '.png', '.bmp')): + return f.lower().endswith(extensions) + + +def _make_dataset(dir): + dir = os.path.expanduser(dir) + if not os.path.isdir(dir): + raise ('{} should be a dir'.format(dir)) + images = [] + for root, _, fnames in sorted(os.walk(dir, followlinks=True)): + for fname in sorted(fnames): + path = os.path.join(root, fname) + if _is_valid_file(path): + images.append(path) + return images + + +@register +@serializable +class ImageFolder(DetDataset): + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + sample_num=-1, + use_default_label=None, + **kwargs): + super(ImageFolder, self).__init__( + dataset_dir, + image_dir, + anno_path, + sample_num=sample_num, + use_default_label=use_default_label) + self._imid2path = {} + self.roidbs = None + self.sample_num = sample_num + + def check_or_download_dataset(self): + if self.dataset_dir: + # NOTE: ImageFolder is only used for prediction, in + # infer mode, image_dir is set by set_images + # so we only check anno_path here + self.dataset_dir = get_dataset_path(self.dataset_dir, + self.anno_path, None) + + def parse_dataset(self, ): + if not self.roidbs: + self.roidbs = self._load_images() + + def _parse(self): + image_dir = self.image_dir + if not isinstance(image_dir, Sequence): + image_dir = [image_dir] + images = [] + for im_dir in image_dir: + if os.path.isdir(im_dir): + im_dir = os.path.join(self.dataset_dir, im_dir) + images.extend(_make_dataset(im_dir)) + elif os.path.isfile(im_dir) and _is_valid_file(im_dir): + images.append(im_dir) + return images + + def _load_images(self): + images = self._parse() + ct = 0 + records = [] + for image in images: + assert image != '' and os.path.isfile(image), \ + "Image {} not found".format(image) + if self.sample_num > 0 and ct >= self.sample_num: + break + rec = {'im_id': np.array([ct]), 'im_file': image} + self._imid2path[ct] = image + ct += 1 + records.append(rec) + assert len(records) > 0, "No image file found" + return records + + def get_imid2path(self): + return self._imid2path + + def set_images(self, images): + self.image_dir = images + self.roidbs = self._load_images() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/keypoint_coco.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/keypoint_coco.py new file mode 100644 index 000000000..fdea57ada --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/keypoint_coco.py @@ -0,0 +1,674 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import cv2 +import numpy as np +import json +import copy +import pycocotools +from pycocotools.coco import COCO +from .dataset import DetDataset +from ppdet.core.workspace import register, serializable + + +@serializable +class KeypointBottomUpBaseDataset(DetDataset): + """Base class for bottom-up datasets. Adapted from + https://github.com/open-mmlab/mmpose + + All datasets should subclass it. + All subclasses should overwrite: + Methods:`_get_imganno` + + Args: + dataset_dir (str): Root path to the dataset. + anno_path (str): Relative path to the annotation file. + image_dir (str): Path to a directory where images are held. + Default: None. + num_joints (int): keypoint numbers + transform (composed(operators)): A sequence of data transforms. + shard (list): [rank, worldsize], the distributed env params + test_mode (bool): Store True when building test or + validation dataset. Default: False. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + transform=[], + shard=[0, 1], + test_mode=False): + super().__init__(dataset_dir, image_dir, anno_path) + self.image_info = {} + self.ann_info = {} + + self.img_prefix = os.path.join(dataset_dir, image_dir) + self.transform = transform + self.test_mode = test_mode + + self.ann_info['num_joints'] = num_joints + self.img_ids = [] + + def parse_dataset(self): + pass + + def __len__(self): + """Get dataset length.""" + return len(self.img_ids) + + def _get_imganno(self, idx): + """Get anno for a single image.""" + raise NotImplementedError + + def __getitem__(self, idx): + """Prepare image for training given the index.""" + records = copy.deepcopy(self._get_imganno(idx)) + records['image'] = cv2.imread(records['image_file']) + records['image'] = cv2.cvtColor(records['image'], cv2.COLOR_BGR2RGB) + records['mask'] = (records['mask'] + 0).astype('uint8') + records = self.transform(records) + return records + + def parse_dataset(self): + return + + +@register +@serializable +class KeypointBottomUpCocoDataset(KeypointBottomUpBaseDataset): + """COCO dataset for bottom-up pose estimation. Adapted from + https://github.com/open-mmlab/mmpose + + The dataset loads raw features and apply specified transforms + to return a dict containing the image tensors and other information. + + COCO keypoint indexes:: + + 0: 'nose', + 1: 'left_eye', + 2: 'right_eye', + 3: 'left_ear', + 4: 'right_ear', + 5: 'left_shoulder', + 6: 'right_shoulder', + 7: 'left_elbow', + 8: 'right_elbow', + 9: 'left_wrist', + 10: 'right_wrist', + 11: 'left_hip', + 12: 'right_hip', + 13: 'left_knee', + 14: 'right_knee', + 15: 'left_ankle', + 16: 'right_ankle' + + Args: + dataset_dir (str): Root path to the dataset. + anno_path (str): Relative path to the annotation file. + image_dir (str): Path to a directory where images are held. + Default: None. + num_joints (int): keypoint numbers + transform (composed(operators)): A sequence of data transforms. + shard (list): [rank, worldsize], the distributed env params + test_mode (bool): Store True when building test or + validation dataset. Default: False. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + transform=[], + shard=[0, 1], + test_mode=False): + super().__init__(dataset_dir, image_dir, anno_path, num_joints, + transform, shard, test_mode) + + self.ann_file = os.path.join(dataset_dir, anno_path) + self.shard = shard + self.test_mode = test_mode + + def parse_dataset(self): + self.coco = COCO(self.ann_file) + + self.img_ids = self.coco.getImgIds() + if not self.test_mode: + self.img_ids = [ + img_id for img_id in self.img_ids + if len(self.coco.getAnnIds( + imgIds=img_id, iscrowd=None)) > 0 + ] + blocknum = int(len(self.img_ids) / self.shard[1]) + self.img_ids = self.img_ids[(blocknum * self.shard[0]):(blocknum * ( + self.shard[0] + 1))] + self.num_images = len(self.img_ids) + self.id2name, self.name2id = self._get_mapping_id_name(self.coco.imgs) + self.dataset_name = 'coco' + + cat_ids = self.coco.getCatIds() + self.catid2clsid = dict({catid: i for i, catid in enumerate(cat_ids)}) + print('=> num_images: {}'.format(self.num_images)) + + @staticmethod + def _get_mapping_id_name(imgs): + """ + Args: + imgs (dict): dict of image info. + + Returns: + tuple: Image name & id mapping dicts. + + - id2name (dict): Mapping image id to name. + - name2id (dict): Mapping image name to id. + """ + id2name = {} + name2id = {} + for image_id, image in imgs.items(): + file_name = image['file_name'] + id2name[image_id] = file_name + name2id[file_name] = image_id + + return id2name, name2id + + def _get_imganno(self, idx): + """Get anno for a single image. + + Args: + idx (int): image idx + + Returns: + dict: info for model training + """ + coco = self.coco + img_id = self.img_ids[idx] + ann_ids = coco.getAnnIds(imgIds=img_id) + anno = coco.loadAnns(ann_ids) + + mask = self._get_mask(anno, idx) + anno = [ + obj for obj in anno + if obj['iscrowd'] == 0 or obj['num_keypoints'] > 0 + ] + + joints, orgsize = self._get_joints(anno, idx) + + db_rec = {} + db_rec['im_id'] = img_id + db_rec['image_file'] = os.path.join(self.img_prefix, + self.id2name[img_id]) + db_rec['mask'] = mask + db_rec['joints'] = joints + db_rec['im_shape'] = orgsize + + return db_rec + + def _get_joints(self, anno, idx): + """Get joints for all people in an image.""" + num_people = len(anno) + + joints = np.zeros( + (num_people, self.ann_info['num_joints'], 3), dtype=np.float32) + + for i, obj in enumerate(anno): + joints[i, :self.ann_info['num_joints'], :3] = \ + np.array(obj['keypoints']).reshape([-1, 3]) + + img_info = self.coco.loadImgs(self.img_ids[idx])[0] + joints[..., 0] /= img_info['width'] + joints[..., 1] /= img_info['height'] + orgsize = np.array([img_info['height'], img_info['width']]) + + return joints, orgsize + + def _get_mask(self, anno, idx): + """Get ignore masks to mask out losses.""" + coco = self.coco + img_info = coco.loadImgs(self.img_ids[idx])[0] + + m = np.zeros((img_info['height'], img_info['width']), dtype=np.float32) + + for obj in anno: + if 'segmentation' in obj: + if obj['iscrowd']: + rle = pycocotools.mask.frPyObjects(obj['segmentation'], + img_info['height'], + img_info['width']) + m += pycocotools.mask.decode(rle) + elif obj['num_keypoints'] == 0: + rles = pycocotools.mask.frPyObjects(obj['segmentation'], + img_info['height'], + img_info['width']) + for rle in rles: + m += pycocotools.mask.decode(rle) + + return m < 0.5 + + +@register +@serializable +class KeypointBottomUpCrowdPoseDataset(KeypointBottomUpCocoDataset): + """CrowdPose dataset for bottom-up pose estimation. Adapted from + https://github.com/open-mmlab/mmpose + + The dataset loads raw features and apply specified transforms + to return a dict containing the image tensors and other information. + + CrowdPose keypoint indexes:: + + 0: 'left_shoulder', + 1: 'right_shoulder', + 2: 'left_elbow', + 3: 'right_elbow', + 4: 'left_wrist', + 5: 'right_wrist', + 6: 'left_hip', + 7: 'right_hip', + 8: 'left_knee', + 9: 'right_knee', + 10: 'left_ankle', + 11: 'right_ankle', + 12: 'top_head', + 13: 'neck' + + Args: + dataset_dir (str): Root path to the dataset. + anno_path (str): Relative path to the annotation file. + image_dir (str): Path to a directory where images are held. + Default: None. + num_joints (int): keypoint numbers + transform (composed(operators)): A sequence of data transforms. + shard (list): [rank, worldsize], the distributed env params + test_mode (bool): Store True when building test or + validation dataset. Default: False. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + transform=[], + shard=[0, 1], + test_mode=False): + super().__init__(dataset_dir, image_dir, anno_path, num_joints, + transform, shard, test_mode) + + self.ann_file = os.path.join(dataset_dir, anno_path) + self.shard = shard + self.test_mode = test_mode + + def parse_dataset(self): + self.coco = COCO(self.ann_file) + + self.img_ids = self.coco.getImgIds() + if not self.test_mode: + self.img_ids = [ + img_id for img_id in self.img_ids + if len(self.coco.getAnnIds( + imgIds=img_id, iscrowd=None)) > 0 + ] + blocknum = int(len(self.img_ids) / self.shard[1]) + self.img_ids = self.img_ids[(blocknum * self.shard[0]):(blocknum * ( + self.shard[0] + 1))] + self.num_images = len(self.img_ids) + self.id2name, self.name2id = self._get_mapping_id_name(self.coco.imgs) + + self.dataset_name = 'crowdpose' + print('=> num_images: {}'.format(self.num_images)) + + +@serializable +class KeypointTopDownBaseDataset(DetDataset): + """Base class for top_down datasets. + + All datasets should subclass it. + All subclasses should overwrite: + Methods:`_get_db` + + Args: + dataset_dir (str): Root path to the dataset. + image_dir (str): Path to a directory where images are held. + anno_path (str): Relative path to the annotation file. + num_joints (int): keypoint numbers + transform (composed(operators)): A sequence of data transforms. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + transform=[]): + super().__init__(dataset_dir, image_dir, anno_path) + self.image_info = {} + self.ann_info = {} + + self.img_prefix = os.path.join(dataset_dir, image_dir) + self.transform = transform + + self.ann_info['num_joints'] = num_joints + self.db = [] + + def __len__(self): + """Get dataset length.""" + return len(self.db) + + def _get_db(self): + """Get a sample""" + raise NotImplementedError + + def __getitem__(self, idx): + """Prepare sample for training given the index.""" + records = copy.deepcopy(self.db[idx]) + records['image'] = cv2.imread(records['image_file'], cv2.IMREAD_COLOR | + cv2.IMREAD_IGNORE_ORIENTATION) + records['image'] = cv2.cvtColor(records['image'], cv2.COLOR_BGR2RGB) + records['score'] = records['score'] if 'score' in records else 1 + records = self.transform(records) + # print('records', records) + return records + + +@register +@serializable +class KeypointTopDownCocoDataset(KeypointTopDownBaseDataset): + """COCO dataset for top-down pose estimation. Adapted from + https://github.com/leoxiaobin/deep-high-resolution-net.pytorch + Copyright (c) Microsoft, under the MIT License. + + The dataset loads raw features and apply specified transforms + to return a dict containing the image tensors and other information. + + COCO keypoint indexes: + + 0: 'nose', + 1: 'left_eye', + 2: 'right_eye', + 3: 'left_ear', + 4: 'right_ear', + 5: 'left_shoulder', + 6: 'right_shoulder', + 7: 'left_elbow', + 8: 'right_elbow', + 9: 'left_wrist', + 10: 'right_wrist', + 11: 'left_hip', + 12: 'right_hip', + 13: 'left_knee', + 14: 'right_knee', + 15: 'left_ankle', + 16: 'right_ankle' + + Args: + dataset_dir (str): Root path to the dataset. + image_dir (str): Path to a directory where images are held. + anno_path (str): Relative path to the annotation file. + num_joints (int): Keypoint numbers + trainsize (list):[w, h] Image target size + transform (composed(operators)): A sequence of data transforms. + bbox_file (str): Path to a detection bbox file + Default: None. + use_gt_bbox (bool): Whether to use ground truth bbox + Default: True. + pixel_std (int): The pixel std of the scale + Default: 200. + image_thre (float): The threshold to filter the detection box + Default: 0.0. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + trainsize, + transform=[], + bbox_file=None, + use_gt_bbox=True, + pixel_std=200, + image_thre=0.0): + super().__init__(dataset_dir, image_dir, anno_path, num_joints, + transform) + + self.bbox_file = bbox_file + self.use_gt_bbox = use_gt_bbox + self.trainsize = trainsize + self.pixel_std = pixel_std + self.image_thre = image_thre + self.dataset_name = 'coco' + + def parse_dataset(self): + if self.use_gt_bbox: + self.db = self._load_coco_keypoint_annotations() + else: + self.db = self._load_coco_person_detection_results() + + def _load_coco_keypoint_annotations(self): + coco = COCO(self.get_anno()) + img_ids = coco.getImgIds() + gt_db = [] + for index in img_ids: + im_ann = coco.loadImgs(index)[0] + width = im_ann['width'] + height = im_ann['height'] + file_name = im_ann['file_name'] + im_id = int(im_ann["id"]) + + annIds = coco.getAnnIds(imgIds=index, iscrowd=False) + objs = coco.loadAnns(annIds) + + valid_objs = [] + for obj in objs: + x, y, w, h = obj['bbox'] + x1 = np.max((0, x)) + y1 = np.max((0, y)) + x2 = np.min((width - 1, x1 + np.max((0, w - 1)))) + y2 = np.min((height - 1, y1 + np.max((0, h - 1)))) + if obj['area'] > 0 and x2 >= x1 and y2 >= y1: + obj['clean_bbox'] = [x1, y1, x2 - x1, y2 - y1] + valid_objs.append(obj) + objs = valid_objs + + rec = [] + for obj in objs: + if max(obj['keypoints']) == 0: + continue + + joints = np.zeros( + (self.ann_info['num_joints'], 3), dtype=np.float) + joints_vis = np.zeros( + (self.ann_info['num_joints'], 3), dtype=np.float) + for ipt in range(self.ann_info['num_joints']): + joints[ipt, 0] = obj['keypoints'][ipt * 3 + 0] + joints[ipt, 1] = obj['keypoints'][ipt * 3 + 1] + joints[ipt, 2] = 0 + t_vis = obj['keypoints'][ipt * 3 + 2] + if t_vis > 1: + t_vis = 1 + joints_vis[ipt, 0] = t_vis + joints_vis[ipt, 1] = t_vis + joints_vis[ipt, 2] = 0 + + center, scale = self._box2cs(obj['clean_bbox'][:4]) + rec.append({ + 'image_file': os.path.join(self.img_prefix, file_name), + 'center': center, + 'scale': scale, + 'joints': joints, + 'joints_vis': joints_vis, + 'im_id': im_id, + }) + gt_db.extend(rec) + + return gt_db + + def _box2cs(self, box): + x, y, w, h = box[:4] + center = np.zeros((2), dtype=np.float32) + center[0] = x + w * 0.5 + center[1] = y + h * 0.5 + aspect_ratio = self.trainsize[0] * 1.0 / self.trainsize[1] + + if w > aspect_ratio * h: + h = w * 1.0 / aspect_ratio + elif w < aspect_ratio * h: + w = h * aspect_ratio + scale = np.array( + [w * 1.0 / self.pixel_std, h * 1.0 / self.pixel_std], + dtype=np.float32) + if center[0] != -1: + scale = scale * 1.25 + + return center, scale + + def _load_coco_person_detection_results(self): + all_boxes = None + bbox_file_path = os.path.join(self.dataset_dir, self.bbox_file) + with open(bbox_file_path, 'r') as f: + all_boxes = json.load(f) + + if not all_boxes: + print('=> Load %s fail!' % bbox_file_path) + return None + + kpt_db = [] + for n_img in range(0, len(all_boxes)): + det_res = all_boxes[n_img] + if det_res['category_id'] != 1: + continue + file_name = det_res[ + 'filename'] if 'filename' in det_res else '%012d.jpg' % det_res[ + 'image_id'] + img_name = os.path.join(self.img_prefix, file_name) + box = det_res['bbox'] + score = det_res['score'] + im_id = int(det_res['image_id']) + + if score < self.image_thre: + continue + + center, scale = self._box2cs(box) + joints = np.zeros((self.ann_info['num_joints'], 3), dtype=np.float) + joints_vis = np.ones( + (self.ann_info['num_joints'], 3), dtype=np.float) + kpt_db.append({ + 'image_file': img_name, + 'im_id': im_id, + 'center': center, + 'scale': scale, + 'score': score, + 'joints': joints, + 'joints_vis': joints_vis, + }) + + return kpt_db + + +@register +@serializable +class KeypointTopDownMPIIDataset(KeypointTopDownBaseDataset): + """MPII dataset for topdown pose estimation. Adapted from + https://github.com/leoxiaobin/deep-high-resolution-net.pytorch + Copyright (c) Microsoft, under the MIT License. + + The dataset loads raw features and apply specified transforms + to return a dict containing the image tensors and other information. + + MPII keypoint indexes:: + + 0: 'right_ankle', + 1: 'right_knee', + 2: 'right_hip', + 3: 'left_hip', + 4: 'left_knee', + 5: 'left_ankle', + 6: 'pelvis', + 7: 'thorax', + 8: 'upper_neck', + 9: 'head_top', + 10: 'right_wrist', + 11: 'right_elbow', + 12: 'right_shoulder', + 13: 'left_shoulder', + 14: 'left_elbow', + 15: 'left_wrist', + + Args: + dataset_dir (str): Root path to the dataset. + image_dir (str): Path to a directory where images are held. + anno_path (str): Relative path to the annotation file. + num_joints (int): Keypoint numbers + trainsize (list):[w, h] Image target size + transform (composed(operators)): A sequence of data transforms. + """ + + def __init__(self, + dataset_dir, + image_dir, + anno_path, + num_joints, + transform=[]): + super().__init__(dataset_dir, image_dir, anno_path, num_joints, + transform) + + self.dataset_name = 'mpii' + + def parse_dataset(self): + with open(self.get_anno()) as anno_file: + anno = json.load(anno_file) + + gt_db = [] + for a in anno: + image_name = a['image'] + im_id = a['image_id'] if 'image_id' in a else int( + os.path.splitext(image_name)[0]) + + c = np.array(a['center'], dtype=np.float) + s = np.array([a['scale'], a['scale']], dtype=np.float) + + # Adjust center/scale slightly to avoid cropping limbs + if c[0] != -1: + c[1] = c[1] + 15 * s[1] + s = s * 1.25 + c = c - 1 + + joints = np.zeros((self.ann_info['num_joints'], 3), dtype=np.float) + joints_vis = np.zeros( + (self.ann_info['num_joints'], 3), dtype=np.float) + if 'joints' in a: + joints_ = np.array(a['joints']) + joints_[:, 0:2] = joints_[:, 0:2] - 1 + joints_vis_ = np.array(a['joints_vis']) + assert len(joints_) == self.ann_info[ + 'num_joints'], 'joint num diff: {} vs {}'.format( + len(joints_), self.ann_info['num_joints']) + + joints[:, 0:2] = joints_[:, 0:2] + joints_vis[:, 0] = joints_vis_[:] + joints_vis[:, 1] = joints_vis_[:] + + gt_db.append({ + 'image_file': os.path.join(self.img_prefix, image_name), + 'im_id': im_id, + 'center': c, + 'scale': s, + 'joints': joints, + 'joints_vis': joints_vis + }) + print("number length: {}".format(len(gt_db))) + self.db = gt_db diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/mot.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/mot.py new file mode 100644 index 000000000..d46c02f52 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/mot.py @@ -0,0 +1,628 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import sys +import cv2 +import glob +import numpy as np +from collections import OrderedDict, defaultdict +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence +from .dataset import DetDataset, _make_dataset, _is_valid_file +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class MOTDataSet(DetDataset): + """ + Load dataset with MOT format, only support single class MOT. + + Args: + dataset_dir (str): root directory for dataset. + image_lists (str|list): mot data image lists, muiti-source mot dataset. + data_fields (list): key name of data dictionary, at least have 'image'. + sample_num (int): number of samples to load, -1 means all. + + Notes: + MOT datasets root directory following this: + dataset/mot + |鈥斺斺斺斺斺攊mage_lists + | |鈥斺斺斺斺斺攃altech.train + | |鈥斺斺斺斺斺攃altech.val + | |鈥斺斺斺斺斺攎ot16.train + | |鈥斺斺斺斺斺攎ot17.train + | ...... + |鈥斺斺斺斺斺擟altech + |鈥斺斺斺斺斺擬OT17 + |鈥斺斺斺斺斺...... + + All the MOT datasets have the following structure: + Caltech + |鈥斺斺斺斺斺攊mages + | 鈹斺斺斺斺斺斺00001.jpg + | |鈥斺斺斺斺斺 ... + | 鈹斺斺斺斺斺斺0000N.jpg + 鈹斺斺斺斺斺斺攍abels_with_ids + 鈹斺斺斺斺斺斺00001.txt + |鈥斺斺斺斺斺 ... + 鈹斺斺斺斺斺斺0000N.txt + or + + MOT17 + |鈥斺斺斺斺斺攊mages + | 鈹斺斺斺斺斺斺攖rain + | 鈹斺斺斺斺斺斺攖est + 鈹斺斺斺斺斺斺攍abels_with_ids + 鈹斺斺斺斺斺斺攖rain + """ + + def __init__(self, + dataset_dir=None, + image_lists=[], + data_fields=['image'], + sample_num=-1): + super(MOTDataSet, self).__init__( + dataset_dir=dataset_dir, + data_fields=data_fields, + sample_num=sample_num) + self.dataset_dir = dataset_dir + self.image_lists = image_lists + if isinstance(self.image_lists, str): + self.image_lists = [self.image_lists] + self.roidbs = None + self.cname2cid = None + + def get_anno(self): + if self.image_lists == []: + return + # only used to get categories and metric + # only check first data, but the label_list of all data should be same. + first_mot_data = self.image_lists[0].split('.')[0] + anno_file = os.path.join(self.dataset_dir, first_mot_data, 'label_list.txt') + return anno_file + + def parse_dataset(self): + self.img_files = OrderedDict() + self.img_start_index = OrderedDict() + self.label_files = OrderedDict() + self.tid_num = OrderedDict() + self.tid_start_index = OrderedDict() + + img_index = 0 + for data_name in self.image_lists: + # check every data image list + image_lists_dir = os.path.join(self.dataset_dir, 'image_lists') + assert os.path.isdir(image_lists_dir), \ + "The {} is not a directory.".format(image_lists_dir) + + list_path = os.path.join(image_lists_dir, data_name) + assert os.path.exists(list_path), \ + "The list path {} does not exist.".format(list_path) + + # record img_files, filter out empty ones + with open(list_path, 'r') as file: + self.img_files[data_name] = file.readlines() + self.img_files[data_name] = [ + os.path.join(self.dataset_dir, x.strip()) + for x in self.img_files[data_name] + ] + self.img_files[data_name] = list( + filter(lambda x: len(x) > 0, self.img_files[data_name])) + + self.img_start_index[data_name] = img_index + img_index += len(self.img_files[data_name]) + + # record label_files + self.label_files[data_name] = [ + x.replace('images', 'labels_with_ids').replace( + '.png', '.txt').replace('.jpg', '.txt') + for x in self.img_files[data_name] + ] + + for data_name, label_paths in self.label_files.items(): + max_index = -1 + for lp in label_paths: + lb = np.loadtxt(lp) + if len(lb) < 1: + continue + if len(lb.shape) < 2: + img_max = lb[1] + else: + img_max = np.max(lb[:, 1]) + if img_max > max_index: + max_index = img_max + self.tid_num[data_name] = int(max_index + 1) + + last_index = 0 + for i, (k, v) in enumerate(self.tid_num.items()): + self.tid_start_index[k] = last_index + last_index += v + + self.num_identities_dict = defaultdict(int) + self.num_identities_dict[0] = int(last_index + 1) # single class + self.num_imgs_each_data = [len(x) for x in self.img_files.values()] + self.total_imgs = sum(self.num_imgs_each_data) + + logger.info('MOT dataset summary: ') + logger.info(self.tid_num) + logger.info('Total images: {}'.format(self.total_imgs)) + logger.info('Image start index: {}'.format(self.img_start_index)) + logger.info('Total identities: {}'.format(self.num_identities_dict[0])) + logger.info('Identity start index: {}'.format(self.tid_start_index)) + + records = [] + cname2cid = mot_label() + + for img_index in range(self.total_imgs): + for i, (k, v) in enumerate(self.img_start_index.items()): + if img_index >= v: + data_name = list(self.label_files.keys())[i] + start_index = v + img_file = self.img_files[data_name][img_index - start_index] + lbl_file = self.label_files[data_name][img_index - start_index] + + if not os.path.exists(img_file): + logger.warning('Illegal image file: {}, and it will be ignored'. + format(img_file)) + continue + if not os.path.isfile(lbl_file): + logger.warning('Illegal label file: {}, and it will be ignored'. + format(lbl_file)) + continue + + labels = np.loadtxt(lbl_file, dtype=np.float32).reshape(-1, 6) + # each row in labels (N, 6) is [gt_class, gt_identity, cx, cy, w, h] + + cx, cy = labels[:, 2], labels[:, 3] + w, h = labels[:, 4], labels[:, 5] + gt_bbox = np.stack((cx, cy, w, h)).T.astype('float32') + gt_class = labels[:, 0:1].astype('int32') + gt_score = np.ones((len(labels), 1)).astype('float32') + gt_ide = labels[:, 1:2].astype('int32') + for i, _ in enumerate(gt_ide): + if gt_ide[i] > -1: + gt_ide[i] += self.tid_start_index[data_name] + + mot_rec = { + 'im_file': img_file, + 'im_id': img_index, + } if 'image' in self.data_fields else {} + + gt_rec = { + 'gt_class': gt_class, + 'gt_score': gt_score, + 'gt_bbox': gt_bbox, + 'gt_ide': gt_ide, + } + + for k, v in gt_rec.items(): + if k in self.data_fields: + mot_rec[k] = v + + records.append(mot_rec) + if self.sample_num > 0 and img_index >= self.sample_num: + break + assert len(records) > 0, 'not found any mot record in %s' % ( + self.image_lists) + self.roidbs, self.cname2cid = records, cname2cid + + +@register +@serializable +class MCMOTDataSet(DetDataset): + """ + Load dataset with MOT format, support multi-class MOT. + + Args: + dataset_dir (str): root directory for dataset. + image_lists (list(str)): mcmot data image lists, muiti-source mcmot dataset. + data_fields (list): key name of data dictionary, at least have 'image'. + label_list (str): if use_default_label is False, will load + mapping between category and class index. + sample_num (int): number of samples to load, -1 means all. + + Notes: + MCMOT datasets root directory following this: + dataset/mot + |鈥斺斺斺斺斺攊mage_lists + | |鈥斺斺斺斺斺攙isdrone_mcmot.train + | |鈥斺斺斺斺斺攙isdrone_mcmot.val + visdrone_mcmot + |鈥斺斺斺斺斺攊mages + | 鈹斺斺斺斺斺斺攖rain + | 鈹斺斺斺斺斺斺攙al + 鈹斺斺斺斺斺斺攍abels_with_ids + 鈹斺斺斺斺斺斺攖rain + """ + + def __init__(self, + dataset_dir=None, + image_lists=[], + data_fields=['image'], + label_list=None, + sample_num=-1): + super(MCMOTDataSet, self).__init__( + dataset_dir=dataset_dir, + data_fields=data_fields, + sample_num=sample_num) + self.dataset_dir = dataset_dir + self.image_lists = image_lists + if isinstance(self.image_lists, str): + self.image_lists = [self.image_lists] + self.label_list = label_list + self.roidbs = None + self.cname2cid = None + + def get_anno(self): + if self.image_lists == []: + return + # only used to get categories and metric + # only check first data, but the label_list of all data should be same. + first_mot_data = self.image_lists[0].split('.')[0] + anno_file = os.path.join(self.dataset_dir, first_mot_data, 'label_list.txt') + return anno_file + + def parse_dataset(self): + self.img_files = OrderedDict() + self.img_start_index = OrderedDict() + self.label_files = OrderedDict() + self.tid_num = OrderedDict() + self.tid_start_idx_of_cls_ids = defaultdict(dict) # for MCMOT + + img_index = 0 + for data_name in self.image_lists: + # check every data image list + image_lists_dir = os.path.join(self.dataset_dir, 'image_lists') + assert os.path.isdir(image_lists_dir), \ + "The {} is not a directory.".format(image_lists_dir) + + list_path = os.path.join(image_lists_dir, data_name) + assert os.path.exists(list_path), \ + "The list path {} does not exist.".format(list_path) + + # record img_files, filter out empty ones + with open(list_path, 'r') as file: + self.img_files[data_name] = file.readlines() + self.img_files[data_name] = [ + os.path.join(self.dataset_dir, x.strip()) + for x in self.img_files[data_name] + ] + self.img_files[data_name] = list( + filter(lambda x: len(x) > 0, self.img_files[data_name])) + + self.img_start_index[data_name] = img_index + img_index += len(self.img_files[data_name]) + + # record label_files + self.label_files[data_name] = [ + x.replace('images', 'labels_with_ids').replace( + '.png', '.txt').replace('.jpg', '.txt') + for x in self.img_files[data_name] + ] + + for data_name, label_paths in self.label_files.items(): + # using max_ids_dict rather than max_index + max_ids_dict = defaultdict(int) + for lp in label_paths: + lb = np.loadtxt(lp) + if len(lb) < 1: + continue + lb = lb.reshape(-1, 6) + for item in lb: + if item[1] > max_ids_dict[int(item[0])]: + # item[0]: cls_id + # item[1]: track id + max_ids_dict[int(item[0])] = int(item[1]) + # track id number + self.tid_num[data_name] = max_ids_dict + + last_idx_dict = defaultdict(int) + for i, (k, v) in enumerate(self.tid_num.items()): # each sub dataset + for cls_id, id_num in v.items(): # v is a max_ids_dict + self.tid_start_idx_of_cls_ids[k][cls_id] = last_idx_dict[cls_id] + last_idx_dict[cls_id] += id_num + + self.num_identities_dict = defaultdict(int) + for k, v in last_idx_dict.items(): + self.num_identities_dict[k] = int(v) # total ids of each category + + self.num_imgs_each_data = [len(x) for x in self.img_files.values()] + self.total_imgs = sum(self.num_imgs_each_data) + + # cname2cid and cid2cname + cname2cid = {} + if self.label_list is not None: + # if use label_list for multi source mix dataset, + # please make sure label_list in the first sub_dataset at least. + sub_dataset = self.image_lists[0].split('.')[0] + label_path = os.path.join(self.dataset_dir, sub_dataset, + self.label_list) + if not os.path.exists(label_path): + logger.info( + "Note: label_list {} does not exists, use VisDrone 10 classes labels as default.". + format(label_path)) + cname2cid = visdrone_mcmot_label() + else: + with open(label_path, 'r') as fr: + label_id = 0 + for line in fr.readlines(): + cname2cid[line.strip()] = label_id + label_id += 1 + else: + cname2cid = visdrone_mcmot_label() + + cid2cname = dict([(v, k) for (k, v) in cname2cid.items()]) + + logger.info('MCMOT dataset summary: ') + logger.info(self.tid_num) + logger.info('Total images: {}'.format(self.total_imgs)) + logger.info('Image start index: {}'.format(self.img_start_index)) + + logger.info('Total identities of each category: ') + num_identities_dict = sorted( + self.num_identities_dict.items(), key=lambda x: x[0]) + total_IDs_all_cats = 0 + for (k, v) in num_identities_dict: + logger.info('Category {} [{}] has {} IDs.'.format(k, cid2cname[k], + v)) + total_IDs_all_cats += v + logger.info('Total identities of all categories: {}'.format( + total_IDs_all_cats)) + + logger.info('Identity start index of each category: ') + for k, v in self.tid_start_idx_of_cls_ids.items(): + sorted_v = sorted(v.items(), key=lambda x: x[0]) + for (cls_id, start_idx) in sorted_v: + logger.info('Start index of dataset {} category {:d} is {:d}' + .format(k, cls_id, start_idx)) + + records = [] + for img_index in range(self.total_imgs): + for i, (k, v) in enumerate(self.img_start_index.items()): + if img_index >= v: + data_name = list(self.label_files.keys())[i] + start_index = v + img_file = self.img_files[data_name][img_index - start_index] + lbl_file = self.label_files[data_name][img_index - start_index] + + if not os.path.exists(img_file): + logger.warning('Illegal image file: {}, and it will be ignored'. + format(img_file)) + continue + if not os.path.isfile(lbl_file): + logger.warning('Illegal label file: {}, and it will be ignored'. + format(lbl_file)) + continue + + labels = np.loadtxt(lbl_file, dtype=np.float32).reshape(-1, 6) + # each row in labels (N, 6) is [gt_class, gt_identity, cx, cy, w, h] + + cx, cy = labels[:, 2], labels[:, 3] + w, h = labels[:, 4], labels[:, 5] + gt_bbox = np.stack((cx, cy, w, h)).T.astype('float32') + gt_class = labels[:, 0:1].astype('int32') + gt_score = np.ones((len(labels), 1)).astype('float32') + gt_ide = labels[:, 1:2].astype('int32') + for i, _ in enumerate(gt_ide): + if gt_ide[i] > -1: + cls_id = int(gt_class[i]) + start_idx = self.tid_start_idx_of_cls_ids[data_name][cls_id] + gt_ide[i] += start_idx + + mot_rec = { + 'im_file': img_file, + 'im_id': img_index, + } if 'image' in self.data_fields else {} + + gt_rec = { + 'gt_class': gt_class, + 'gt_score': gt_score, + 'gt_bbox': gt_bbox, + 'gt_ide': gt_ide, + } + + for k, v in gt_rec.items(): + if k in self.data_fields: + mot_rec[k] = v + + records.append(mot_rec) + if self.sample_num > 0 and img_index >= self.sample_num: + break + assert len(records) > 0, 'not found any mot record in %s' % ( + self.image_lists) + self.roidbs, self.cname2cid = records, cname2cid + + +@register +@serializable +class MOTImageFolder(DetDataset): + """ + Load MOT dataset with MOT format from image folder or video . + Args: + video_file (str): path of the video file, default ''. + frame_rate (int): frame rate of the video, use cv2 VideoCapture if not set. + dataset_dir (str): root directory for dataset. + keep_ori_im (bool): whether to keep original image, default False. + Set True when used during MOT model inference while saving + images or video, or used in DeepSORT. + """ + + def __init__(self, + video_file=None, + frame_rate=-1, + dataset_dir=None, + data_root=None, + image_dir=None, + sample_num=-1, + keep_ori_im=False, + **kwargs): + super(MOTImageFolder, self).__init__( + dataset_dir, image_dir, sample_num=sample_num) + self.video_file = video_file + self.data_root = data_root + self.keep_ori_im = keep_ori_im + self._imid2path = {} + self.roidbs = None + self.frame_rate = frame_rate + + def check_or_download_dataset(self): + return + + def parse_dataset(self, ): + if not self.roidbs: + if self.video_file is None: + self.frame_rate = 30 # set as default if infer image folder + self.roidbs = self._load_images() + else: + self.roidbs = self._load_video_images() + + def _load_video_images(self): + if self.frame_rate == -1: + # if frame_rate is not set for video, use cv2.VideoCapture + cap = cv2.VideoCapture(self.video_file) + self.frame_rate = int(cap.get(cv2.CAP_PROP_FPS)) + + extension = self.video_file.split('.')[-1] + output_path = self.video_file.replace('.{}'.format(extension), '') + frames_path = video2frames(self.video_file, output_path, + self.frame_rate) + self.video_frames = sorted( + glob.glob(os.path.join(frames_path, '*.png'))) + + self.video_length = len(self.video_frames) + logger.info('Length of the video: {:d} frames.'.format( + self.video_length)) + ct = 0 + records = [] + for image in self.video_frames: + assert image != '' and os.path.isfile(image), \ + "Image {} not found".format(image) + if self.sample_num > 0 and ct >= self.sample_num: + break + rec = {'im_id': np.array([ct]), 'im_file': image} + if self.keep_ori_im: + rec.update({'keep_ori_im': 1}) + self._imid2path[ct] = image + ct += 1 + records.append(rec) + assert len(records) > 0, "No image file found" + return records + + def _find_images(self): + image_dir = self.image_dir + if not isinstance(image_dir, Sequence): + image_dir = [image_dir] + images = [] + for im_dir in image_dir: + if os.path.isdir(im_dir): + im_dir = os.path.join(self.dataset_dir, im_dir) + images.extend(_make_dataset(im_dir)) + elif os.path.isfile(im_dir) and _is_valid_file(im_dir): + images.append(im_dir) + return images + + def _load_images(self): + images = self._find_images() + ct = 0 + records = [] + for image in images: + assert image != '' and os.path.isfile(image), \ + "Image {} not found".format(image) + if self.sample_num > 0 and ct >= self.sample_num: + break + rec = {'im_id': np.array([ct]), 'im_file': image} + if self.keep_ori_im: + rec.update({'keep_ori_im': 1}) + self._imid2path[ct] = image + ct += 1 + records.append(rec) + assert len(records) > 0, "No image file found" + return records + + def get_imid2path(self): + return self._imid2path + + def set_images(self, images): + self.image_dir = images + self.roidbs = self._load_images() + + def set_video(self, video_file, frame_rate): + # update video_file and frame_rate by command line of tools/infer_mot.py + self.video_file = video_file + self.frame_rate = frame_rate + assert os.path.isfile(self.video_file) and _is_valid_video(self.video_file), \ + "wrong or unsupported file format: {}".format(self.video_file) + self.roidbs = self._load_video_images() + + +def _is_valid_video(f, extensions=('.mp4', '.avi', '.mov', '.rmvb', 'flv')): + return f.lower().endswith(extensions) + + +def video2frames(video_path, outpath, frame_rate, **kargs): + def _dict2str(kargs): + cmd_str = '' + for k, v in kargs.items(): + cmd_str += (' ' + str(k) + ' ' + str(v)) + return cmd_str + + ffmpeg = ['ffmpeg ', ' -y -loglevel ', ' error '] + vid_name = os.path.basename(video_path).split('.')[0] + out_full_path = os.path.join(outpath, vid_name) + + if not os.path.exists(out_full_path): + os.makedirs(out_full_path) + + # video file name + outformat = os.path.join(out_full_path, '%08d.png') + + cmd = ffmpeg + cmd = ffmpeg + [ + ' -i ', video_path, ' -r ', str(frame_rate), ' -f image2 ', outformat + ] + cmd = ''.join(cmd) + _dict2str(kargs) + + if os.system(cmd) != 0: + raise RuntimeError('ffmpeg process video: {} error'.format(video_path)) + sys.exit(-1) + + sys.stdout.flush() + return out_full_path + + +def mot_label(): + labels_map = {'person': 0} + return labels_map + + +def visdrone_mcmot_label(): + labels_map = { + 'pedestrian': 0, + 'people': 1, + 'bicycle': 2, + 'car': 3, + 'van': 4, + 'truck': 5, + 'tricycle': 6, + 'awning-tricycle': 7, + 'bus': 8, + 'motor': 9, + } + return labels_map diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/sniper_coco.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/sniper_coco.py new file mode 100644 index 000000000..1b07e7a31 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/sniper_coco.py @@ -0,0 +1,194 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import cv2 +import json +import copy +import numpy as np + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence + +from ppdet.core.workspace import register, serializable +from ppdet.data.crop_utils.annotation_cropper import AnnoCropper +from .coco import COCODataSet +from .dataset import _make_dataset, _is_valid_file +from ppdet.utils.logger import setup_logger + +logger = setup_logger('sniper_coco_dataset') + + +@register +@serializable +class SniperCOCODataSet(COCODataSet): + """SniperCOCODataSet""" + + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + proposals_file=None, + data_fields=['image'], + sample_num=-1, + load_crowd=False, + allow_empty=True, + empty_ratio=1., + is_trainset=True, + image_target_sizes=[2000, 1000], + valid_box_ratio_ranges=[[-1, 0.1],[0.08, -1]], + chip_target_size=500, + chip_target_stride=200, + use_neg_chip=False, + max_neg_num_per_im=8, + max_per_img=-1, + nms_thresh=0.5): + super(SniperCOCODataSet, self).__init__( + dataset_dir=dataset_dir, + image_dir=image_dir, + anno_path=anno_path, + data_fields=data_fields, + sample_num=sample_num, + load_crowd=load_crowd, + allow_empty=allow_empty, + empty_ratio=empty_ratio + ) + self.proposals_file = proposals_file + self.proposals = None + self.anno_cropper = None + self.is_trainset = is_trainset + self.image_target_sizes = image_target_sizes + self.valid_box_ratio_ranges = valid_box_ratio_ranges + self.chip_target_size = chip_target_size + self.chip_target_stride = chip_target_stride + self.use_neg_chip = use_neg_chip + self.max_neg_num_per_im = max_neg_num_per_im + self.max_per_img = max_per_img + self.nms_thresh = nms_thresh + + + def parse_dataset(self): + if not hasattr(self, "roidbs"): + super(SniperCOCODataSet, self).parse_dataset() + if self.is_trainset: + self._parse_proposals() + self._merge_anno_proposals() + self.ori_roidbs = copy.deepcopy(self.roidbs) + self.init_anno_cropper() + self.roidbs = self.generate_chips_roidbs(self.roidbs, self.is_trainset) + + def set_proposals_file(self, file_path): + self.proposals_file = file_path + + def init_anno_cropper(self): + logger.info("Init AnnoCropper...") + self.anno_cropper = AnnoCropper( + image_target_sizes=self.image_target_sizes, + valid_box_ratio_ranges=self.valid_box_ratio_ranges, + chip_target_size=self.chip_target_size, + chip_target_stride=self.chip_target_stride, + use_neg_chip=self.use_neg_chip, + max_neg_num_per_im=self.max_neg_num_per_im, + max_per_img=self.max_per_img, + nms_thresh=self.nms_thresh + ) + + def generate_chips_roidbs(self, roidbs, is_trainset): + if is_trainset: + roidbs = self.anno_cropper.crop_anno_records(roidbs) + else: + roidbs = self.anno_cropper.crop_infer_anno_records(roidbs) + return roidbs + + def _parse_proposals(self): + if self.proposals_file: + self.proposals = {} + logger.info("Parse proposals file:{}".format(self.proposals_file)) + with open(self.proposals_file, 'r') as f: + proposals = json.load(f) + for prop in proposals: + image_id = prop["image_id"] + if image_id not in self.proposals: + self.proposals[image_id] = [] + x, y, w, h = prop["bbox"] + self.proposals[image_id].append([x, y, x + w, y + h]) + + def _merge_anno_proposals(self): + assert self.roidbs + if self.proposals and len(self.proposals.keys()) > 0: + logger.info("merge proposals to annos") + for id, record in enumerate(self.roidbs): + image_id = int(record["im_id"]) + if image_id not in self.proposals.keys(): + logger.info("image id :{} no proposals".format(image_id)) + record["proposals"] = np.array(self.proposals.get(image_id, []), dtype=np.float32) + self.roidbs[id] = record + + def get_ori_roidbs(self): + if not hasattr(self, "ori_roidbs"): + return None + return self.ori_roidbs + + def get_roidbs(self): + if not hasattr(self, "roidbs"): + self.parse_dataset() + return self.roidbs + + def set_roidbs(self, roidbs): + self.roidbs = roidbs + + def check_or_download_dataset(self): + return + + def _parse(self): + image_dir = self.image_dir + if not isinstance(image_dir, Sequence): + image_dir = [image_dir] + images = [] + for im_dir in image_dir: + if os.path.isdir(im_dir): + im_dir = os.path.join(self.dataset_dir, im_dir) + images.extend(_make_dataset(im_dir)) + elif os.path.isfile(im_dir) and _is_valid_file(im_dir): + images.append(im_dir) + return images + + def _load_images(self): + images = self._parse() + ct = 0 + records = [] + for image in images: + assert image != '' and os.path.isfile(image), \ + "Image {} not found".format(image) + if self.sample_num > 0 and ct >= self.sample_num: + break + im = cv2.imread(image) + h, w, c = im.shape + rec = {'im_id': np.array([ct]), 'im_file': image, "h": h, "w": w} + self._imid2path[ct] = image + ct += 1 + records.append(rec) + assert len(records) > 0, "No image file found" + return records + + def get_imid2path(self): + return self._imid2path + + def set_images(self, images): + self._imid2path = {} + self.image_dir = images + self.roidbs = self._load_images() + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/voc.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/voc.py new file mode 100644 index 000000000..1c2a7ef98 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/voc.py @@ -0,0 +1,231 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import numpy as np + +import xml.etree.ElementTree as ET + +from ppdet.core.workspace import register, serializable + +from .dataset import DetDataset + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class VOCDataSet(DetDataset): + """ + Load dataset with PascalVOC format. + + Notes: + `anno_path` must contains xml file and image file path for annotations. + + Args: + dataset_dir (str): root directory for dataset. + image_dir (str): directory for images. + anno_path (str): voc annotation file path. + data_fields (list): key name of data dictionary, at least have 'image'. + sample_num (int): number of samples to load, -1 means all. + label_list (str): if use_default_label is False, will load + mapping between category and class index. + allow_empty (bool): whether to load empty entry. False as default + empty_ratio (float): the ratio of empty record number to total + record's, if empty_ratio is out of [0. ,1.), do not sample the + records and use all the empty entries. 1. as default + """ + + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + data_fields=['image'], + sample_num=-1, + label_list=None, + allow_empty=False, + empty_ratio=1.): + super(VOCDataSet, self).__init__( + dataset_dir=dataset_dir, + image_dir=image_dir, + anno_path=anno_path, + data_fields=data_fields, + sample_num=sample_num) + self.label_list = label_list + self.allow_empty = allow_empty + self.empty_ratio = empty_ratio + + def _sample_empty(self, records, num): + # if empty_ratio is out of [0. ,1.), do not sample the records + if self.empty_ratio < 0. or self.empty_ratio >= 1.: + return records + import random + sample_num = min( + int(num * self.empty_ratio / (1 - self.empty_ratio)), len(records)) + records = random.sample(records, sample_num) + return records + + def parse_dataset(self, ): + anno_path = os.path.join(self.dataset_dir, self.anno_path) + image_dir = os.path.join(self.dataset_dir, self.image_dir) + + # mapping category name to class id + # first_class:0, second_class:1, ... + records = [] + empty_records = [] + ct = 0 + cname2cid = {} + if self.label_list: + label_path = os.path.join(self.dataset_dir, self.label_list) + if not os.path.exists(label_path): + raise ValueError("label_list {} does not exists".format( + label_path)) + with open(label_path, 'r') as fr: + label_id = 0 + for line in fr.readlines(): + cname2cid[line.strip()] = label_id + label_id += 1 + else: + cname2cid = pascalvoc_label() + + with open(anno_path, 'r') as fr: + while True: + line = fr.readline() + if not line: + break + img_file, xml_file = [os.path.join(image_dir, x) \ + for x in line.strip().split()[:2]] + if not os.path.exists(img_file): + logger.warning( + 'Illegal image file: {}, and it will be ignored'.format( + img_file)) + continue + if not os.path.isfile(xml_file): + logger.warning( + 'Illegal xml file: {}, and it will be ignored'.format( + xml_file)) + continue + tree = ET.parse(xml_file) + if tree.find('id') is None: + im_id = np.array([ct]) + else: + im_id = np.array([int(tree.find('id').text)]) + + objs = tree.findall('object') + im_w = float(tree.find('size').find('width').text) + im_h = float(tree.find('size').find('height').text) + if im_w < 0 or im_h < 0: + logger.warning( + 'Illegal width: {} or height: {} in annotation, ' + 'and {} will be ignored'.format(im_w, im_h, xml_file)) + continue + + num_bbox, i = len(objs), 0 + gt_bbox = np.zeros((num_bbox, 4), dtype=np.float32) + gt_class = np.zeros((num_bbox, 1), dtype=np.int32) + gt_score = np.zeros((num_bbox, 1), dtype=np.float32) + difficult = np.zeros((num_bbox, 1), dtype=np.int32) + for obj in objs: + cname = obj.find('name').text + + # user dataset may not contain difficult field + _difficult = obj.find('difficult') + _difficult = int( + _difficult.text) if _difficult is not None else 0 + + x1 = float(obj.find('bndbox').find('xmin').text) + y1 = float(obj.find('bndbox').find('ymin').text) + x2 = float(obj.find('bndbox').find('xmax').text) + y2 = float(obj.find('bndbox').find('ymax').text) + x1 = max(0, x1) + y1 = max(0, y1) + x2 = min(im_w - 1, x2) + y2 = min(im_h - 1, y2) + if x2 > x1 and y2 > y1: + gt_bbox[i, :] = [x1, y1, x2, y2] + gt_class[i, 0] = cname2cid[cname] + gt_score[i, 0] = 1. + difficult[i, 0] = _difficult + i += 1 + else: + logger.warning( + 'Found an invalid bbox in annotations: xml_file: {}' + ', x1: {}, y1: {}, x2: {}, y2: {}.'.format( + xml_file, x1, y1, x2, y2)) + gt_bbox = gt_bbox[:i, :] + gt_class = gt_class[:i, :] + gt_score = gt_score[:i, :] + difficult = difficult[:i, :] + + voc_rec = { + 'im_file': img_file, + 'im_id': im_id, + 'h': im_h, + 'w': im_w + } if 'image' in self.data_fields else {} + + gt_rec = { + 'gt_class': gt_class, + 'gt_score': gt_score, + 'gt_bbox': gt_bbox, + 'difficult': difficult + } + for k, v in gt_rec.items(): + if k in self.data_fields: + voc_rec[k] = v + + if len(objs) == 0: + empty_records.append(voc_rec) + else: + records.append(voc_rec) + + ct += 1 + if self.sample_num > 0 and ct >= self.sample_num: + break + assert ct > 0, 'not found any voc record in %s' % (self.anno_path) + logger.debug('{} samples in file {}'.format(ct, anno_path)) + if self.allow_empty and len(empty_records) > 0: + empty_records = self._sample_empty(empty_records, len(records)) + records += empty_records + self.roidbs, self.cname2cid = records, cname2cid + + def get_label_list(self): + return os.path.join(self.dataset_dir, self.label_list) + + +def pascalvoc_label(): + labels_map = { + 'aeroplane': 0, + 'bicycle': 1, + 'bird': 2, + 'boat': 3, + 'bottle': 4, + 'bus': 5, + 'car': 6, + 'cat': 7, + 'chair': 8, + 'cow': 9, + 'diningtable': 10, + 'dog': 11, + 'horse': 12, + 'motorbike': 13, + 'person': 14, + 'pottedplant': 15, + 'sheep': 16, + 'sofa': 17, + 'train': 18, + 'tvmonitor': 19 + } + return labels_map diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/widerface.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/widerface.py new file mode 100644 index 000000000..a17c2aaf8 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/source/widerface.py @@ -0,0 +1,180 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import numpy as np + +from ppdet.core.workspace import register, serializable +from .dataset import DetDataset + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class WIDERFaceDataSet(DetDataset): + """ + Load WiderFace records with 'anno_path' + + Args: + dataset_dir (str): root directory for dataset. + image_dir (str): directory for images. + anno_path (str): WiderFace annotation data. + data_fields (list): key name of data dictionary, at least have 'image'. + sample_num (int): number of samples to load, -1 means all. + with_lmk (bool): whether to load face landmark keypoint labels. + """ + + def __init__(self, + dataset_dir=None, + image_dir=None, + anno_path=None, + data_fields=['image'], + sample_num=-1, + with_lmk=False): + super(WIDERFaceDataSet, self).__init__( + dataset_dir=dataset_dir, + image_dir=image_dir, + anno_path=anno_path, + data_fields=data_fields, + sample_num=sample_num, + with_lmk=with_lmk) + self.anno_path = anno_path + self.sample_num = sample_num + self.roidbs = None + self.cname2cid = None + self.with_lmk = with_lmk + + def parse_dataset(self): + anno_path = os.path.join(self.dataset_dir, self.anno_path) + image_dir = os.path.join(self.dataset_dir, self.image_dir) + + txt_file = anno_path + + records = [] + ct = 0 + file_lists = self._load_file_list(txt_file) + cname2cid = widerface_label() + + for item in file_lists: + im_fname = item[0] + im_id = np.array([ct]) + gt_bbox = np.zeros((len(item) - 1, 4), dtype=np.float32) + gt_class = np.zeros((len(item) - 1, 1), dtype=np.int32) + gt_lmk_labels = np.zeros((len(item) - 1, 10), dtype=np.float32) + lmk_ignore_flag = np.zeros((len(item) - 1, 1), dtype=np.int32) + for index_box in range(len(item)): + if index_box < 1: + continue + gt_bbox[index_box - 1] = item[index_box][0] + if self.with_lmk: + gt_lmk_labels[index_box - 1] = item[index_box][1] + lmk_ignore_flag[index_box - 1] = item[index_box][2] + im_fname = os.path.join(image_dir, + im_fname) if image_dir else im_fname + widerface_rec = { + 'im_file': im_fname, + 'im_id': im_id, + } if 'image' in self.data_fields else {} + gt_rec = { + 'gt_bbox': gt_bbox, + 'gt_class': gt_class, + } + for k, v in gt_rec.items(): + if k in self.data_fields: + widerface_rec[k] = v + if self.with_lmk: + widerface_rec['gt_keypoint'] = gt_lmk_labels + widerface_rec['keypoint_ignore'] = lmk_ignore_flag + + if len(item) != 0: + records.append(widerface_rec) + + ct += 1 + if self.sample_num > 0 and ct >= self.sample_num: + break + assert len(records) > 0, 'not found any widerface in %s' % (anno_path) + logger.debug('{} samples in file {}'.format(ct, anno_path)) + self.roidbs, self.cname2cid = records, cname2cid + + def _load_file_list(self, input_txt): + with open(input_txt, 'r') as f_dir: + lines_input_txt = f_dir.readlines() + + file_dict = {} + num_class = 0 + exts = ['jpg', 'jpeg', 'png', 'bmp'] + exts += [ext.upper() for ext in exts] + for i in range(len(lines_input_txt)): + line_txt = lines_input_txt[i].strip('\n\t\r') + split_str = line_txt.split(' ') + if len(split_str) == 1: + img_file_name = os.path.split(split_str[0])[1] + split_txt = img_file_name.split('.') + if len(split_txt) < 2: + continue + elif split_txt[-1] in exts: + if i != 0: + num_class += 1 + file_dict[num_class] = [line_txt] + else: + if len(line_txt) <= 6: + continue + result_boxs = [] + xmin = float(split_str[0]) + ymin = float(split_str[1]) + w = float(split_str[2]) + h = float(split_str[3]) + # Filter out wrong labels + if w < 0 or h < 0: + logger.warning('Illegal box with w: {}, h: {} in ' + 'img: {}, and it will be ignored'.format( + w, h, file_dict[num_class][0])) + continue + xmin = max(0, xmin) + ymin = max(0, ymin) + xmax = xmin + w + ymax = ymin + h + gt_bbox = [xmin, ymin, xmax, ymax] + result_boxs.append(gt_bbox) + if self.with_lmk: + assert len(split_str) > 18, 'When `with_lmk=True`, the number' \ + 'of characters per line in the annotation file should' \ + 'exceed 18.' + lmk0_x = float(split_str[5]) + lmk0_y = float(split_str[6]) + lmk1_x = float(split_str[8]) + lmk1_y = float(split_str[9]) + lmk2_x = float(split_str[11]) + lmk2_y = float(split_str[12]) + lmk3_x = float(split_str[14]) + lmk3_y = float(split_str[15]) + lmk4_x = float(split_str[17]) + lmk4_y = float(split_str[18]) + lmk_ignore_flag = 0 if lmk0_x == -1 else 1 + gt_lmk_label = [ + lmk0_x, lmk0_y, lmk1_x, lmk1_y, lmk2_x, lmk2_y, lmk3_x, + lmk3_y, lmk4_x, lmk4_y + ] + result_boxs.append(gt_lmk_label) + result_boxs.append(lmk_ignore_flag) + file_dict[num_class].append(result_boxs) + + return list(file_dict.values()) + + +def widerface_label(): + labels_map = {'face': 0} + return labels_map diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__init__.py new file mode 100644 index 000000000..fb8a1a449 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__init__.py @@ -0,0 +1,28 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import operators +from . import batch_operators +from . import keypoint_operators +from . import mot_operators + +from .operators import * +from .batch_operators import * +from .keypoint_operators import * +from .mot_operators import * + +__all__ = [] +__all__ += registered_ops +__all__ += keypoint_operators.__all__ +__all__ += mot_operators.__all__ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..ada30eb03 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/atss_assigner.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/atss_assigner.cpython-37.pyc new file mode 100644 index 000000000..937f1cda9 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/atss_assigner.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/batch_operators.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/batch_operators.cpython-37.pyc new file mode 100644 index 000000000..06f9ac59e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/batch_operators.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/keypoint_operators.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/keypoint_operators.cpython-37.pyc new file mode 100644 index 000000000..54cc72243 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/keypoint_operators.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/mot_operators.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/mot_operators.cpython-37.pyc new file mode 100644 index 000000000..0de9ed2c8 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/mot_operators.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/op_helper.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/op_helper.cpython-37.pyc new file mode 100644 index 000000000..a2570b09e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/op_helper.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/operators.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/operators.cpython-37.pyc new file mode 100644 index 000000000..8f1e42647 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/__pycache__/operators.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/atss_assigner.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/atss_assigner.py new file mode 100644 index 000000000..178d94fb6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/atss_assigner.py @@ -0,0 +1,269 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/core/bbox/assigners/atss_assigner.py + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import numpy as np +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +def bbox_overlaps(bboxes1, bboxes2, mode='iou', is_aligned=False, eps=1e-6): + """Calculate overlap between two set of bboxes. + If ``is_aligned `` is ``False``, then calculate the overlaps between each + bbox of bboxes1 and bboxes2, otherwise the overlaps between each aligned + pair of bboxes1 and bboxes2. + Args: + bboxes1 (Tensor): shape (B, m, 4) in format or empty. + bboxes2 (Tensor): shape (B, n, 4) in format or empty. + B indicates the batch dim, in shape (B1, B2, ..., Bn). + If ``is_aligned `` is ``True``, then m and n must be equal. + mode (str): "iou" (intersection over union) or "iof" (intersection over + foreground). + is_aligned (bool, optional): If True, then m and n must be equal. + Default False. + eps (float, optional): A value added to the denominator for numerical + stability. Default 1e-6. + Returns: + Tensor: shape (m, n) if ``is_aligned `` is False else shape (m,) + """ + assert mode in ['iou', 'iof', 'giou'], 'Unsupported mode {}'.format(mode) + # Either the boxes are empty or the length of boxes's last dimenstion is 4 + assert (bboxes1.shape[-1] == 4 or bboxes1.shape[0] == 0) + assert (bboxes2.shape[-1] == 4 or bboxes2.shape[0] == 0) + + # Batch dim must be the same + # Batch dim: (B1, B2, ... Bn) + assert bboxes1.shape[:-2] == bboxes2.shape[:-2] + batch_shape = bboxes1.shape[:-2] + + rows = bboxes1.shape[-2] if bboxes1.shape[0] > 0 else 0 + cols = bboxes2.shape[-2] if bboxes2.shape[0] > 0 else 0 + if is_aligned: + assert rows == cols + + if rows * cols == 0: + if is_aligned: + return np.random.random(batch_shape + (rows, )) + else: + return np.random.random(batch_shape + (rows, cols)) + + area1 = (bboxes1[..., 2] - bboxes1[..., 0]) * ( + bboxes1[..., 3] - bboxes1[..., 1]) + area2 = (bboxes2[..., 2] - bboxes2[..., 0]) * ( + bboxes2[..., 3] - bboxes2[..., 1]) + + if is_aligned: + lt = np.maximum(bboxes1[..., :2], bboxes2[..., :2]) # [B, rows, 2] + rb = np.minimum(bboxes1[..., 2:], bboxes2[..., 2:]) # [B, rows, 2] + + wh = (rb - lt).clip(min=0) # [B, rows, 2] + overlap = wh[..., 0] * wh[..., 1] + + if mode in ['iou', 'giou']: + union = area1 + area2 - overlap + else: + union = area1 + if mode == 'giou': + enclosed_lt = np.minimum(bboxes1[..., :2], bboxes2[..., :2]) + enclosed_rb = np.maximum(bboxes1[..., 2:], bboxes2[..., 2:]) + else: + lt = np.maximum(bboxes1[..., :, None, :2], + bboxes2[..., None, :, :2]) # [B, rows, cols, 2] + rb = np.minimum(bboxes1[..., :, None, 2:], + bboxes2[..., None, :, 2:]) # [B, rows, cols, 2] + + wh = (rb - lt).clip(min=0) # [B, rows, cols, 2] + overlap = wh[..., 0] * wh[..., 1] + + if mode in ['iou', 'giou']: + union = area1[..., None] + area2[..., None, :] - overlap + else: + union = area1[..., None] + if mode == 'giou': + enclosed_lt = np.minimum(bboxes1[..., :, None, :2], + bboxes2[..., None, :, :2]) + enclosed_rb = np.maximum(bboxes1[..., :, None, 2:], + bboxes2[..., None, :, 2:]) + + eps = np.array([eps]) + union = np.maximum(union, eps) + ious = overlap / union + if mode in ['iou', 'iof']: + return ious + # calculate gious + enclose_wh = (enclosed_rb - enclosed_lt).clip(min=0) + enclose_area = enclose_wh[..., 0] * enclose_wh[..., 1] + enclose_area = np.maximum(enclose_area, eps) + gious = ious - (enclose_area - union) / enclose_area + return gious + + +def topk_(input, k, axis=1, largest=True): + x = -input if largest else input + if axis == 0: + row_index = np.arange(input.shape[1 - axis]) + topk_index = np.argpartition(x, k, axis=axis)[0:k, :] + topk_data = x[topk_index, row_index] + + topk_index_sort = np.argsort(topk_data, axis=axis) + topk_data_sort = topk_data[topk_index_sort, row_index] + topk_index_sort = topk_index[0:k, :][topk_index_sort, row_index] + else: + column_index = np.arange(x.shape[1 - axis])[:, None] + topk_index = np.argpartition(x, k, axis=axis)[:, 0:k] + topk_data = x[column_index, topk_index] + topk_data = -topk_data if largest else topk_data + topk_index_sort = np.argsort(topk_data, axis=axis) + topk_data_sort = topk_data[column_index, topk_index_sort] + topk_index_sort = topk_index[:, 0:k][column_index, topk_index_sort] + + return topk_data_sort, topk_index_sort + + +class ATSSAssigner(object): + """Assign a corresponding gt bbox or background to each bbox. + + Each proposals will be assigned with `0` or a positive integer + indicating the ground truth index. + + - 0: negative sample, no assigned gt + - positive integer: positive sample, index (1-based) of assigned gt + + Args: + topk (float): number of bbox selected in each level + """ + + def __init__(self, topk=9): + self.topk = topk + + def __call__(self, + bboxes, + num_level_bboxes, + gt_bboxes, + gt_bboxes_ignore=None, + gt_labels=None): + """Assign gt to bboxes. + The assignment is done in following steps + 1. compute iou between all bbox (bbox of all pyramid levels) and gt + 2. compute center distance between all bbox and gt + 3. on each pyramid level, for each gt, select k bbox whose center + are closest to the gt center, so we total select k*l bbox as + candidates for each gt + 4. get corresponding iou for the these candidates, and compute the + mean and std, set mean + std as the iou threshold + 5. select these candidates whose iou are greater than or equal to + the threshold as postive + 6. limit the positive sample's center in gt + Args: + bboxes (np.array): Bounding boxes to be assigned, shape(n, 4). + num_level_bboxes (List): num of bboxes in each level + gt_bboxes (np.array): Groundtruth boxes, shape (k, 4). + gt_bboxes_ignore (np.array, optional): Ground truth bboxes that are + labelled as `ignored`, e.g., crowd boxes in COCO. + gt_labels (np.array, optional): Label of gt_bboxes, shape (k, ). + """ + bboxes = bboxes[:, :4] + num_gt, num_bboxes = gt_bboxes.shape[0], bboxes.shape[0] + + # assign 0 by default + assigned_gt_inds = np.zeros((num_bboxes, ), dtype=np.int64) + + if num_gt == 0 or num_bboxes == 0: + # No ground truth or boxes, return empty assignment + max_overlaps = np.zeros((num_bboxes, )) + if num_gt == 0: + # No truth, assign everything to background + assigned_gt_inds[:] = 0 + if not np.any(gt_labels): + assigned_labels = None + else: + assigned_labels = -np.ones((num_bboxes, ), dtype=np.int64) + return assigned_gt_inds, max_overlaps + + # compute iou between all bbox and gt + overlaps = bbox_overlaps(bboxes, gt_bboxes) + # compute center distance between all bbox and gt + gt_cx = (gt_bboxes[:, 0] + gt_bboxes[:, 2]) / 2.0 + gt_cy = (gt_bboxes[:, 1] + gt_bboxes[:, 3]) / 2.0 + gt_points = np.stack((gt_cx, gt_cy), axis=1) + + bboxes_cx = (bboxes[:, 0] + bboxes[:, 2]) / 2.0 + bboxes_cy = (bboxes[:, 1] + bboxes[:, 3]) / 2.0 + bboxes_points = np.stack((bboxes_cx, bboxes_cy), axis=1) + + distances = np.sqrt( + np.power((bboxes_points[:, None, :] - gt_points[None, :, :]), 2) + .sum(-1)) + + # Selecting candidates based on the center distance + candidate_idxs = [] + start_idx = 0 + for bboxes_per_level in num_level_bboxes: + # on each pyramid level, for each gt, + # select k bbox whose center are closest to the gt center + end_idx = start_idx + bboxes_per_level + distances_per_level = distances[start_idx:end_idx, :] + selectable_k = min(self.topk, bboxes_per_level) + _, topk_idxs_per_level = topk_( + distances_per_level, selectable_k, axis=0, largest=False) + candidate_idxs.append(topk_idxs_per_level + start_idx) + start_idx = end_idx + candidate_idxs = np.concatenate(candidate_idxs, axis=0) + + # get corresponding iou for the these candidates, and compute the + # mean and std, set mean + std as the iou threshold + candidate_overlaps = overlaps[candidate_idxs, np.arange(num_gt)] + overlaps_mean_per_gt = candidate_overlaps.mean(0) + overlaps_std_per_gt = candidate_overlaps.std(0) + overlaps_thr_per_gt = overlaps_mean_per_gt + overlaps_std_per_gt + + is_pos = candidate_overlaps >= overlaps_thr_per_gt[None, :] + + # limit the positive sample's center in gt + for gt_idx in range(num_gt): + candidate_idxs[:, gt_idx] += gt_idx * num_bboxes + ep_bboxes_cx = np.broadcast_to( + bboxes_cx.reshape(1, -1), [num_gt, num_bboxes]).reshape(-1) + ep_bboxes_cy = np.broadcast_to( + bboxes_cy.reshape(1, -1), [num_gt, num_bboxes]).reshape(-1) + candidate_idxs = candidate_idxs.reshape(-1) + + # calculate the left, top, right, bottom distance between positive + # bbox center and gt side + l_ = ep_bboxes_cx[candidate_idxs].reshape(-1, num_gt) - gt_bboxes[:, 0] + t_ = ep_bboxes_cy[candidate_idxs].reshape(-1, num_gt) - gt_bboxes[:, 1] + r_ = gt_bboxes[:, 2] - ep_bboxes_cx[candidate_idxs].reshape(-1, num_gt) + b_ = gt_bboxes[:, 3] - ep_bboxes_cy[candidate_idxs].reshape(-1, num_gt) + is_in_gts = np.stack([l_, t_, r_, b_], axis=1).min(axis=1) > 0.01 + is_pos = is_pos & is_in_gts + + # if an anchor box is assigned to multiple gts, + # the one with the highest IoU will be selected. + overlaps_inf = -np.inf * np.ones_like(overlaps).T.reshape(-1) + index = candidate_idxs.reshape(-1)[is_pos.reshape(-1)] + overlaps_inf[index] = overlaps.T.reshape(-1)[index] + overlaps_inf = overlaps_inf.reshape(num_gt, -1).T + + max_overlaps = overlaps_inf.max(axis=1) + argmax_overlaps = overlaps_inf.argmax(axis=1) + assigned_gt_inds[max_overlaps != + -np.inf] = argmax_overlaps[max_overlaps != -np.inf] + 1 + + return assigned_gt_inds, max_overlaps diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/autoaugment_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/autoaugment_utils.py new file mode 100644 index 000000000..cfa89d374 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/autoaugment_utils.py @@ -0,0 +1,1586 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# Reference: +# https://github.com/tensorflow/tpu/blob/master/models/official/detection/utils/autoaugment_utils.py +"""AutoAugment util file.""" + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import inspect +import math +from PIL import Image, ImageEnhance +import numpy as np +import cv2 +from copy import deepcopy + +# This signifies the max integer that the controller RNN could predict for the +# augmentation scheme. +_MAX_LEVEL = 10. + +# Represents an invalid bounding box that is used for checking for padding +# lists of bounding box coordinates for a few augmentation operations +_INVALID_BOX = [[-1.0, -1.0, -1.0, -1.0]] + + +def policy_v0(): + """Autoaugment policy that was used in AutoAugment Detection Paper.""" + # Each tuple is an augmentation operation of the form + # (operation, probability, magnitude). Each element in policy is a + # sub-policy that will be applied sequentially on the image. + policy = [ + [('TranslateX_BBox', 0.6, 4), ('Equalize', 0.8, 10)], + [('TranslateY_Only_BBoxes', 0.2, 2), ('Cutout', 0.8, 8)], + [('Sharpness', 0.0, 8), ('ShearX_BBox', 0.4, 0)], + [('ShearY_BBox', 1.0, 2), ('TranslateY_Only_BBoxes', 0.6, 6)], + [('Rotate_BBox', 0.6, 10), ('Color', 1.0, 6)], + ] + return policy + + +def policy_v1(): + """Autoaugment policy that was used in AutoAugment Detection Paper.""" + # Each tuple is an augmentation operation of the form + # (operation, probability, magnitude). Each element in policy is a + # sub-policy that will be applied sequentially on the image. + policy = [ + [('TranslateX_BBox', 0.6, 4), ('Equalize', 0.8, 10)], + [('TranslateY_Only_BBoxes', 0.2, 2), ('Cutout', 0.8, 8)], + [('Sharpness', 0.0, 8), ('ShearX_BBox', 0.4, 0)], + [('ShearY_BBox', 1.0, 2), ('TranslateY_Only_BBoxes', 0.6, 6)], + [('Rotate_BBox', 0.6, 10), ('Color', 1.0, 6)], + [('Color', 0.0, 0), ('ShearX_Only_BBoxes', 0.8, 4)], + [('ShearY_Only_BBoxes', 0.8, 2), ('Flip_Only_BBoxes', 0.0, 10)], + [('Equalize', 0.6, 10), ('TranslateX_BBox', 0.2, 2)], + [('Color', 1.0, 10), ('TranslateY_Only_BBoxes', 0.4, 6)], + [('Rotate_BBox', 0.8, 10), ('Contrast', 0.0, 10)], # , + [('Cutout', 0.2, 2), ('Brightness', 0.8, 10)], + [('Color', 1.0, 6), ('Equalize', 1.0, 2)], + [('Cutout_Only_BBoxes', 0.4, 6), ('TranslateY_Only_BBoxes', 0.8, 2)], + [('Color', 0.2, 8), ('Rotate_BBox', 0.8, 10)], + [('Sharpness', 0.4, 4), ('TranslateY_Only_BBoxes', 0.0, 4)], + [('Sharpness', 1.0, 4), ('SolarizeAdd', 0.4, 4)], + [('Rotate_BBox', 1.0, 8), ('Sharpness', 0.2, 8)], + [('ShearY_BBox', 0.6, 10), ('Equalize_Only_BBoxes', 0.6, 8)], + [('ShearX_BBox', 0.2, 6), ('TranslateY_Only_BBoxes', 0.2, 10)], + [('SolarizeAdd', 0.6, 8), ('Brightness', 0.8, 10)], + ] + return policy + + +def policy_vtest(): + """Autoaugment test policy for debugging.""" + # Each tuple is an augmentation operation of the form + # (operation, probability, magnitude). Each element in policy is a + # sub-policy that will be applied sequentially on the image. + policy = [[('TranslateX_BBox', 1.0, 4), ('Equalize', 1.0, 10)], ] + return policy + + +def policy_v2(): + """Additional policy that performs well on object detection.""" + # Each tuple is an augmentation operation of the form + # (operation, probability, magnitude). Each element in policy is a + # sub-policy that will be applied sequentially on the image. + policy = [ + [('Color', 0.0, 6), ('Cutout', 0.6, 8), ('Sharpness', 0.4, 8)], + [('Rotate_BBox', 0.4, 8), ('Sharpness', 0.4, 2), + ('Rotate_BBox', 0.8, 10)], + [('TranslateY_BBox', 1.0, 8), ('AutoContrast', 0.8, 2)], + [('AutoContrast', 0.4, 6), ('ShearX_BBox', 0.8, 8), + ('Brightness', 0.0, 10)], + [('SolarizeAdd', 0.2, 6), ('Contrast', 0.0, 10), + ('AutoContrast', 0.6, 0)], + [('Cutout', 0.2, 0), ('Solarize', 0.8, 8), ('Color', 1.0, 4)], + [('TranslateY_BBox', 0.0, 4), ('Equalize', 0.6, 8), + ('Solarize', 0.0, 10)], + [('TranslateY_BBox', 0.2, 2), ('ShearY_BBox', 0.8, 8), + ('Rotate_BBox', 0.8, 8)], + [('Cutout', 0.8, 8), ('Brightness', 0.8, 8), ('Cutout', 0.2, 2)], + [('Color', 0.8, 4), ('TranslateY_BBox', 1.0, 6), + ('Rotate_BBox', 0.6, 6)], + [('Rotate_BBox', 0.6, 10), ('BBox_Cutout', 1.0, 4), ('Cutout', 0.2, 8)], + [('Rotate_BBox', 0.0, 0), ('Equalize', 0.6, 6), + ('ShearY_BBox', 0.6, 8)], + [('Brightness', 0.8, 8), ('AutoContrast', 0.4, 2), + ('Brightness', 0.2, 2)], + [('TranslateY_BBox', 0.4, 8), ('Solarize', 0.4, 6), + ('SolarizeAdd', 0.2, 10)], + [('Contrast', 1.0, 10), ('SolarizeAdd', 0.2, 8), ('Equalize', 0.2, 4)], + ] + return policy + + +def policy_v3(): + """"Additional policy that performs well on object detection.""" + # Each tuple is an augmentation operation of the form + # (operation, probability, magnitude). Each element in policy is a + # sub-policy that will be applied sequentially on the image. + policy = [ + [('Posterize', 0.8, 2), ('TranslateX_BBox', 1.0, 8)], + [('BBox_Cutout', 0.2, 10), ('Sharpness', 1.0, 8)], + [('Rotate_BBox', 0.6, 8), ('Rotate_BBox', 0.8, 10)], + [('Equalize', 0.8, 10), ('AutoContrast', 0.2, 10)], + [('SolarizeAdd', 0.2, 2), ('TranslateY_BBox', 0.2, 8)], + [('Sharpness', 0.0, 2), ('Color', 0.4, 8)], + [('Equalize', 1.0, 8), ('TranslateY_BBox', 1.0, 8)], + [('Posterize', 0.6, 2), ('Rotate_BBox', 0.0, 10)], + [('AutoContrast', 0.6, 0), ('Rotate_BBox', 1.0, 6)], + [('Equalize', 0.0, 4), ('Cutout', 0.8, 10)], + [('Brightness', 1.0, 2), ('TranslateY_BBox', 1.0, 6)], + [('Contrast', 0.0, 2), ('ShearY_BBox', 0.8, 0)], + [('AutoContrast', 0.8, 10), ('Contrast', 0.2, 10)], + [('Rotate_BBox', 1.0, 10), ('Cutout', 1.0, 10)], + [('SolarizeAdd', 0.8, 6), ('Equalize', 0.8, 8)], + ] + return policy + + +def _equal(val1, val2, eps=1e-8): + return abs(val1 - val2) <= eps + + +def blend(image1, image2, factor): + """Blend image1 and image2 using 'factor'. + + Factor can be above 0.0. A value of 0.0 means only image1 is used. + A value of 1.0 means only image2 is used. A value between 0.0 and + 1.0 means we linearly interpolate the pixel values between the two + images. A value greater than 1.0 "extrapolates" the difference + between the two pixel values, and we clip the results to values + between 0 and 255. + + Args: + image1: An image Tensor of type uint8. + image2: An image Tensor of type uint8. + factor: A floating point value above 0.0. + + Returns: + A blended image Tensor of type uint8. + """ + if factor == 0.0: + return image1 + if factor == 1.0: + return image2 + + image1 = image1.astype(np.float32) + image2 = image2.astype(np.float32) + + difference = image2 - image1 + scaled = factor * difference + + # Do addition in float. + temp = image1 + scaled + + # Interpolate + if factor > 0.0 and factor < 1.0: + # Interpolation means we always stay within 0 and 255. + return temp.astype(np.uint8) + + # Extrapolate: + # + # We need to clip and then cast. + return np.clip(temp, a_min=0, a_max=255).astype(np.uint8) + + +def cutout(image, pad_size, replace=0): + """Apply cutout (https://arxiv.org/abs/1708.04552) to image. + + This operation applies a (2*pad_size x 2*pad_size) mask of zeros to + a random location within `img`. The pixel values filled in will be of the + value `replace`. The located where the mask will be applied is randomly + chosen uniformly over the whole image. + + Args: + image: An image Tensor of type uint8. + pad_size: Specifies how big the zero mask that will be generated is that + is applied to the image. The mask will be of size + (2*pad_size x 2*pad_size). + replace: What pixel value to fill in the image in the area that has + the cutout mask applied to it. + + Returns: + An image Tensor that is of type uint8. + Example: + img = cv2.imread( "/home/vis/gry/train/img_data/test.jpg", cv2.COLOR_BGR2RGB ) + new_img = cutout(img, pad_size=50, replace=0) + """ + image_height, image_width = image.shape[0], image.shape[1] + + cutout_center_height = np.random.randint(low=0, high=image_height) + cutout_center_width = np.random.randint(low=0, high=image_width) + + lower_pad = np.maximum(0, cutout_center_height - pad_size) + upper_pad = np.maximum(0, image_height - cutout_center_height - pad_size) + left_pad = np.maximum(0, cutout_center_width - pad_size) + right_pad = np.maximum(0, image_width - cutout_center_width - pad_size) + + cutout_shape = [ + image_height - (lower_pad + upper_pad), + image_width - (left_pad + right_pad) + ] + padding_dims = [[lower_pad, upper_pad], [left_pad, right_pad]] + mask = np.pad(np.zeros( + cutout_shape, dtype=image.dtype), + padding_dims, + 'constant', + constant_values=1) + mask = np.expand_dims(mask, -1) + mask = np.tile(mask, [1, 1, 3]) + image = np.where( + np.equal(mask, 0), + np.ones_like( + image, dtype=image.dtype) * replace, + image) + return image.astype(np.uint8) + + +def solarize(image, threshold=128): + # For each pixel in the image, select the pixel + # if the value is less than the threshold. + # Otherwise, subtract 255 from the pixel. + return np.where(image < threshold, image, 255 - image) + + +def solarize_add(image, addition=0, threshold=128): + # For each pixel in the image less than threshold + # we add 'addition' amount to it and then clip the + # pixel value to be between 0 and 255. The value + # of 'addition' is between -128 and 128. + added_image = image.astype(np.int64) + addition + added_image = np.clip(added_image, a_min=0, a_max=255).astype(np.uint8) + return np.where(image < threshold, added_image, image) + + +def color(image, factor): + """use cv2 to deal""" + gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) + degenerate = cv2.cvtColor(gray, cv2.COLOR_GRAY2BGR) + return blend(degenerate, image, factor) + + +# refer to https://github.com/4uiiurz1/pytorch-auto-augment/blob/024b2eac4140c38df8342f09998e307234cafc80/auto_augment.py#L197 +def contrast(img, factor): + img = ImageEnhance.Contrast(Image.fromarray(img)).enhance(factor) + return np.array(img) + + +def brightness(image, factor): + """Equivalent of PIL Brightness.""" + degenerate = np.zeros_like(image) + return blend(degenerate, image, factor) + + +def posterize(image, bits): + """Equivalent of PIL Posterize.""" + shift = 8 - bits + return np.left_shift(np.right_shift(image, shift), shift) + + +def rotate(image, degrees, replace): + """Rotates the image by degrees either clockwise or counterclockwise. + + Args: + image: An image Tensor of type uint8. + degrees: Float, a scalar angle in degrees to rotate all images by. If + degrees is positive the image will be rotated clockwise otherwise it will + be rotated counterclockwise. + replace: A one or three value 1D tensor to fill empty pixels caused by + the rotate operation. + + Returns: + The rotated version of image. + """ + image = wrap(image) + image = Image.fromarray(image) + image = image.rotate(degrees) + image = np.array(image, dtype=np.uint8) + return unwrap(image, replace) + + +def random_shift_bbox(image, + bbox, + pixel_scaling, + replace, + new_min_bbox_coords=None): + """Move the bbox and the image content to a slightly new random location. + + Args: + image: 3D uint8 Tensor. + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + The potential values for the new min corner of the bbox will be between + [old_min - pixel_scaling * bbox_height/2, + old_min - pixel_scaling * bbox_height/2]. + pixel_scaling: A float between 0 and 1 that specifies the pixel range + that the new bbox location will be sampled from. + replace: A one or three value 1D tensor to fill empty pixels. + new_min_bbox_coords: If not None, then this is a tuple that specifies the + (min_y, min_x) coordinates of the new bbox. Normally this is randomly + specified, but this allows it to be manually set. The coordinates are + the absolute coordinates between 0 and image height/width and are int32. + + Returns: + The new image that will have the shifted bbox location in it along with + the new bbox that contains the new coordinates. + """ + # Obtains image height and width and create helper clip functions. + image_height, image_width = image.shape[0], image.shape[1] + image_height = float(image_height) + image_width = float(image_width) + + def clip_y(val): + return np.clip(val, a_min=0, a_max=image_height - 1).astype(np.int32) + + def clip_x(val): + return np.clip(val, a_min=0, a_max=image_width - 1).astype(np.int32) + + # Convert bbox to pixel coordinates. + min_y = int(image_height * bbox[0]) + min_x = int(image_width * bbox[1]) + max_y = clip_y(image_height * bbox[2]) + max_x = clip_x(image_width * bbox[3]) + + bbox_height, bbox_width = (max_y - min_y + 1, max_x - min_x + 1) + image_height = int(image_height) + image_width = int(image_width) + + # Select the new min/max bbox ranges that are used for sampling the + # new min x/y coordinates of the shifted bbox. + minval_y = clip_y(min_y - np.int32(pixel_scaling * float(bbox_height) / + 2.0)) + maxval_y = clip_y(min_y + np.int32(pixel_scaling * float(bbox_height) / + 2.0)) + minval_x = clip_x(min_x - np.int32(pixel_scaling * float(bbox_width) / 2.0)) + maxval_x = clip_x(min_x + np.int32(pixel_scaling * float(bbox_width) / 2.0)) + + # Sample and calculate the new unclipped min/max coordinates of the new bbox. + if new_min_bbox_coords is None: + unclipped_new_min_y = np.random.randint( + low=minval_y, high=maxval_y, dtype=np.int32) + unclipped_new_min_x = np.random.randint( + low=minval_x, high=maxval_x, dtype=np.int32) + else: + unclipped_new_min_y, unclipped_new_min_x = ( + clip_y(new_min_bbox_coords[0]), clip_x(new_min_bbox_coords[1])) + unclipped_new_max_y = unclipped_new_min_y + bbox_height - 1 + unclipped_new_max_x = unclipped_new_min_x + bbox_width - 1 + + # Determine if any of the new bbox was shifted outside the current image. + # This is used for determining if any of the original bbox content should be + # discarded. + new_min_y, new_min_x, new_max_y, new_max_x = ( + clip_y(unclipped_new_min_y), clip_x(unclipped_new_min_x), + clip_y(unclipped_new_max_y), clip_x(unclipped_new_max_x)) + shifted_min_y = (new_min_y - unclipped_new_min_y) + min_y + shifted_max_y = max_y - (unclipped_new_max_y - new_max_y) + shifted_min_x = (new_min_x - unclipped_new_min_x) + min_x + shifted_max_x = max_x - (unclipped_new_max_x - new_max_x) + + # Create the new bbox tensor by converting pixel integer values to floats. + new_bbox = np.stack([ + float(new_min_y) / float(image_height), float(new_min_x) / + float(image_width), float(new_max_y) / float(image_height), + float(new_max_x) / float(image_width) + ]) + + # Copy the contents in the bbox and fill the old bbox location + # with gray (128). + bbox_content = image[shifted_min_y:shifted_max_y + 1, shifted_min_x: + shifted_max_x + 1, :] + + def mask_and_add_image(min_y_, min_x_, max_y_, max_x_, mask, content_tensor, + image_): + """Applies mask to bbox region in image then adds content_tensor to it.""" + mask = np.pad(mask, [[min_y_, (image_height - 1) - max_y_], + [min_x_, (image_width - 1) - max_x_], [0, 0]], + 'constant', + constant_values=1) + + content_tensor = np.pad(content_tensor, + [[min_y_, (image_height - 1) - max_y_], + [min_x_, (image_width - 1) - max_x_], [0, 0]], + 'constant', + constant_values=0) + return image_ * mask + content_tensor + + # Zero out original bbox location. + mask = np.zeros_like(image)[min_y:max_y + 1, min_x:max_x + 1, :] + grey_tensor = np.zeros_like(mask) + replace[0] + image = mask_and_add_image(min_y, min_x, max_y, max_x, mask, grey_tensor, + image) + + # Fill in bbox content to new bbox location. + mask = np.zeros_like(bbox_content) + image = mask_and_add_image(new_min_y, new_min_x, new_max_y, new_max_x, mask, + bbox_content, image) + + return image.astype(np.uint8), new_bbox + + +def _clip_bbox(min_y, min_x, max_y, max_x): + """Clip bounding box coordinates between 0 and 1. + + Args: + min_y: Normalized bbox coordinate of type float between 0 and 1. + min_x: Normalized bbox coordinate of type float between 0 and 1. + max_y: Normalized bbox coordinate of type float between 0 and 1. + max_x: Normalized bbox coordinate of type float between 0 and 1. + + Returns: + Clipped coordinate values between 0 and 1. + """ + min_y = np.clip(min_y, a_min=0, a_max=1.0) + min_x = np.clip(min_x, a_min=0, a_max=1.0) + max_y = np.clip(max_y, a_min=0, a_max=1.0) + max_x = np.clip(max_x, a_min=0, a_max=1.0) + return min_y, min_x, max_y, max_x + + +def _check_bbox_area(min_y, min_x, max_y, max_x, delta=0.05): + """Adjusts bbox coordinates to make sure the area is > 0. + + Args: + min_y: Normalized bbox coordinate of type float between 0 and 1. + min_x: Normalized bbox coordinate of type float between 0 and 1. + max_y: Normalized bbox coordinate of type float between 0 and 1. + max_x: Normalized bbox coordinate of type float between 0 and 1. + delta: Float, this is used to create a gap of size 2 * delta between + bbox min/max coordinates that are the same on the boundary. + This prevents the bbox from having an area of zero. + + Returns: + Tuple of new bbox coordinates between 0 and 1 that will now have a + guaranteed area > 0. + """ + height = max_y - min_y + width = max_x - min_x + + def _adjust_bbox_boundaries(min_coord, max_coord): + # Make sure max is never 0 and min is never 1. + max_coord = np.maximum(max_coord, 0.0 + delta) + min_coord = np.minimum(min_coord, 1.0 - delta) + return min_coord, max_coord + + if _equal(height, 0): + min_y, max_y = _adjust_bbox_boundaries(min_y, max_y) + + if _equal(width, 0): + min_x, max_x = _adjust_bbox_boundaries(min_x, max_x) + + return min_y, min_x, max_y, max_x + + +def _scale_bbox_only_op_probability(prob): + """Reduce the probability of the bbox-only operation. + + Probability is reduced so that we do not distort the content of too many + bounding boxes that are close to each other. The value of 3.0 was a chosen + hyper parameter when designing the autoaugment algorithm that we found + empirically to work well. + + Args: + prob: Float that is the probability of applying the bbox-only operation. + + Returns: + Reduced probability. + """ + return prob / 3.0 + + +def _apply_bbox_augmentation(image, bbox, augmentation_func, *args): + """Applies augmentation_func to the subsection of image indicated by bbox. + + Args: + image: 3D uint8 Tensor. + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + augmentation_func: Augmentation function that will be applied to the + subsection of image. + *args: Additional parameters that will be passed into augmentation_func + when it is called. + + Returns: + A modified version of image, where the bbox location in the image will + have `ugmentation_func applied to it. + """ + image_height = image.shape[0] + image_width = image.shape[1] + + min_y = int(image_height * bbox[0]) + min_x = int(image_width * bbox[1]) + max_y = int(image_height * bbox[2]) + max_x = int(image_width * bbox[3]) + + # Clip to be sure the max values do not fall out of range. + max_y = np.minimum(max_y, image_height - 1) + max_x = np.minimum(max_x, image_width - 1) + + # Get the sub-tensor that is the image within the bounding box region. + bbox_content = image[min_y:max_y + 1, min_x:max_x + 1, :] + + # Apply the augmentation function to the bbox portion of the image. + augmented_bbox_content = augmentation_func(bbox_content, *args) + + # Pad the augmented_bbox_content and the mask to match the shape of original + # image. + augmented_bbox_content = np.pad( + augmented_bbox_content, [[min_y, (image_height - 1) - max_y], + [min_x, (image_width - 1) - max_x], [0, 0]], + 'constant', + constant_values=1) + + # Create a mask that will be used to zero out a part of the original image. + mask_tensor = np.zeros_like(bbox_content) + + mask_tensor = np.pad(mask_tensor, + [[min_y, (image_height - 1) - max_y], + [min_x, (image_width - 1) - max_x], [0, 0]], + 'constant', + constant_values=1) + # Replace the old bbox content with the new augmented content. + image = image * mask_tensor + augmented_bbox_content + return image.astype(np.uint8) + + +def _concat_bbox(bbox, bboxes): + """Helper function that concates bbox to bboxes along the first dimension.""" + + # Note if all elements in bboxes are -1 (_INVALID_BOX), then this means + # we discard bboxes and start the bboxes Tensor with the current bbox. + bboxes_sum_check = np.sum(bboxes) + bbox = np.expand_dims(bbox, 0) + # This check will be true when it is an _INVALID_BOX + if _equal(bboxes_sum_check, -4): + bboxes = bbox + else: + bboxes = np.concatenate([bboxes, bbox], 0) + return bboxes + + +def _apply_bbox_augmentation_wrapper(image, bbox, new_bboxes, prob, + augmentation_func, func_changes_bbox, + *args): + """Applies _apply_bbox_augmentation with probability prob. + + Args: + image: 3D uint8 Tensor. + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + new_bboxes: 2D Tensor that is a list of the bboxes in the image after they + have been altered by aug_func. These will only be changed when + func_changes_bbox is set to true. Each bbox has 4 elements + (min_y, min_x, max_y, max_x) of type float that are the normalized + bbox coordinates between 0 and 1. + prob: Float that is the probability of applying _apply_bbox_augmentation. + augmentation_func: Augmentation function that will be applied to the + subsection of image. + func_changes_bbox: Boolean. Does augmentation_func return bbox in addition + to image. + *args: Additional parameters that will be passed into augmentation_func + when it is called. + + Returns: + A tuple. Fist element is a modified version of image, where the bbox + location in the image will have augmentation_func applied to it if it is + chosen to be called with probability `prob`. The second element is a + Tensor of Tensors of length 4 that will contain the altered bbox after + applying augmentation_func. + """ + should_apply_op = (np.random.rand() + prob >= 1) + if func_changes_bbox: + if should_apply_op: + augmented_image, bbox = augmentation_func(image, bbox, *args) + else: + augmented_image, bbox = (image, bbox) + else: + if should_apply_op: + augmented_image = _apply_bbox_augmentation(image, bbox, + augmentation_func, *args) + else: + augmented_image = image + new_bboxes = _concat_bbox(bbox, new_bboxes) + return augmented_image.astype(np.uint8), new_bboxes + + +def _apply_multi_bbox_augmentation(image, bboxes, prob, aug_func, + func_changes_bbox, *args): + """Applies aug_func to the image for each bbox in bboxes. + + Args: + image: 3D uint8 Tensor. + bboxes: 2D Tensor that is a list of the bboxes in the image. Each bbox + has 4 elements (min_y, min_x, max_y, max_x) of type float. + prob: Float that is the probability of applying aug_func to a specific + bounding box within the image. + aug_func: Augmentation function that will be applied to the + subsections of image indicated by the bbox values in bboxes. + func_changes_bbox: Boolean. Does augmentation_func return bbox in addition + to image. + *args: Additional parameters that will be passed into augmentation_func + when it is called. + + Returns: + A modified version of image, where each bbox location in the image will + have augmentation_func applied to it if it is chosen to be called with + probability prob independently across all bboxes. Also the final + bboxes are returned that will be unchanged if func_changes_bbox is set to + false and if true, the new altered ones will be returned. + """ + # Will keep track of the new altered bboxes after aug_func is repeatedly + # applied. The -1 values are a dummy value and this first Tensor will be + # removed upon appending the first real bbox. + new_bboxes = np.array(_INVALID_BOX) + + # If the bboxes are empty, then just give it _INVALID_BOX. The result + # will be thrown away. + bboxes = np.array((_INVALID_BOX)) if bboxes.size == 0 else bboxes + + assert bboxes.shape[1] == 4, "bboxes.shape[1] must be 4!!!!" + + # pylint:disable=g-long-lambda + # pylint:disable=line-too-long + wrapped_aug_func = lambda _image, bbox, _new_bboxes: _apply_bbox_augmentation_wrapper(_image, bbox, _new_bboxes, prob, aug_func, func_changes_bbox, *args) + # pylint:enable=g-long-lambda + # pylint:enable=line-too-long + + # Setup the while_loop. + num_bboxes = bboxes.shape[0] # We loop until we go over all bboxes. + idx = 0 # Counter for the while loop. + + # Conditional function when to end the loop once we go over all bboxes + # images_and_bboxes contain (_image, _new_bboxes) + def cond(_idx, _images_and_bboxes): + return _idx < num_bboxes + + # Shuffle the bboxes so that the augmentation order is not deterministic if + # we are not changing the bboxes with aug_func. + # if not func_changes_bbox: + # print(bboxes) + # loop_bboxes = np.take(bboxes,np.random.permutation(bboxes.shape[0]),axis=0) + # print(loop_bboxes) + # else: + # loop_bboxes = bboxes + # we can not shuffle the bbox because it does not contain class information here + loop_bboxes = deepcopy(bboxes) + + # Main function of while_loop where we repeatedly apply augmentation on the + # bboxes in the image. + # pylint:disable=g-long-lambda + body = lambda _idx, _images_and_bboxes: [ + _idx + 1, wrapped_aug_func(_images_and_bboxes[0], + loop_bboxes[_idx], + _images_and_bboxes[1])] + while (cond(idx, (image, new_bboxes))): + idx, (image, new_bboxes) = body(idx, (image, new_bboxes)) + + # Either return the altered bboxes or the original ones depending on if + # we altered them in anyway. + if func_changes_bbox: + final_bboxes = new_bboxes + else: + final_bboxes = bboxes + return image, final_bboxes + + +def _apply_multi_bbox_augmentation_wrapper(image, bboxes, prob, aug_func, + func_changes_bbox, *args): + """Checks to be sure num bboxes > 0 before calling inner function.""" + num_bboxes = len(bboxes) + new_image = deepcopy(image) + new_bboxes = deepcopy(bboxes) + if num_bboxes != 0: + new_image, new_bboxes = _apply_multi_bbox_augmentation( + new_image, new_bboxes, prob, aug_func, func_changes_bbox, *args) + return new_image, new_bboxes + + +def rotate_only_bboxes(image, bboxes, prob, degrees, replace): + """Apply rotate to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, rotate, func_changes_bbox, degrees, replace) + + +def shear_x_only_bboxes(image, bboxes, prob, level, replace): + """Apply shear_x to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, shear_x, func_changes_bbox, level, replace) + + +def shear_y_only_bboxes(image, bboxes, prob, level, replace): + """Apply shear_y to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, shear_y, func_changes_bbox, level, replace) + + +def translate_x_only_bboxes(image, bboxes, prob, pixels, replace): + """Apply translate_x to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, translate_x, func_changes_bbox, pixels, replace) + + +def translate_y_only_bboxes(image, bboxes, prob, pixels, replace): + """Apply translate_y to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, translate_y, func_changes_bbox, pixels, replace) + + +def flip_only_bboxes(image, bboxes, prob): + """Apply flip_lr to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper(image, bboxes, prob, + np.fliplr, func_changes_bbox) + + +def solarize_only_bboxes(image, bboxes, prob, threshold): + """Apply solarize to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper(image, bboxes, prob, solarize, + func_changes_bbox, threshold) + + +def equalize_only_bboxes(image, bboxes, prob): + """Apply equalize to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper(image, bboxes, prob, equalize, + func_changes_bbox) + + +def cutout_only_bboxes(image, bboxes, prob, pad_size, replace): + """Apply cutout to each bbox in the image with probability prob.""" + func_changes_bbox = False + prob = _scale_bbox_only_op_probability(prob) + return _apply_multi_bbox_augmentation_wrapper( + image, bboxes, prob, cutout, func_changes_bbox, pad_size, replace) + + +def _rotate_bbox(bbox, image_height, image_width, degrees): + """Rotates the bbox coordinated by degrees. + + Args: + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + image_height: Int, height of the image. + image_width: Int, height of the image. + degrees: Float, a scalar angle in degrees to rotate all images by. If + degrees is positive the image will be rotated clockwise otherwise it will + be rotated counterclockwise. + + Returns: + A tensor of the same shape as bbox, but now with the rotated coordinates. + """ + image_height, image_width = (float(image_height), float(image_width)) + + # Convert from degrees to radians. + degrees_to_radians = math.pi / 180.0 + radians = degrees * degrees_to_radians + + # Translate the bbox to the center of the image and turn the normalized 0-1 + # coordinates to absolute pixel locations. + # Y coordinates are made negative as the y axis of images goes down with + # increasing pixel values, so we negate to make sure x axis and y axis points + # are in the traditionally positive direction. + min_y = -int(image_height * (bbox[0] - 0.5)) + min_x = int(image_width * (bbox[1] - 0.5)) + max_y = -int(image_height * (bbox[2] - 0.5)) + max_x = int(image_width * (bbox[3] - 0.5)) + coordinates = np.stack([[min_y, min_x], [min_y, max_x], [max_y, min_x], + [max_y, max_x]]).astype(np.float32) + # Rotate the coordinates according to the rotation matrix clockwise if + # radians is positive, else negative + rotation_matrix = np.stack([[math.cos(radians), math.sin(radians)], + [-math.sin(radians), math.cos(radians)]]) + new_coords = np.matmul(rotation_matrix, + np.transpose(coordinates)).astype(np.int32) + + # Find min/max values and convert them back to normalized 0-1 floats. + min_y = -(float(np.max(new_coords[0, :])) / image_height - 0.5) + min_x = float(np.min(new_coords[1, :])) / image_width + 0.5 + max_y = -(float(np.min(new_coords[0, :])) / image_height - 0.5) + max_x = float(np.max(new_coords[1, :])) / image_width + 0.5 + + # Clip the bboxes to be sure the fall between [0, 1]. + min_y, min_x, max_y, max_x = _clip_bbox(min_y, min_x, max_y, max_x) + min_y, min_x, max_y, max_x = _check_bbox_area(min_y, min_x, max_y, max_x) + return np.stack([min_y, min_x, max_y, max_x]) + + +def rotate_with_bboxes(image, bboxes, degrees, replace): + # Rotate the image. + image = rotate(image, degrees, replace) + + # Convert bbox coordinates to pixel values. + image_height, image_width = image.shape[:2] + # pylint:disable=g-long-lambda + wrapped_rotate_bbox = lambda bbox: _rotate_bbox(bbox, image_height, image_width, degrees) + # pylint:enable=g-long-lambda + new_bboxes = np.zeros_like(bboxes) + for idx in range(len(bboxes)): + new_bboxes[idx] = wrapped_rotate_bbox(bboxes[idx]) + return image, new_bboxes + + +def translate_x(image, pixels, replace): + """Equivalent of PIL Translate in X dimension.""" + image = Image.fromarray(wrap(image)) + image = image.transform(image.size, Image.AFFINE, (1, 0, pixels, 0, 1, 0)) + return unwrap(np.array(image), replace) + + +def translate_y(image, pixels, replace): + """Equivalent of PIL Translate in Y dimension.""" + image = Image.fromarray(wrap(image)) + image = image.transform(image.size, Image.AFFINE, (1, 0, 0, 0, 1, pixels)) + return unwrap(np.array(image), replace) + + +def _shift_bbox(bbox, image_height, image_width, pixels, shift_horizontal): + """Shifts the bbox coordinates by pixels. + + Args: + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + image_height: Int, height of the image. + image_width: Int, width of the image. + pixels: An int. How many pixels to shift the bbox. + shift_horizontal: Boolean. If true then shift in X dimension else shift in + Y dimension. + + Returns: + A tensor of the same shape as bbox, but now with the shifted coordinates. + """ + pixels = int(pixels) + # Convert bbox to integer pixel locations. + min_y = int(float(image_height) * bbox[0]) + min_x = int(float(image_width) * bbox[1]) + max_y = int(float(image_height) * bbox[2]) + max_x = int(float(image_width) * bbox[3]) + + if shift_horizontal: + min_x = np.maximum(0, min_x - pixels) + max_x = np.minimum(image_width, max_x - pixels) + else: + min_y = np.maximum(0, min_y - pixels) + max_y = np.minimum(image_height, max_y - pixels) + + # Convert bbox back to floats. + min_y = float(min_y) / float(image_height) + min_x = float(min_x) / float(image_width) + max_y = float(max_y) / float(image_height) + max_x = float(max_x) / float(image_width) + + # Clip the bboxes to be sure the fall between [0, 1]. + min_y, min_x, max_y, max_x = _clip_bbox(min_y, min_x, max_y, max_x) + min_y, min_x, max_y, max_x = _check_bbox_area(min_y, min_x, max_y, max_x) + return np.stack([min_y, min_x, max_y, max_x]) + + +def translate_bbox(image, bboxes, pixels, replace, shift_horizontal): + """Equivalent of PIL Translate in X/Y dimension that shifts image and bbox. + + Args: + image: 3D uint8 Tensor. + bboxes: 2D Tensor that is a list of the bboxes in the image. Each bbox + has 4 elements (min_y, min_x, max_y, max_x) of type float with values + between [0, 1]. + pixels: An int. How many pixels to shift the image and bboxes + replace: A one or three value 1D tensor to fill empty pixels. + shift_horizontal: Boolean. If true then shift in X dimension else shift in + Y dimension. + + Returns: + A tuple containing a 3D uint8 Tensor that will be the result of translating + image by pixels. The second element of the tuple is bboxes, where now + the coordinates will be shifted to reflect the shifted image. + """ + if shift_horizontal: + image = translate_x(image, pixels, replace) + else: + image = translate_y(image, pixels, replace) + + # Convert bbox coordinates to pixel values. + image_height, image_width = image.shape[0], image.shape[1] + # pylint:disable=g-long-lambda + wrapped_shift_bbox = lambda bbox: _shift_bbox(bbox, image_height, image_width, pixels, shift_horizontal) + # pylint:enable=g-long-lambda + new_bboxes = deepcopy(bboxes) + num_bboxes = len(bboxes) + for idx in range(num_bboxes): + new_bboxes[idx] = wrapped_shift_bbox(bboxes[idx]) + return image.astype(np.uint8), new_bboxes + + +def shear_x(image, level, replace): + """Equivalent of PIL Shearing in X dimension.""" + # Shear parallel to x axis is a projective transform + # with a matrix form of: + # [1 level + # 0 1]. + image = Image.fromarray(wrap(image)) + image = image.transform(image.size, Image.AFFINE, (1, level, 0, 0, 1, 0)) + return unwrap(np.array(image), replace) + + +def shear_y(image, level, replace): + """Equivalent of PIL Shearing in Y dimension.""" + # Shear parallel to y axis is a projective transform + # with a matrix form of: + # [1 0 + # level 1]. + image = Image.fromarray(wrap(image)) + image = image.transform(image.size, Image.AFFINE, (1, 0, 0, level, 1, 0)) + return unwrap(np.array(image), replace) + + +def _shear_bbox(bbox, image_height, image_width, level, shear_horizontal): + """Shifts the bbox according to how the image was sheared. + + Args: + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + image_height: Int, height of the image. + image_width: Int, height of the image. + level: Float. How much to shear the image. + shear_horizontal: If true then shear in X dimension else shear in + the Y dimension. + + Returns: + A tensor of the same shape as bbox, but now with the shifted coordinates. + """ + image_height, image_width = (float(image_height), float(image_width)) + + # Change bbox coordinates to be pixels. + min_y = int(image_height * bbox[0]) + min_x = int(image_width * bbox[1]) + max_y = int(image_height * bbox[2]) + max_x = int(image_width * bbox[3]) + coordinates = np.stack( + [[min_y, min_x], [min_y, max_x], [max_y, min_x], [max_y, max_x]]) + coordinates = coordinates.astype(np.float32) + + # Shear the coordinates according to the translation matrix. + if shear_horizontal: + translation_matrix = np.stack([[1, 0], [-level, 1]]) + else: + translation_matrix = np.stack([[1, -level], [0, 1]]) + translation_matrix = translation_matrix.astype(np.float32) + new_coords = np.matmul(translation_matrix, + np.transpose(coordinates)).astype(np.int32) + + # Find min/max values and convert them back to floats. + min_y = float(np.min(new_coords[0, :])) / image_height + min_x = float(np.min(new_coords[1, :])) / image_width + max_y = float(np.max(new_coords[0, :])) / image_height + max_x = float(np.max(new_coords[1, :])) / image_width + + # Clip the bboxes to be sure the fall between [0, 1]. + min_y, min_x, max_y, max_x = _clip_bbox(min_y, min_x, max_y, max_x) + min_y, min_x, max_y, max_x = _check_bbox_area(min_y, min_x, max_y, max_x) + return np.stack([min_y, min_x, max_y, max_x]) + + +def shear_with_bboxes(image, bboxes, level, replace, shear_horizontal): + """Applies Shear Transformation to the image and shifts the bboxes. + + Args: + image: 3D uint8 Tensor. + bboxes: 2D Tensor that is a list of the bboxes in the image. Each bbox + has 4 elements (min_y, min_x, max_y, max_x) of type float with values + between [0, 1]. + level: Float. How much to shear the image. This value will be between + -0.3 to 0.3. + replace: A one or three value 1D tensor to fill empty pixels. + shear_horizontal: Boolean. If true then shear in X dimension else shear in + the Y dimension. + + Returns: + A tuple containing a 3D uint8 Tensor that will be the result of shearing + image by level. The second element of the tuple is bboxes, where now + the coordinates will be shifted to reflect the sheared image. + """ + if shear_horizontal: + image = shear_x(image, level, replace) + else: + image = shear_y(image, level, replace) + + # Convert bbox coordinates to pixel values. + image_height, image_width = image.shape[:2] + # pylint:disable=g-long-lambda + wrapped_shear_bbox = lambda bbox: _shear_bbox(bbox, image_height, image_width, level, shear_horizontal) + # pylint:enable=g-long-lambda + new_bboxes = deepcopy(bboxes) + num_bboxes = len(bboxes) + for idx in range(num_bboxes): + new_bboxes[idx] = wrapped_shear_bbox(bboxes[idx]) + return image.astype(np.uint8), new_bboxes + + +def autocontrast(image): + """Implements Autocontrast function from PIL. + + Args: + image: A 3D uint8 tensor. + + Returns: + The image after it has had autocontrast applied to it and will be of type + uint8. + """ + + def scale_channel(image): + """Scale the 2D image using the autocontrast rule.""" + # A possibly cheaper version can be done using cumsum/unique_with_counts + # over the histogram values, rather than iterating over the entire image. + # to compute mins and maxes. + lo = float(np.min(image)) + hi = float(np.max(image)) + + # Scale the image, making the lowest value 0 and the highest value 255. + def scale_values(im): + scale = 255.0 / (hi - lo) + offset = -lo * scale + im = im.astype(np.float32) * scale + offset + img = np.clip(im, a_min=0, a_max=255.0) + return im.astype(np.uint8) + + result = scale_values(image) if hi > lo else image + return result + + # Assumes RGB for now. Scales each channel independently + # and then stacks the result. + s1 = scale_channel(image[:, :, 0]) + s2 = scale_channel(image[:, :, 1]) + s3 = scale_channel(image[:, :, 2]) + image = np.stack([s1, s2, s3], 2) + return image + + +def sharpness(image, factor): + """Implements Sharpness function from PIL.""" + orig_image = image + image = image.astype(np.float32) + # Make image 4D for conv operation. + # SMOOTH PIL Kernel. + kernel = np.array([[1, 1, 1], [1, 5, 1], [1, 1, 1]], dtype=np.float32) / 13. + result = cv2.filter2D(image, -1, kernel).astype(np.uint8) + + # Blend the final result. + return blend(result, orig_image, factor) + + +def equalize(image): + """Implements Equalize function from PIL using.""" + + def scale_channel(im, c): + """Scale the data in the channel to implement equalize.""" + im = im[:, :, c].astype(np.int32) + # Compute the histogram of the image channel. + histo, _ = np.histogram(im, range=[0, 255], bins=256) + + # For the purposes of computing the step, filter out the nonzeros. + nonzero = np.where(np.not_equal(histo, 0)) + nonzero_histo = np.reshape(np.take(histo, nonzero), [-1]) + step = (np.sum(nonzero_histo) - nonzero_histo[-1]) // 255 + + def build_lut(histo, step): + # Compute the cumulative sum, shifting by step // 2 + # and then normalization by step. + lut = (np.cumsum(histo) + (step // 2)) // step + # Shift lut, prepending with 0. + lut = np.concatenate([[0], lut[:-1]], 0) + # Clip the counts to be in range. This is done + # in the C code for image.point. + return np.clip(lut, a_min=0, a_max=255).astype(np.uint8) + + # If step is zero, return the original image. Otherwise, build + # lut from the full histogram and step and then index from it. + if step == 0: + result = im + else: + result = np.take(build_lut(histo, step), im) + + return result.astype(np.uint8) + + # Assumes RGB for now. Scales each channel independently + # and then stacks the result. + s1 = scale_channel(image, 0) + s2 = scale_channel(image, 1) + s3 = scale_channel(image, 2) + image = np.stack([s1, s2, s3], 2) + return image + + +def wrap(image): + """Returns 'image' with an extra channel set to all 1s.""" + shape = image.shape + extended_channel = 255 * np.ones([shape[0], shape[1], 1], image.dtype) + extended = np.concatenate([image, extended_channel], 2).astype(image.dtype) + return extended + + +def unwrap(image, replace): + """Unwraps an image produced by wrap. + + Where there is a 0 in the last channel for every spatial position, + the rest of the three channels in that spatial dimension are grayed + (set to 128). Operations like translate and shear on a wrapped + Tensor will leave 0s in empty locations. Some transformations look + at the intensity of values to do preprocessing, and we want these + empty pixels to assume the 'average' value, rather than pure black. + + + Args: + image: A 3D Image Tensor with 4 channels. + replace: A one or three value 1D tensor to fill empty pixels. + + Returns: + image: A 3D image Tensor with 3 channels. + """ + image_shape = image.shape + # Flatten the spatial dimensions. + flattened_image = np.reshape(image, [-1, image_shape[2]]) + + # Find all pixels where the last channel is zero. + alpha_channel = flattened_image[:, 3] + + replace = np.concatenate([replace, np.ones([1], image.dtype)], 0) + + # Where they are zero, fill them in with 'replace'. + alpha_channel = np.reshape(alpha_channel, (-1, 1)) + alpha_channel = np.tile(alpha_channel, reps=(1, flattened_image.shape[1])) + + flattened_image = np.where( + np.equal(alpha_channel, 0), + np.ones_like( + flattened_image, dtype=image.dtype) * replace, + flattened_image) + + image = np.reshape(flattened_image, image_shape) + image = image[:, :, :3] + return image.astype(np.uint8) + + +def _cutout_inside_bbox(image, bbox, pad_fraction): + """Generates cutout mask and the mean pixel value of the bbox. + + First a location is randomly chosen within the image as the center where the + cutout mask will be applied. Note this can be towards the boundaries of the + image, so the full cutout mask may not be applied. + + Args: + image: 3D uint8 Tensor. + bbox: 1D Tensor that has 4 elements (min_y, min_x, max_y, max_x) + of type float that represents the normalized coordinates between 0 and 1. + pad_fraction: Float that specifies how large the cutout mask should be in + in reference to the size of the original bbox. If pad_fraction is 0.25, + then the cutout mask will be of shape + (0.25 * bbox height, 0.25 * bbox width). + + Returns: + A tuple. Fist element is a tensor of the same shape as image where each + element is either a 1 or 0 that is used to determine where the image + will have cutout applied. The second element is the mean of the pixels + in the image where the bbox is located. + mask value: [0,1] + """ + image_height, image_width = image.shape[0], image.shape[1] + # Transform from shape [1, 4] to [4]. + bbox = np.squeeze(bbox) + + min_y = int(float(image_height) * bbox[0]) + min_x = int(float(image_width) * bbox[1]) + max_y = int(float(image_height) * bbox[2]) + max_x = int(float(image_width) * bbox[3]) + + # Calculate the mean pixel values in the bounding box, which will be used + # to fill the cutout region. + mean = np.mean(image[min_y:max_y + 1, min_x:max_x + 1], axis=(0, 1)) + # Cutout mask will be size pad_size_heigh * 2 by pad_size_width * 2 if the + # region lies entirely within the bbox. + box_height = max_y - min_y + 1 + box_width = max_x - min_x + 1 + pad_size_height = int(pad_fraction * (box_height / 2)) + pad_size_width = int(pad_fraction * (box_width / 2)) + + # Sample the center location in the image where the zero mask will be applied. + cutout_center_height = np.random.randint(min_y, max_y + 1, dtype=np.int32) + cutout_center_width = np.random.randint(min_x, max_x + 1, dtype=np.int32) + + lower_pad = np.maximum(0, cutout_center_height - pad_size_height) + upper_pad = np.maximum( + 0, image_height - cutout_center_height - pad_size_height) + left_pad = np.maximum(0, cutout_center_width - pad_size_width) + right_pad = np.maximum(0, + image_width - cutout_center_width - pad_size_width) + + cutout_shape = [ + image_height - (lower_pad + upper_pad), + image_width - (left_pad + right_pad) + ] + padding_dims = [[lower_pad, upper_pad], [left_pad, right_pad]] + + mask = np.pad(np.zeros( + cutout_shape, dtype=image.dtype), + padding_dims, + 'constant', + constant_values=1) + + mask = np.expand_dims(mask, 2) + mask = np.tile(mask, [1, 1, 3]) + return mask, mean + + +def bbox_cutout(image, bboxes, pad_fraction, replace_with_mean): + """Applies cutout to the image according to bbox information. + + This is a cutout variant that using bbox information to make more informed + decisions on where to place the cutout mask. + + Args: + image: 3D uint8 Tensor. + bboxes: 2D Tensor that is a list of the bboxes in the image. Each bbox + has 4 elements (min_y, min_x, max_y, max_x) of type float with values + between [0, 1]. + pad_fraction: Float that specifies how large the cutout mask should be in + in reference to the size of the original bbox. If pad_fraction is 0.25, + then the cutout mask will be of shape + (0.25 * bbox height, 0.25 * bbox width). + replace_with_mean: Boolean that specified what value should be filled in + where the cutout mask is applied. Since the incoming image will be of + uint8 and will not have had any mean normalization applied, by default + we set the value to be 128. If replace_with_mean is True then we find + the mean pixel values across the channel dimension and use those to fill + in where the cutout mask is applied. + + Returns: + A tuple. First element is a tensor of the same shape as image that has + cutout applied to it. Second element is the bboxes that were passed in + that will be unchanged. + """ + + def apply_bbox_cutout(image, bboxes, pad_fraction): + """Applies cutout to a single bounding box within image.""" + # Choose a single bounding box to apply cutout to. + random_index = np.random.randint(0, bboxes.shape[0], dtype=np.int32) + # Select the corresponding bbox and apply cutout. + chosen_bbox = np.take(bboxes, random_index, axis=0) + mask, mean = _cutout_inside_bbox(image, chosen_bbox, pad_fraction) + + # When applying cutout we either set the pixel value to 128 or to the mean + # value inside the bbox. + replace = mean if replace_with_mean else [128] * 3 + + # Apply the cutout mask to the image. Where the mask is 0 we fill it with + # `replace`. + image = np.where( + np.equal(mask, 0), + np.ones_like( + image, dtype=image.dtype) * replace, + image).astype(image.dtype) + return image + + # Check to see if there are boxes, if so then apply boxcutout. + if len(bboxes) != 0: + image = apply_bbox_cutout(image, bboxes, pad_fraction) + + return image, bboxes + + +NAME_TO_FUNC = { + 'AutoContrast': autocontrast, + 'Equalize': equalize, + 'Posterize': posterize, + 'Solarize': solarize, + 'SolarizeAdd': solarize_add, + 'Color': color, + 'Contrast': contrast, + 'Brightness': brightness, + 'Sharpness': sharpness, + 'Cutout': cutout, + 'BBox_Cutout': bbox_cutout, + 'Rotate_BBox': rotate_with_bboxes, + # pylint:disable=g-long-lambda + 'TranslateX_BBox': lambda image, bboxes, pixels, replace: translate_bbox( + image, bboxes, pixels, replace, shift_horizontal=True), + 'TranslateY_BBox': lambda image, bboxes, pixels, replace: translate_bbox( + image, bboxes, pixels, replace, shift_horizontal=False), + 'ShearX_BBox': lambda image, bboxes, level, replace: shear_with_bboxes( + image, bboxes, level, replace, shear_horizontal=True), + 'ShearY_BBox': lambda image, bboxes, level, replace: shear_with_bboxes( + image, bboxes, level, replace, shear_horizontal=False), + # pylint:enable=g-long-lambda + 'Rotate_Only_BBoxes': rotate_only_bboxes, + 'ShearX_Only_BBoxes': shear_x_only_bboxes, + 'ShearY_Only_BBoxes': shear_y_only_bboxes, + 'TranslateX_Only_BBoxes': translate_x_only_bboxes, + 'TranslateY_Only_BBoxes': translate_y_only_bboxes, + 'Flip_Only_BBoxes': flip_only_bboxes, + 'Solarize_Only_BBoxes': solarize_only_bboxes, + 'Equalize_Only_BBoxes': equalize_only_bboxes, + 'Cutout_Only_BBoxes': cutout_only_bboxes, +} + + +def _randomly_negate_tensor(tensor): + """With 50% prob turn the tensor negative.""" + should_flip = np.floor(np.random.rand() + 0.5) >= 1 + final_tensor = tensor if should_flip else -tensor + return final_tensor + + +def _rotate_level_to_arg(level): + level = (level / _MAX_LEVEL) * 30. + level = _randomly_negate_tensor(level) + return (level, ) + + +def _shrink_level_to_arg(level): + """Converts level to ratio by which we shrink the image content.""" + if level == 0: + return (1.0, ) # if level is zero, do not shrink the image + # Maximum shrinking ratio is 2.9. + level = 2. / (_MAX_LEVEL / level) + 0.9 + return (level, ) + + +def _enhance_level_to_arg(level): + return ((level / _MAX_LEVEL) * 1.8 + 0.1, ) + + +def _shear_level_to_arg(level): + level = (level / _MAX_LEVEL) * 0.3 + # Flip level to negative with 50% chance. + level = _randomly_negate_tensor(level) + return (level, ) + + +def _translate_level_to_arg(level, translate_const): + level = (level / _MAX_LEVEL) * float(translate_const) + # Flip level to negative with 50% chance. + level = _randomly_negate_tensor(level) + return (level, ) + + +def _bbox_cutout_level_to_arg(level, hparams): + cutout_pad_fraction = (level / + _MAX_LEVEL) * 0.75 # hparams.cutout_max_pad_fraction + return (cutout_pad_fraction, False) # hparams.cutout_bbox_replace_with_mean + + +def level_to_arg(hparams): + return { + 'AutoContrast': lambda level: (), + 'Equalize': lambda level: (), + 'Posterize': lambda level: (int((level / _MAX_LEVEL) * 4), ), + 'Solarize': lambda level: (int((level / _MAX_LEVEL) * 256), ), + 'SolarizeAdd': lambda level: (int((level / _MAX_LEVEL) * 110), ), + 'Color': _enhance_level_to_arg, + 'Contrast': _enhance_level_to_arg, + 'Brightness': _enhance_level_to_arg, + 'Sharpness': _enhance_level_to_arg, + 'Cutout': + lambda level: (int((level / _MAX_LEVEL) * 100), ), # hparams.cutout_const=100 + # pylint:disable=g-long-lambda + 'BBox_Cutout': lambda level: _bbox_cutout_level_to_arg(level, hparams), + 'TranslateX_BBox': + lambda level: _translate_level_to_arg(level, 250), # hparams.translate_const=250 + 'TranslateY_BBox': + lambda level: _translate_level_to_arg(level, 250), # hparams.translate_cons + # pylint:enable=g-long-lambda + 'ShearX_BBox': _shear_level_to_arg, + 'ShearY_BBox': _shear_level_to_arg, + 'Rotate_BBox': _rotate_level_to_arg, + 'Rotate_Only_BBoxes': _rotate_level_to_arg, + 'ShearX_Only_BBoxes': _shear_level_to_arg, + 'ShearY_Only_BBoxes': _shear_level_to_arg, + # pylint:disable=g-long-lambda + 'TranslateX_Only_BBoxes': + lambda level: _translate_level_to_arg(level, 120), # hparams.translate_bbox_const + 'TranslateY_Only_BBoxes': + lambda level: _translate_level_to_arg(level, 120), # hparams.translate_bbox_const + # pylint:enable=g-long-lambda + 'Flip_Only_BBoxes': lambda level: (), + 'Solarize_Only_BBoxes': + lambda level: (int((level / _MAX_LEVEL) * 256), ), + 'Equalize_Only_BBoxes': lambda level: (), + # pylint:disable=g-long-lambda + 'Cutout_Only_BBoxes': + lambda level: (int((level / _MAX_LEVEL) * 50), ), # hparams.cutout_bbox_const + # pylint:enable=g-long-lambda + } + + +def bbox_wrapper(func): + """Adds a bboxes function argument to func and returns unchanged bboxes.""" + + def wrapper(images, bboxes, *args, **kwargs): + return (func(images, *args, **kwargs), bboxes) + + return wrapper + + +def _parse_policy_info(name, prob, level, replace_value, augmentation_hparams): + """Return the function that corresponds to `name` and update `level` param.""" + func = NAME_TO_FUNC[name] + args = level_to_arg(augmentation_hparams)[name](level) + + # Check to see if prob is passed into function. This is used for operations + # where we alter bboxes independently. + # pytype:disable=wrong-arg-types + if 'prob' in inspect.getfullargspec(func)[0]: + args = tuple([prob] + list(args)) + # pytype:enable=wrong-arg-types + + # Add in replace arg if it is required for the function that is being called. + if 'replace' in inspect.getfullargspec(func)[0]: + # Make sure replace is the final argument + assert 'replace' == inspect.getfullargspec(func)[0][-1] + args = tuple(list(args) + [replace_value]) + + # Add bboxes as the second positional argument for the function if it does + # not already exist. + if 'bboxes' not in inspect.getfullargspec(func)[0]: + func = bbox_wrapper(func) + return (func, prob, args) + + +def _apply_func_with_prob(func, image, args, prob, bboxes): + """Apply `func` to image w/ `args` as input with probability `prob`.""" + assert isinstance(args, tuple) + assert 'bboxes' == inspect.getfullargspec(func)[0][1] + + # If prob is a function argument, then this randomness is being handled + # inside the function, so make sure it is always called. + if 'prob' in inspect.getfullargspec(func)[0]: + prob = 1.0 + + # Apply the function with probability `prob`. + should_apply_op = np.floor(np.random.rand() + 0.5) >= 1 + if should_apply_op: + augmented_image, augmented_bboxes = func(image, bboxes, *args) + else: + augmented_image, augmented_bboxes = (image, bboxes) + return augmented_image, augmented_bboxes + + +def select_and_apply_random_policy(policies, image, bboxes): + """Select a random policy from `policies` and apply it to `image`.""" + policy_to_select = np.random.randint(0, len(policies), dtype=np.int32) + # policy_to_select = 6 # for test + for (i, policy) in enumerate(policies): + if i == policy_to_select: + image, bboxes = policy(image, bboxes) + return (image, bboxes) + + +def build_and_apply_nas_policy(policies, image, bboxes, augmentation_hparams): + """Build a policy from the given policies passed in and apply to image. + + Args: + policies: list of lists of tuples in the form `(func, prob, level)`, `func` + is a string name of the augmentation function, `prob` is the probability + of applying the `func` operation, `level` is the input argument for + `func`. + image: numpy array that the resulting policy will be applied to. + bboxes: + augmentation_hparams: Hparams associated with the NAS learned policy. + + Returns: + A version of image that now has data augmentation applied to it based on + the `policies` pass into the function. Additionally, returns bboxes if + a value for them is passed in that is not None + """ + replace_value = [128, 128, 128] + + # func is the string name of the augmentation function, prob is the + # probability of applying the operation and level is the parameter associated + + # tf_policies are functions that take in an image and return an augmented + # image. + tf_policies = [] + for policy in policies: + tf_policy = [] + # Link string name to the correct python function and make sure the correct + # argument is passed into that function. + for policy_info in policy: + policy_info = list( + policy_info) + [replace_value, augmentation_hparams] + + tf_policy.append(_parse_policy_info(*policy_info)) + # Now build the tf policy that will apply the augmentation procedue + # on image. + def make_final_policy(tf_policy_): + def final_policy(image_, bboxes_): + for func, prob, args in tf_policy_: + image_, bboxes_ = _apply_func_with_prob(func, image_, args, + prob, bboxes_) + return image_, bboxes_ + + return final_policy + + tf_policies.append(make_final_policy(tf_policy)) + + augmented_images, augmented_bboxes = select_and_apply_random_policy( + tf_policies, image, bboxes) + # If no bounding boxes were specified, then just return the images. + return (augmented_images, augmented_bboxes) + + +# TODO(barretzoph): Add in ArXiv link once paper is out. +def distort_image_with_autoaugment(image, bboxes, augmentation_name): + """Applies the AutoAugment policy to `image` and `bboxes`. + + Args: + image: `Tensor` of shape [height, width, 3] representing an image. + bboxes: `Tensor` of shape [N, 4] representing ground truth boxes that are + normalized between [0, 1]. + augmentation_name: The name of the AutoAugment policy to use. The available + options are `v0`, `v1`, `v2`, `v3` and `test`. `v0` is the policy used for + all of the results in the paper and was found to achieve the best results + on the COCO dataset. `v1`, `v2` and `v3` are additional good policies + found on the COCO dataset that have slight variation in what operations + were used during the search procedure along with how many operations are + applied in parallel to a single image (2 vs 3). + + Returns: + A tuple containing the augmented versions of `image` and `bboxes`. + """ + available_policies = { + 'v0': policy_v0, + 'v1': policy_v1, + 'v2': policy_v2, + 'v3': policy_v3, + 'test': policy_vtest + } + if augmentation_name not in available_policies: + raise ValueError('Invalid augmentation_name: {}'.format( + augmentation_name)) + + policy = available_policies[augmentation_name]() + augmentation_hparams = {} + return build_and_apply_nas_policy(policy, image, bboxes, + augmentation_hparams) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/batch_operators.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/batch_operators.py new file mode 100644 index 000000000..e43fb7d20 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/batch_operators.py @@ -0,0 +1,1060 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence + +import cv2 +import math +import numpy as np +from .operators import register_op, BaseOperator, Resize +from .op_helper import jaccard_overlap, gaussian2D, gaussian_radius, draw_umich_gaussian +from .atss_assigner import ATSSAssigner +from scipy import ndimage + +from ppdet.modeling import bbox_utils +from ppdet.utils.logger import setup_logger +from ppdet.modeling.keypoint_utils import get_affine_transform, affine_transform +logger = setup_logger(__name__) + +__all__ = [ + 'PadBatch', + 'BatchRandomResize', + 'Gt2YoloTarget', + 'Gt2FCOSTarget', + 'Gt2TTFTarget', + 'Gt2Solov2Target', + 'Gt2SparseRCNNTarget', + 'PadMaskBatch', + 'Gt2GFLTarget', + 'Gt2CenterNetTarget', +] + + +@register_op +class PadBatch(BaseOperator): + """ + Pad a batch of samples so they can be divisible by a stride. + The layout of each image should be 'CHW'. + Args: + pad_to_stride (int): If `pad_to_stride > 0`, pad zeros to ensure + height and width is divisible by `pad_to_stride`. + """ + + def __init__(self, pad_to_stride=0): + super(PadBatch, self).__init__() + self.pad_to_stride = pad_to_stride + + def __call__(self, samples, context=None): + """ + Args: + samples (list): a batch of sample, each is dict. + """ + coarsest_stride = self.pad_to_stride + + max_shape = np.array([data['image'].shape for data in samples]).max( + axis=0) + if coarsest_stride > 0: + max_shape[1] = int( + np.ceil(max_shape[1] / coarsest_stride) * coarsest_stride) + max_shape[2] = int( + np.ceil(max_shape[2] / coarsest_stride) * coarsest_stride) + + for data in samples: + im = data['image'] + im_c, im_h, im_w = im.shape[:] + padding_im = np.zeros( + (im_c, max_shape[1], max_shape[2]), dtype=np.float32) + padding_im[:, :im_h, :im_w] = im + data['image'] = padding_im + if 'semantic' in data and data['semantic'] is not None: + semantic = data['semantic'] + padding_sem = np.zeros( + (1, max_shape[1], max_shape[2]), dtype=np.float32) + padding_sem[:, :im_h, :im_w] = semantic + data['semantic'] = padding_sem + if 'gt_segm' in data and data['gt_segm'] is not None: + gt_segm = data['gt_segm'] + padding_segm = np.zeros( + (gt_segm.shape[0], max_shape[1], max_shape[2]), + dtype=np.uint8) + padding_segm[:, :im_h, :im_w] = gt_segm + data['gt_segm'] = padding_segm + + if 'gt_rbox2poly' in data and data['gt_rbox2poly'] is not None: + # ploy to rbox + polys = data['gt_rbox2poly'] + rbox = bbox_utils.poly2rbox(polys) + data['gt_rbox'] = rbox + + return samples + + +@register_op +class BatchRandomResize(BaseOperator): + """ + Resize image to target size randomly. random target_size and interpolation method + Args: + target_size (int, list, tuple): image target size, if random size is True, must be list or tuple + keep_ratio (bool): whether keep_raio or not, default true + interp (int): the interpolation method + random_size (bool): whether random select target size of image + random_interp (bool): whether random select interpolation method + """ + + def __init__(self, + target_size, + keep_ratio, + interp=cv2.INTER_NEAREST, + random_size=True, + random_interp=False): + super(BatchRandomResize, self).__init__() + self.keep_ratio = keep_ratio + self.interps = [ + cv2.INTER_NEAREST, + cv2.INTER_LINEAR, + cv2.INTER_AREA, + cv2.INTER_CUBIC, + cv2.INTER_LANCZOS4, + ] + self.interp = interp + assert isinstance(target_size, ( + int, Sequence)), "target_size must be int, list or tuple" + if random_size and not isinstance(target_size, list): + raise TypeError( + "Type of target_size is invalid when random_size is True. Must be List, now is {}". + format(type(target_size))) + self.target_size = target_size + self.random_size = random_size + self.random_interp = random_interp + + def __call__(self, samples, context=None): + if self.random_size: + index = np.random.choice(len(self.target_size)) + target_size = self.target_size[index] + else: + target_size = self.target_size + + if self.random_interp: + interp = np.random.choice(self.interps) + else: + interp = self.interp + + resizer = Resize(target_size, keep_ratio=self.keep_ratio, interp=interp) + return resizer(samples, context=context) + + +@register_op +class Gt2YoloTarget(BaseOperator): + """ + Generate YOLOv3 targets by groud truth data, this operator is only used in + fine grained YOLOv3 loss mode + """ + + def __init__(self, + anchors, + anchor_masks, + downsample_ratios, + num_classes=80, + iou_thresh=1.): + super(Gt2YoloTarget, self).__init__() + self.anchors = anchors + self.anchor_masks = anchor_masks + self.downsample_ratios = downsample_ratios + self.num_classes = num_classes + self.iou_thresh = iou_thresh + + def __call__(self, samples, context=None): + assert len(self.anchor_masks) == len(self.downsample_ratios), \ + "anchor_masks', and 'downsample_ratios' should have same length." + + h, w = samples[0]['image'].shape[1:3] + an_hw = np.array(self.anchors) / np.array([[w, h]]) + for sample in samples: + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + if 'gt_score' not in sample: + sample['gt_score'] = np.ones( + (gt_bbox.shape[0], 1), dtype=np.float32) + gt_score = sample['gt_score'] + for i, ( + mask, downsample_ratio + ) in enumerate(zip(self.anchor_masks, self.downsample_ratios)): + grid_h = int(h / downsample_ratio) + grid_w = int(w / downsample_ratio) + target = np.zeros( + (len(mask), 6 + self.num_classes, grid_h, grid_w), + dtype=np.float32) + for b in range(gt_bbox.shape[0]): + gx, gy, gw, gh = gt_bbox[b, :] + cls = gt_class[b] + score = gt_score[b] + if gw <= 0. or gh <= 0. or score <= 0.: + continue + + # find best match anchor index + best_iou = 0. + best_idx = -1 + for an_idx in range(an_hw.shape[0]): + iou = jaccard_overlap( + [0., 0., gw, gh], + [0., 0., an_hw[an_idx, 0], an_hw[an_idx, 1]]) + if iou > best_iou: + best_iou = iou + best_idx = an_idx + + gi = int(gx * grid_w) + gj = int(gy * grid_h) + + # gtbox should be regresed in this layes if best match + # anchor index in anchor mask of this layer + if best_idx in mask: + best_n = mask.index(best_idx) + + # x, y, w, h, scale + target[best_n, 0, gj, gi] = gx * grid_w - gi + target[best_n, 1, gj, gi] = gy * grid_h - gj + target[best_n, 2, gj, gi] = np.log( + gw * w / self.anchors[best_idx][0]) + target[best_n, 3, gj, gi] = np.log( + gh * h / self.anchors[best_idx][1]) + target[best_n, 4, gj, gi] = 2.0 - gw * gh + + # objectness record gt_score + target[best_n, 5, gj, gi] = score + + # classification + target[best_n, 6 + cls, gj, gi] = 1. + + # For non-matched anchors, calculate the target if the iou + # between anchor and gt is larger than iou_thresh + if self.iou_thresh < 1: + for idx, mask_i in enumerate(mask): + if mask_i == best_idx: continue + iou = jaccard_overlap( + [0., 0., gw, gh], + [0., 0., an_hw[mask_i, 0], an_hw[mask_i, 1]]) + if iou > self.iou_thresh and target[idx, 5, gj, + gi] == 0.: + # x, y, w, h, scale + target[idx, 0, gj, gi] = gx * grid_w - gi + target[idx, 1, gj, gi] = gy * grid_h - gj + target[idx, 2, gj, gi] = np.log( + gw * w / self.anchors[mask_i][0]) + target[idx, 3, gj, gi] = np.log( + gh * h / self.anchors[mask_i][1]) + target[idx, 4, gj, gi] = 2.0 - gw * gh + + # objectness record gt_score + target[idx, 5, gj, gi] = score + + # classification + target[idx, 6 + cls, gj, gi] = 1. + sample['target{}'.format(i)] = target + + # remove useless gt_class and gt_score after target calculated + sample.pop('gt_class') + sample.pop('gt_score') + + return samples + + +@register_op +class Gt2FCOSTarget(BaseOperator): + """ + Generate FCOS targets by groud truth data + """ + + def __init__(self, + object_sizes_boundary, + center_sampling_radius, + downsample_ratios, + norm_reg_targets=False): + super(Gt2FCOSTarget, self).__init__() + self.center_sampling_radius = center_sampling_radius + self.downsample_ratios = downsample_ratios + self.INF = np.inf + self.object_sizes_boundary = [-1] + object_sizes_boundary + [self.INF] + object_sizes_of_interest = [] + for i in range(len(self.object_sizes_boundary) - 1): + object_sizes_of_interest.append([ + self.object_sizes_boundary[i], self.object_sizes_boundary[i + 1] + ]) + self.object_sizes_of_interest = object_sizes_of_interest + self.norm_reg_targets = norm_reg_targets + + def _compute_points(self, w, h): + """ + compute the corresponding points in each feature map + :param h: image height + :param w: image width + :return: points from all feature map + """ + locations = [] + for stride in self.downsample_ratios: + shift_x = np.arange(0, w, stride).astype(np.float32) + shift_y = np.arange(0, h, stride).astype(np.float32) + shift_x, shift_y = np.meshgrid(shift_x, shift_y) + shift_x = shift_x.flatten() + shift_y = shift_y.flatten() + location = np.stack([shift_x, shift_y], axis=1) + stride // 2 + locations.append(location) + num_points_each_level = [len(location) for location in locations] + locations = np.concatenate(locations, axis=0) + return locations, num_points_each_level + + def _convert_xywh2xyxy(self, gt_bbox, w, h): + """ + convert the bounding box from style xywh to xyxy + :param gt_bbox: bounding boxes normalized into [0, 1] + :param w: image width + :param h: image height + :return: bounding boxes in xyxy style + """ + bboxes = gt_bbox.copy() + bboxes[:, [0, 2]] = bboxes[:, [0, 2]] * w + bboxes[:, [1, 3]] = bboxes[:, [1, 3]] * h + bboxes[:, 2] = bboxes[:, 0] + bboxes[:, 2] + bboxes[:, 3] = bboxes[:, 1] + bboxes[:, 3] + return bboxes + + def _check_inside_boxes_limited(self, gt_bbox, xs, ys, + num_points_each_level): + """ + check if points is within the clipped boxes + :param gt_bbox: bounding boxes + :param xs: horizontal coordinate of points + :param ys: vertical coordinate of points + :return: the mask of points is within gt_box or not + """ + bboxes = np.reshape( + gt_bbox, newshape=[1, gt_bbox.shape[0], gt_bbox.shape[1]]) + bboxes = np.tile(bboxes, reps=[xs.shape[0], 1, 1]) + ct_x = (bboxes[:, :, 0] + bboxes[:, :, 2]) / 2 + ct_y = (bboxes[:, :, 1] + bboxes[:, :, 3]) / 2 + beg = 0 + clipped_box = bboxes.copy() + for lvl, stride in enumerate(self.downsample_ratios): + end = beg + num_points_each_level[lvl] + stride_exp = self.center_sampling_radius * stride + clipped_box[beg:end, :, 0] = np.maximum( + bboxes[beg:end, :, 0], ct_x[beg:end, :] - stride_exp) + clipped_box[beg:end, :, 1] = np.maximum( + bboxes[beg:end, :, 1], ct_y[beg:end, :] - stride_exp) + clipped_box[beg:end, :, 2] = np.minimum( + bboxes[beg:end, :, 2], ct_x[beg:end, :] + stride_exp) + clipped_box[beg:end, :, 3] = np.minimum( + bboxes[beg:end, :, 3], ct_y[beg:end, :] + stride_exp) + beg = end + l_res = xs - clipped_box[:, :, 0] + r_res = clipped_box[:, :, 2] - xs + t_res = ys - clipped_box[:, :, 1] + b_res = clipped_box[:, :, 3] - ys + clipped_box_reg_targets = np.stack([l_res, t_res, r_res, b_res], axis=2) + inside_gt_box = np.min(clipped_box_reg_targets, axis=2) > 0 + return inside_gt_box + + def __call__(self, samples, context=None): + assert len(self.object_sizes_of_interest) == len(self.downsample_ratios), \ + "object_sizes_of_interest', and 'downsample_ratios' should have same length." + + for sample in samples: + im = sample['image'] + bboxes = sample['gt_bbox'] + gt_class = sample['gt_class'] + # calculate the locations + h, w = im.shape[1:3] + points, num_points_each_level = self._compute_points(w, h) + object_scale_exp = [] + for i, num_pts in enumerate(num_points_each_level): + object_scale_exp.append( + np.tile( + np.array([self.object_sizes_of_interest[i]]), + reps=[num_pts, 1])) + object_scale_exp = np.concatenate(object_scale_exp, axis=0) + + gt_area = (bboxes[:, 2] - bboxes[:, 0]) * ( + bboxes[:, 3] - bboxes[:, 1]) + xs, ys = points[:, 0], points[:, 1] + xs = np.reshape(xs, newshape=[xs.shape[0], 1]) + xs = np.tile(xs, reps=[1, bboxes.shape[0]]) + ys = np.reshape(ys, newshape=[ys.shape[0], 1]) + ys = np.tile(ys, reps=[1, bboxes.shape[0]]) + + l_res = xs - bboxes[:, 0] + r_res = bboxes[:, 2] - xs + t_res = ys - bboxes[:, 1] + b_res = bboxes[:, 3] - ys + reg_targets = np.stack([l_res, t_res, r_res, b_res], axis=2) + if self.center_sampling_radius > 0: + is_inside_box = self._check_inside_boxes_limited( + bboxes, xs, ys, num_points_each_level) + else: + is_inside_box = np.min(reg_targets, axis=2) > 0 + # check if the targets is inside the corresponding level + max_reg_targets = np.max(reg_targets, axis=2) + lower_bound = np.tile( + np.expand_dims( + object_scale_exp[:, 0], axis=1), + reps=[1, max_reg_targets.shape[1]]) + high_bound = np.tile( + np.expand_dims( + object_scale_exp[:, 1], axis=1), + reps=[1, max_reg_targets.shape[1]]) + is_match_current_level = \ + (max_reg_targets > lower_bound) & \ + (max_reg_targets < high_bound) + points2gtarea = np.tile( + np.expand_dims( + gt_area, axis=0), reps=[xs.shape[0], 1]) + points2gtarea[is_inside_box == 0] = self.INF + points2gtarea[is_match_current_level == 0] = self.INF + points2min_area = points2gtarea.min(axis=1) + points2min_area_ind = points2gtarea.argmin(axis=1) + labels = gt_class[points2min_area_ind] + 1 + labels[points2min_area == self.INF] = 0 + reg_targets = reg_targets[range(xs.shape[0]), points2min_area_ind] + ctn_targets = np.sqrt((reg_targets[:, [0, 2]].min(axis=1) / \ + reg_targets[:, [0, 2]].max(axis=1)) * \ + (reg_targets[:, [1, 3]].min(axis=1) / \ + reg_targets[:, [1, 3]].max(axis=1))).astype(np.float32) + ctn_targets = np.reshape( + ctn_targets, newshape=[ctn_targets.shape[0], 1]) + ctn_targets[labels <= 0] = 0 + pos_ind = np.nonzero(labels != 0) + reg_targets_pos = reg_targets[pos_ind[0], :] + split_sections = [] + beg = 0 + for lvl in range(len(num_points_each_level)): + end = beg + num_points_each_level[lvl] + split_sections.append(end) + beg = end + labels_by_level = np.split(labels, split_sections, axis=0) + reg_targets_by_level = np.split(reg_targets, split_sections, axis=0) + ctn_targets_by_level = np.split(ctn_targets, split_sections, axis=0) + for lvl in range(len(self.downsample_ratios)): + grid_w = int(np.ceil(w / self.downsample_ratios[lvl])) + grid_h = int(np.ceil(h / self.downsample_ratios[lvl])) + if self.norm_reg_targets: + sample['reg_target{}'.format(lvl)] = \ + np.reshape( + reg_targets_by_level[lvl] / \ + self.downsample_ratios[lvl], + newshape=[grid_h, grid_w, 4]) + else: + sample['reg_target{}'.format(lvl)] = np.reshape( + reg_targets_by_level[lvl], + newshape=[grid_h, grid_w, 4]) + sample['labels{}'.format(lvl)] = np.reshape( + labels_by_level[lvl], newshape=[grid_h, grid_w, 1]) + sample['centerness{}'.format(lvl)] = np.reshape( + ctn_targets_by_level[lvl], newshape=[grid_h, grid_w, 1]) + + sample.pop('is_crowd', None) + sample.pop('difficult', None) + sample.pop('gt_class', None) + sample.pop('gt_bbox', None) + return samples + + +@register_op +class Gt2GFLTarget(BaseOperator): + """ + Generate GFocal loss targets by groud truth data + """ + + def __init__(self, + num_classes=80, + downsample_ratios=[8, 16, 32, 64, 128], + grid_cell_scale=4, + cell_offset=0): + super(Gt2GFLTarget, self).__init__() + self.num_classes = num_classes + self.downsample_ratios = downsample_ratios + self.grid_cell_scale = grid_cell_scale + self.cell_offset = cell_offset + + self.assigner = ATSSAssigner() + + def get_grid_cells(self, featmap_size, scale, stride, offset=0): + """ + Generate grid cells of a feature map for target assignment. + Args: + featmap_size: Size of a single level feature map. + scale: Grid cell scale. + stride: Down sample stride of the feature map. + offset: Offset of grid cells. + return: + Grid_cells xyxy position. Size should be [feat_w * feat_h, 4] + """ + cell_size = stride * scale + h, w = featmap_size + x_range = (np.arange(w, dtype=np.float32) + offset) * stride + y_range = (np.arange(h, dtype=np.float32) + offset) * stride + x, y = np.meshgrid(x_range, y_range) + y = y.flatten() + x = x.flatten() + grid_cells = np.stack( + [ + x - 0.5 * cell_size, y - 0.5 * cell_size, x + 0.5 * cell_size, + y + 0.5 * cell_size + ], + axis=-1) + return grid_cells + + def get_sample(self, assign_gt_inds, gt_bboxes): + pos_inds = np.unique(np.nonzero(assign_gt_inds > 0)[0]) + neg_inds = np.unique(np.nonzero(assign_gt_inds == 0)[0]) + pos_assigned_gt_inds = assign_gt_inds[pos_inds] - 1 + + if gt_bboxes.size == 0: + # hack for index error case + assert pos_assigned_gt_inds.size == 0 + pos_gt_bboxes = np.empty_like(gt_bboxes).reshape(-1, 4) + else: + if len(gt_bboxes.shape) < 2: + gt_bboxes = gt_bboxes.resize(-1, 4) + pos_gt_bboxes = gt_bboxes[pos_assigned_gt_inds, :] + return pos_inds, neg_inds, pos_gt_bboxes, pos_assigned_gt_inds + + def __call__(self, samples, context=None): + assert len(samples) > 0 + batch_size = len(samples) + # get grid cells of image + h, w = samples[0]['image'].shape[1:3] + multi_level_grid_cells = [] + for stride in self.downsample_ratios: + featmap_size = (int(math.ceil(h / stride)), + int(math.ceil(w / stride))) + multi_level_grid_cells.append( + self.get_grid_cells(featmap_size, self.grid_cell_scale, stride, + self.cell_offset)) + mlvl_grid_cells_list = [ + multi_level_grid_cells for i in range(batch_size) + ] + # pixel cell number of multi-level feature maps + num_level_cells = [ + grid_cells.shape[0] for grid_cells in mlvl_grid_cells_list[0] + ] + num_level_cells_list = [num_level_cells] * batch_size + # concat all level cells and to a single array + for i in range(batch_size): + mlvl_grid_cells_list[i] = np.concatenate(mlvl_grid_cells_list[i]) + # target assign on all images + for sample, grid_cells, num_level_cells in zip( + samples, mlvl_grid_cells_list, num_level_cells_list): + gt_bboxes = sample['gt_bbox'] + gt_labels = sample['gt_class'].squeeze() + if gt_labels.size == 1: + gt_labels = np.array([gt_labels]).astype(np.int32) + gt_bboxes_ignore = None + assign_gt_inds, _ = self.assigner(grid_cells, num_level_cells, + gt_bboxes, gt_bboxes_ignore, + gt_labels) + pos_inds, neg_inds, pos_gt_bboxes, pos_assigned_gt_inds = self.get_sample( + assign_gt_inds, gt_bboxes) + + num_cells = grid_cells.shape[0] + bbox_targets = np.zeros_like(grid_cells) + bbox_weights = np.zeros_like(grid_cells) + labels = np.ones([num_cells], dtype=np.int64) * self.num_classes + label_weights = np.zeros([num_cells], dtype=np.float32) + + if len(pos_inds) > 0: + pos_bbox_targets = pos_gt_bboxes + bbox_targets[pos_inds, :] = pos_bbox_targets + bbox_weights[pos_inds, :] = 1.0 + if not np.any(gt_labels): + labels[pos_inds] = 0 + else: + labels[pos_inds] = gt_labels[pos_assigned_gt_inds] + + label_weights[pos_inds] = 1.0 + if len(neg_inds) > 0: + label_weights[neg_inds] = 1.0 + sample['grid_cells'] = grid_cells + sample['labels'] = labels + sample['label_weights'] = label_weights + sample['bbox_targets'] = bbox_targets + sample['pos_num'] = max(pos_inds.size, 1) + sample.pop('is_crowd', None) + sample.pop('difficult', None) + sample.pop('gt_class', None) + sample.pop('gt_bbox', None) + sample.pop('gt_score', None) + return samples + + +@register_op +class Gt2TTFTarget(BaseOperator): + __shared__ = ['num_classes'] + """ + Gt2TTFTarget + Generate TTFNet targets by ground truth data + + Args: + num_classes(int): the number of classes. + down_ratio(int): the down ratio from images to heatmap, 4 by default. + alpha(float): the alpha parameter to generate gaussian target. + 0.54 by default. + """ + + def __init__(self, num_classes=80, down_ratio=4, alpha=0.54): + super(Gt2TTFTarget, self).__init__() + self.down_ratio = down_ratio + self.num_classes = num_classes + self.alpha = alpha + + def __call__(self, samples, context=None): + output_size = samples[0]['image'].shape[1] + feat_size = output_size // self.down_ratio + for sample in samples: + heatmap = np.zeros( + (self.num_classes, feat_size, feat_size), dtype='float32') + box_target = np.ones( + (4, feat_size, feat_size), dtype='float32') * -1 + reg_weight = np.zeros((1, feat_size, feat_size), dtype='float32') + + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + + bbox_w = gt_bbox[:, 2] - gt_bbox[:, 0] + 1 + bbox_h = gt_bbox[:, 3] - gt_bbox[:, 1] + 1 + area = bbox_w * bbox_h + boxes_areas_log = np.log(area) + boxes_ind = np.argsort(boxes_areas_log, axis=0)[::-1] + boxes_area_topk_log = boxes_areas_log[boxes_ind] + gt_bbox = gt_bbox[boxes_ind] + gt_class = gt_class[boxes_ind] + + feat_gt_bbox = gt_bbox / self.down_ratio + feat_gt_bbox = np.clip(feat_gt_bbox, 0, feat_size - 1) + feat_hs, feat_ws = (feat_gt_bbox[:, 3] - feat_gt_bbox[:, 1], + feat_gt_bbox[:, 2] - feat_gt_bbox[:, 0]) + + ct_inds = np.stack( + [(gt_bbox[:, 0] + gt_bbox[:, 2]) / 2, + (gt_bbox[:, 1] + gt_bbox[:, 3]) / 2], + axis=1) / self.down_ratio + + h_radiuses_alpha = (feat_hs / 2. * self.alpha).astype('int32') + w_radiuses_alpha = (feat_ws / 2. * self.alpha).astype('int32') + + for k in range(len(gt_bbox)): + cls_id = gt_class[k] + fake_heatmap = np.zeros((feat_size, feat_size), dtype='float32') + self.draw_truncate_gaussian(fake_heatmap, ct_inds[k], + h_radiuses_alpha[k], + w_radiuses_alpha[k]) + + heatmap[cls_id] = np.maximum(heatmap[cls_id], fake_heatmap) + box_target_inds = fake_heatmap > 0 + box_target[:, box_target_inds] = gt_bbox[k][:, None] + + local_heatmap = fake_heatmap[box_target_inds] + ct_div = np.sum(local_heatmap) + local_heatmap *= boxes_area_topk_log[k] + reg_weight[0, box_target_inds] = local_heatmap / ct_div + sample['ttf_heatmap'] = heatmap + sample['ttf_box_target'] = box_target + sample['ttf_reg_weight'] = reg_weight + sample.pop('is_crowd', None) + sample.pop('difficult', None) + sample.pop('gt_class', None) + sample.pop('gt_bbox', None) + sample.pop('gt_score', None) + return samples + + def draw_truncate_gaussian(self, heatmap, center, h_radius, w_radius): + h, w = 2 * h_radius + 1, 2 * w_radius + 1 + sigma_x = w / 6 + sigma_y = h / 6 + gaussian = gaussian2D((h, w), sigma_x, sigma_y) + + x, y = int(center[0]), int(center[1]) + + height, width = heatmap.shape[0:2] + + left, right = min(x, w_radius), min(width - x, w_radius + 1) + top, bottom = min(y, h_radius), min(height - y, h_radius + 1) + + masked_heatmap = heatmap[y - top:y + bottom, x - left:x + right] + masked_gaussian = gaussian[h_radius - top:h_radius + bottom, w_radius - + left:w_radius + right] + if min(masked_gaussian.shape) > 0 and min(masked_heatmap.shape) > 0: + heatmap[y - top:y + bottom, x - left:x + right] = np.maximum( + masked_heatmap, masked_gaussian) + return heatmap + + +@register_op +class Gt2Solov2Target(BaseOperator): + """Assign mask target and labels in SOLOv2 network. + The code of this function is based on: + https://github.com/WXinlong/SOLO/blob/master/mmdet/models/anchor_heads/solov2_head.py#L271 + Args: + num_grids (list): The list of feature map grids size. + scale_ranges (list): The list of mask boundary range. + coord_sigma (float): The coefficient of coordinate area length. + sampling_ratio (float): The ratio of down sampling. + """ + + def __init__(self, + num_grids=[40, 36, 24, 16, 12], + scale_ranges=[[1, 96], [48, 192], [96, 384], [192, 768], + [384, 2048]], + coord_sigma=0.2, + sampling_ratio=4.0): + super(Gt2Solov2Target, self).__init__() + self.num_grids = num_grids + self.scale_ranges = scale_ranges + self.coord_sigma = coord_sigma + self.sampling_ratio = sampling_ratio + + def _scale_size(self, im, scale): + h, w = im.shape[:2] + new_size = (int(w * float(scale) + 0.5), int(h * float(scale) + 0.5)) + resized_img = cv2.resize( + im, None, None, fx=scale, fy=scale, interpolation=cv2.INTER_LINEAR) + return resized_img + + def __call__(self, samples, context=None): + sample_id = 0 + max_ins_num = [0] * len(self.num_grids) + for sample in samples: + gt_bboxes_raw = sample['gt_bbox'] + gt_labels_raw = sample['gt_class'] + 1 + im_c, im_h, im_w = sample['image'].shape[:] + gt_masks_raw = sample['gt_segm'].astype(np.uint8) + mask_feat_size = [ + int(im_h / self.sampling_ratio), int(im_w / self.sampling_ratio) + ] + gt_areas = np.sqrt((gt_bboxes_raw[:, 2] - gt_bboxes_raw[:, 0]) * + (gt_bboxes_raw[:, 3] - gt_bboxes_raw[:, 1])) + ins_ind_label_list = [] + idx = 0 + for (lower_bound, upper_bound), num_grid \ + in zip(self.scale_ranges, self.num_grids): + + hit_indices = ((gt_areas >= lower_bound) & + (gt_areas <= upper_bound)).nonzero()[0] + num_ins = len(hit_indices) + + ins_label = [] + grid_order = [] + cate_label = np.zeros([num_grid, num_grid], dtype=np.int64) + ins_ind_label = np.zeros([num_grid**2], dtype=np.bool) + + if num_ins == 0: + ins_label = np.zeros( + [1, mask_feat_size[0], mask_feat_size[1]], + dtype=np.uint8) + ins_ind_label_list.append(ins_ind_label) + sample['cate_label{}'.format(idx)] = cate_label.flatten() + sample['ins_label{}'.format(idx)] = ins_label + sample['grid_order{}'.format(idx)] = np.asarray( + [sample_id * num_grid * num_grid + 0], dtype=np.int32) + idx += 1 + continue + gt_bboxes = gt_bboxes_raw[hit_indices] + gt_labels = gt_labels_raw[hit_indices] + gt_masks = gt_masks_raw[hit_indices, ...] + + half_ws = 0.5 * ( + gt_bboxes[:, 2] - gt_bboxes[:, 0]) * self.coord_sigma + half_hs = 0.5 * ( + gt_bboxes[:, 3] - gt_bboxes[:, 1]) * self.coord_sigma + + for seg_mask, gt_label, half_h, half_w in zip( + gt_masks, gt_labels, half_hs, half_ws): + if seg_mask.sum() == 0: + continue + # mass center + upsampled_size = (mask_feat_size[0] * 4, + mask_feat_size[1] * 4) + center_h, center_w = ndimage.measurements.center_of_mass( + seg_mask) + coord_w = int( + (center_w / upsampled_size[1]) // (1. / num_grid)) + coord_h = int( + (center_h / upsampled_size[0]) // (1. / num_grid)) + + # left, top, right, down + top_box = max(0, + int(((center_h - half_h) / upsampled_size[0]) + // (1. / num_grid))) + down_box = min(num_grid - 1, + int(((center_h + half_h) / upsampled_size[0]) + // (1. / num_grid))) + left_box = max(0, + int(((center_w - half_w) / upsampled_size[1]) + // (1. / num_grid))) + right_box = min(num_grid - 1, + int(((center_w + half_w) / + upsampled_size[1]) // (1. / num_grid))) + + top = max(top_box, coord_h - 1) + down = min(down_box, coord_h + 1) + left = max(coord_w - 1, left_box) + right = min(right_box, coord_w + 1) + + cate_label[top:(down + 1), left:(right + 1)] = gt_label + seg_mask = self._scale_size( + seg_mask, scale=1. / self.sampling_ratio) + for i in range(top, down + 1): + for j in range(left, right + 1): + label = int(i * num_grid + j) + cur_ins_label = np.zeros( + [mask_feat_size[0], mask_feat_size[1]], + dtype=np.uint8) + cur_ins_label[:seg_mask.shape[0], :seg_mask.shape[ + 1]] = seg_mask + ins_label.append(cur_ins_label) + ins_ind_label[label] = True + grid_order.append(sample_id * num_grid * num_grid + + label) + if ins_label == []: + ins_label = np.zeros( + [1, mask_feat_size[0], mask_feat_size[1]], + dtype=np.uint8) + ins_ind_label_list.append(ins_ind_label) + sample['cate_label{}'.format(idx)] = cate_label.flatten() + sample['ins_label{}'.format(idx)] = ins_label + sample['grid_order{}'.format(idx)] = np.asarray( + [sample_id * num_grid * num_grid + 0], dtype=np.int32) + else: + ins_label = np.stack(ins_label, axis=0) + ins_ind_label_list.append(ins_ind_label) + sample['cate_label{}'.format(idx)] = cate_label.flatten() + sample['ins_label{}'.format(idx)] = ins_label + sample['grid_order{}'.format(idx)] = np.asarray( + grid_order, dtype=np.int32) + assert len(grid_order) > 0 + max_ins_num[idx] = max( + max_ins_num[idx], + sample['ins_label{}'.format(idx)].shape[0]) + idx += 1 + ins_ind_labels = np.concatenate([ + ins_ind_labels_level_img + for ins_ind_labels_level_img in ins_ind_label_list + ]) + fg_num = np.sum(ins_ind_labels) + sample['fg_num'] = fg_num + sample_id += 1 + + sample.pop('is_crowd') + sample.pop('gt_class') + sample.pop('gt_bbox') + sample.pop('gt_poly') + sample.pop('gt_segm') + + # padding batch + for data in samples: + for idx in range(len(self.num_grids)): + gt_ins_data = np.zeros( + [ + max_ins_num[idx], + data['ins_label{}'.format(idx)].shape[1], + data['ins_label{}'.format(idx)].shape[2] + ], + dtype=np.uint8) + gt_ins_data[0:data['ins_label{}'.format(idx)].shape[ + 0], :, :] = data['ins_label{}'.format(idx)] + gt_grid_order = np.zeros([max_ins_num[idx]], dtype=np.int32) + gt_grid_order[0:data['grid_order{}'.format(idx)].shape[ + 0]] = data['grid_order{}'.format(idx)] + data['ins_label{}'.format(idx)] = gt_ins_data + data['grid_order{}'.format(idx)] = gt_grid_order + + return samples + + +@register_op +class Gt2SparseRCNNTarget(BaseOperator): + ''' + Generate SparseRCNN targets by groud truth data + ''' + + def __init__(self): + super(Gt2SparseRCNNTarget, self).__init__() + + def __call__(self, samples, context=None): + for sample in samples: + im = sample["image"] + h, w = im.shape[1:3] + img_whwh = np.array([w, h, w, h], dtype=np.int32) + sample["img_whwh"] = img_whwh + if "scale_factor" in sample: + sample["scale_factor_wh"] = np.array( + [sample["scale_factor"][1], sample["scale_factor"][0]], + dtype=np.float32) + else: + sample["scale_factor_wh"] = np.array( + [1.0, 1.0], dtype=np.float32) + + return samples + + +@register_op +class PadMaskBatch(BaseOperator): + """ + Pad a batch of samples so they can be divisible by a stride. + The layout of each image should be 'CHW'. + Args: + pad_to_stride (int): If `pad_to_stride > 0`, pad zeros to ensure + height and width is divisible by `pad_to_stride`. + return_pad_mask (bool): If `return_pad_mask = True`, return + `pad_mask` for transformer. + """ + + def __init__(self, pad_to_stride=0, return_pad_mask=False): + super(PadMaskBatch, self).__init__() + self.pad_to_stride = pad_to_stride + self.return_pad_mask = return_pad_mask + + def __call__(self, samples, context=None): + """ + Args: + samples (list): a batch of sample, each is dict. + """ + coarsest_stride = self.pad_to_stride + + max_shape = np.array([data['image'].shape for data in samples]).max( + axis=0) + if coarsest_stride > 0: + max_shape[1] = int( + np.ceil(max_shape[1] / coarsest_stride) * coarsest_stride) + max_shape[2] = int( + np.ceil(max_shape[2] / coarsest_stride) * coarsest_stride) + + for data in samples: + im = data['image'] + im_c, im_h, im_w = im.shape[:] + padding_im = np.zeros( + (im_c, max_shape[1], max_shape[2]), dtype=np.float32) + padding_im[:, :im_h, :im_w] = im + data['image'] = padding_im + if 'semantic' in data and data['semantic'] is not None: + semantic = data['semantic'] + padding_sem = np.zeros( + (1, max_shape[1], max_shape[2]), dtype=np.float32) + padding_sem[:, :im_h, :im_w] = semantic + data['semantic'] = padding_sem + if 'gt_segm' in data and data['gt_segm'] is not None: + gt_segm = data['gt_segm'] + padding_segm = np.zeros( + (gt_segm.shape[0], max_shape[1], max_shape[2]), + dtype=np.uint8) + padding_segm[:, :im_h, :im_w] = gt_segm + data['gt_segm'] = padding_segm + if self.return_pad_mask: + padding_mask = np.zeros( + (max_shape[1], max_shape[2]), dtype=np.float32) + padding_mask[:im_h, :im_w] = 1. + data['pad_mask'] = padding_mask + + if 'gt_rbox2poly' in data and data['gt_rbox2poly'] is not None: + # ploy to rbox + polys = data['gt_rbox2poly'] + rbox = bbox_utils.poly2rbox(polys) + data['gt_rbox'] = rbox + + return samples + + +@register_op +class Gt2CenterNetTarget(BaseOperator): + """Gt2CenterNetTarget + Genterate CenterNet targets by ground-truth + Args: + down_ratio (int): The down sample ratio between output feature and + input image. + num_classes (int): The number of classes, 80 by default. + max_objs (int): The maximum objects detected, 128 by default. + """ + + def __init__(self, down_ratio, num_classes=80, max_objs=128): + super(Gt2CenterNetTarget, self).__init__() + self.down_ratio = down_ratio + self.num_classes = num_classes + self.max_objs = max_objs + + def __call__(self, sample, context=None): + input_h, input_w = sample['image'].shape[1:] + output_h = input_h // self.down_ratio + output_w = input_w // self.down_ratio + num_classes = self.num_classes + c = sample['center'] + s = sample['scale'] + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + + hm = np.zeros((num_classes, output_h, output_w), dtype=np.float32) + wh = np.zeros((self.max_objs, 2), dtype=np.float32) + dense_wh = np.zeros((2, output_h, output_w), dtype=np.float32) + reg = np.zeros((self.max_objs, 2), dtype=np.float32) + ind = np.zeros((self.max_objs), dtype=np.int64) + reg_mask = np.zeros((self.max_objs), dtype=np.int32) + cat_spec_wh = np.zeros( + (self.max_objs, num_classes * 2), dtype=np.float32) + cat_spec_mask = np.zeros( + (self.max_objs, num_classes * 2), dtype=np.int32) + + trans_output = get_affine_transform(c, [s, s], 0, [output_w, output_h]) + + gt_det = [] + for i, (bbox, cls) in enumerate(zip(gt_bbox, gt_class)): + cls = int(cls) + bbox[:2] = affine_transform(bbox[:2], trans_output) + bbox[2:] = affine_transform(bbox[2:], trans_output) + bbox[[0, 2]] = np.clip(bbox[[0, 2]], 0, output_w - 1) + bbox[[1, 3]] = np.clip(bbox[[1, 3]], 0, output_h - 1) + h, w = bbox[3] - bbox[1], bbox[2] - bbox[0] + if h > 0 and w > 0: + radius = gaussian_radius((math.ceil(h), math.ceil(w)), 0.7) + radius = max(0, int(radius)) + ct = np.array( + [(bbox[0] + bbox[2]) / 2, (bbox[1] + bbox[3]) / 2], + dtype=np.float32) + ct_int = ct.astype(np.int32) + draw_umich_gaussian(hm[cls], ct_int, radius) + wh[i] = 1. * w, 1. * h + ind[i] = ct_int[1] * output_w + ct_int[0] + reg[i] = ct - ct_int + reg_mask[i] = 1 + cat_spec_wh[i, cls * 2:cls * 2 + 2] = wh[i] + cat_spec_mask[i, cls * 2:cls * 2 + 2] = 1 + gt_det.append([ + ct[0] - w / 2, ct[1] - h / 2, ct[0] + w / 2, ct[1] + h / 2, + 1, cls + ]) + + sample.pop('gt_bbox', None) + sample.pop('gt_class', None) + sample.pop('center', None) + sample.pop('scale', None) + sample.pop('is_crowd', None) + sample.pop('difficult', None) + sample['heatmap'] = hm + sample['index_mask'] = reg_mask + sample['index'] = ind + sample['size'] = wh + sample['offset'] = reg + return sample diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/gridmask_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/gridmask_utils.py new file mode 100644 index 000000000..c18701556 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/gridmask_utils.py @@ -0,0 +1,86 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/dvlab-research/GridMask/blob/master/detection_grid/maskrcnn_benchmark/data/transforms/grid.py + +from __future__ import absolute_import +from __future__ import print_function +from __future__ import division + +import numpy as np +from PIL import Image + + +class Gridmask(object): + def __init__(self, + use_h=True, + use_w=True, + rotate=1, + offset=False, + ratio=0.5, + mode=1, + prob=0.7, + upper_iter=360000): + super(Gridmask, self).__init__() + self.use_h = use_h + self.use_w = use_w + self.rotate = rotate + self.offset = offset + self.ratio = ratio + self.mode = mode + self.prob = prob + self.st_prob = prob + self.upper_iter = upper_iter + + def __call__(self, x, curr_iter): + self.prob = self.st_prob * min(1, 1.0 * curr_iter / self.upper_iter) + if np.random.rand() > self.prob: + return x + h, w, _ = x.shape + hh = int(1.5 * h) + ww = int(1.5 * w) + d = np.random.randint(2, h) + self.l = min(max(int(d * self.ratio + 0.5), 1), d - 1) + mask = np.ones((hh, ww), np.float32) + st_h = np.random.randint(d) + st_w = np.random.randint(d) + if self.use_h: + for i in range(hh // d): + s = d * i + st_h + t = min(s + self.l, hh) + mask[s:t, :] *= 0 + if self.use_w: + for i in range(ww // d): + s = d * i + st_w + t = min(s + self.l, ww) + mask[:, s:t] *= 0 + + r = np.random.randint(self.rotate) + mask = Image.fromarray(np.uint8(mask)) + mask = mask.rotate(r) + mask = np.asarray(mask) + mask = mask[(hh - h) // 2:(hh - h) // 2 + h, (ww - w) // 2:(ww - w) // 2 + + w].astype(np.float32) + + if self.mode == 1: + mask = 1 - mask + mask = np.expand_dims(mask, axis=-1) + if self.offset: + offset = (2 * (np.random.rand(h, w) - 0.5)).astype(np.float32) + x = (x * mask + offset * (1 - mask)).astype(x.dtype) + else: + x = (x * mask).astype(x.dtype) + + return x diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/keypoint_operators.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/keypoint_operators.py new file mode 100644 index 000000000..81770b63e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/keypoint_operators.py @@ -0,0 +1,859 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# function: +# operators to process sample, +# eg: decode/resize/crop image + +from __future__ import absolute_import + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence + +import cv2 +import numpy as np +import math +import copy + +from ...modeling.keypoint_utils import get_affine_mat_kernel, warp_affine_joints, get_affine_transform, affine_transform, get_warp_matrix +from ppdet.core.workspace import serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +registered_ops = [] + +__all__ = [ + 'RandomAffine', + 'KeyPointFlip', + 'TagGenerate', + 'ToHeatmaps', + 'NormalizePermute', + 'EvalAffine', + 'RandomFlipHalfBodyTransform', + 'TopDownAffine', + 'ToHeatmapsTopDown', + 'ToHeatmapsTopDown_DARK', + 'ToHeatmapsTopDown_UDP', + 'TopDownEvalAffine', + 'AugmentationbyInformantionDropping', +] + + +def register_keypointop(cls): + return serializable(cls) + + +@register_keypointop +class KeyPointFlip(object): + """Get the fliped image by flip_prob. flip the coords also + the left coords and right coords should exchange while flip, for the right keypoint will be left keypoint after image fliped + + Args: + flip_permutation (list[17]): the left-right exchange order list corresponding to [0,1,2,...,16] + hmsize (list[2]): output heatmap's shape list of different scale outputs of higherhrnet + flip_prob (float): the ratio whether to flip the image + records(dict): the dict contained the image, mask and coords + + Returns: + records(dict): contain the image, mask and coords after tranformed + + """ + + def __init__(self, flip_permutation, hmsize, flip_prob=0.5): + super(KeyPointFlip, self).__init__() + assert isinstance(flip_permutation, Sequence) + self.flip_permutation = flip_permutation + self.flip_prob = flip_prob + self.hmsize = hmsize + + def __call__(self, records): + image = records['image'] + kpts_lst = records['joints'] + mask_lst = records['mask'] + flip = np.random.random() < self.flip_prob + if flip: + image = image[:, ::-1] + for idx, hmsize in enumerate(self.hmsize): + if len(mask_lst) > idx: + mask_lst[idx] = mask_lst[idx][:, ::-1] + if kpts_lst[idx].ndim == 3: + kpts_lst[idx] = kpts_lst[idx][:, self.flip_permutation] + else: + kpts_lst[idx] = kpts_lst[idx][self.flip_permutation] + kpts_lst[idx][..., 0] = hmsize - kpts_lst[idx][..., 0] + kpts_lst[idx] = kpts_lst[idx].astype(np.int64) + kpts_lst[idx][kpts_lst[idx][..., 0] >= hmsize, 2] = 0 + kpts_lst[idx][kpts_lst[idx][..., 1] >= hmsize, 2] = 0 + kpts_lst[idx][kpts_lst[idx][..., 0] < 0, 2] = 0 + kpts_lst[idx][kpts_lst[idx][..., 1] < 0, 2] = 0 + records['image'] = image + records['joints'] = kpts_lst + records['mask'] = mask_lst + return records + + +@register_keypointop +class RandomAffine(object): + """apply affine transform to image, mask and coords + to achieve the rotate, scale and shift effect for training image + + Args: + max_degree (float): the max abslute rotate degree to apply, transform range is [-max_degree, max_degree] + max_scale (list[2]): the scale range to apply, transform range is [min, max] + max_shift (float): the max abslute shift ratio to apply, transform range is [-max_shift*imagesize, max_shift*imagesize] + hmsize (list[2]): output heatmap's shape list of different scale outputs of higherhrnet + trainsize (int): the standard length used to train, the 'scale_type' of [h,w] will be resize to trainsize for standard + scale_type (str): the length of [h,w] to used for trainsize, chosed between 'short' and 'long' + records(dict): the dict contained the image, mask and coords + + Returns: + records(dict): contain the image, mask and coords after tranformed + + """ + + def __init__(self, + max_degree=30, + scale=[0.75, 1.5], + max_shift=0.2, + hmsize=[128, 256], + trainsize=512, + scale_type='short'): + super(RandomAffine, self).__init__() + self.max_degree = max_degree + self.min_scale = scale[0] + self.max_scale = scale[1] + self.max_shift = max_shift + self.hmsize = hmsize + self.trainsize = trainsize + self.scale_type = scale_type + + def _get_affine_matrix(self, center, scale, res, rot=0): + """Generate transformation matrix.""" + h = scale + t = np.zeros((3, 3), dtype=np.float32) + t[0, 0] = float(res[1]) / h + t[1, 1] = float(res[0]) / h + t[0, 2] = res[1] * (-float(center[0]) / h + .5) + t[1, 2] = res[0] * (-float(center[1]) / h + .5) + t[2, 2] = 1 + if rot != 0: + rot = -rot # To match direction of rotation from cropping + rot_mat = np.zeros((3, 3), dtype=np.float32) + rot_rad = rot * np.pi / 180 + sn, cs = np.sin(rot_rad), np.cos(rot_rad) + rot_mat[0, :2] = [cs, -sn] + rot_mat[1, :2] = [sn, cs] + rot_mat[2, 2] = 1 + # Need to rotate around center + t_mat = np.eye(3) + t_mat[0, 2] = -res[1] / 2 + t_mat[1, 2] = -res[0] / 2 + t_inv = t_mat.copy() + t_inv[:2, 2] *= -1 + t = np.dot(t_inv, np.dot(rot_mat, np.dot(t_mat, t))) + return t + + def __call__(self, records): + image = records['image'] + keypoints = records['joints'] + heatmap_mask = records['mask'] + + degree = (np.random.random() * 2 - 1) * self.max_degree + shape = np.array(image.shape[:2][::-1]) + center = center = np.array((np.array(shape) / 2)) + + aug_scale = np.random.random() * (self.max_scale - self.min_scale + ) + self.min_scale + if self.scale_type == 'long': + scale = max(shape[0], shape[1]) / 1.0 + elif self.scale_type == 'short': + scale = min(shape[0], shape[1]) / 1.0 + else: + raise ValueError('Unknown scale type: {}'.format(self.scale_type)) + roi_size = aug_scale * scale + dx = int(0) + dy = int(0) + if self.max_shift > 0: + + dx = np.random.randint(-self.max_shift * roi_size, + self.max_shift * roi_size) + dy = np.random.randint(-self.max_shift * roi_size, + self.max_shift * roi_size) + + center += np.array([dx, dy]) + input_size = 2 * center + + keypoints[..., :2] *= shape + heatmap_mask *= 255 + kpts_lst = [] + mask_lst = [] + + image_affine_mat = self._get_affine_matrix( + center, roi_size, (self.trainsize, self.trainsize), degree)[:2] + image = cv2.warpAffine( + image, + image_affine_mat, (self.trainsize, self.trainsize), + flags=cv2.INTER_LINEAR) + for hmsize in self.hmsize: + kpts = copy.deepcopy(keypoints) + mask_affine_mat = self._get_affine_matrix( + center, roi_size, (hmsize, hmsize), degree)[:2] + if heatmap_mask is not None: + mask = cv2.warpAffine(heatmap_mask, mask_affine_mat, + (hmsize, hmsize)) + mask = ((mask / 255) > 0.5).astype(np.float32) + kpts[..., 0:2] = warp_affine_joints(kpts[..., 0:2].copy(), + mask_affine_mat) + kpts[np.trunc(kpts[..., 0]) >= hmsize, 2] = 0 + kpts[np.trunc(kpts[..., 1]) >= hmsize, 2] = 0 + kpts[np.trunc(kpts[..., 0]) < 0, 2] = 0 + kpts[np.trunc(kpts[..., 1]) < 0, 2] = 0 + kpts_lst.append(kpts) + mask_lst.append(mask) + records['image'] = image + records['joints'] = kpts_lst + records['mask'] = mask_lst + return records + + +@register_keypointop +class EvalAffine(object): + """apply affine transform to image + resize the short of [h,w] to standard size for eval + + Args: + size (int): the standard length used to train, the 'short' of [h,w] will be resize to trainsize for standard + records(dict): the dict contained the image, mask and coords + + Returns: + records(dict): contain the image, mask and coords after tranformed + + """ + + def __init__(self, size, stride=64): + super(EvalAffine, self).__init__() + self.size = size + self.stride = stride + + def __call__(self, records): + image = records['image'] + mask = records['mask'] if 'mask' in records else None + s = self.size + h, w, _ = image.shape + trans, size_resized = get_affine_mat_kernel(h, w, s, inv=False) + image_resized = cv2.warpAffine(image, trans, size_resized) + if mask is not None: + mask = cv2.warpAffine(mask, trans, size_resized) + records['mask'] = mask + if 'joints' in records: + del records['joints'] + records['image'] = image_resized + return records + + +@register_keypointop +class NormalizePermute(object): + def __init__(self, + mean=[123.675, 116.28, 103.53], + std=[58.395, 57.120, 57.375], + is_scale=True): + super(NormalizePermute, self).__init__() + self.mean = mean + self.std = std + self.is_scale = is_scale + + def __call__(self, records): + image = records['image'] + image = image.astype(np.float32) + if self.is_scale: + image /= 255. + image = image.transpose((2, 0, 1)) + mean = np.array(self.mean, dtype=np.float32) + std = np.array(self.std, dtype=np.float32) + invstd = 1. / std + for v, m, s in zip(image, mean, invstd): + v.__isub__(m).__imul__(s) + records['image'] = image + return records + + +@register_keypointop +class TagGenerate(object): + """record gt coords for aeloss to sample coords value in tagmaps + + Args: + num_joints (int): the keypoint numbers of dataset to train + num_people (int): maxmum people to support for sample aeloss + records(dict): the dict contained the image, mask and coords + + Returns: + records(dict): contain the gt coords used in tagmap + + """ + + def __init__(self, num_joints, max_people=30): + super(TagGenerate, self).__init__() + self.max_people = max_people + self.num_joints = num_joints + + def __call__(self, records): + kpts_lst = records['joints'] + kpts = kpts_lst[0] + tagmap = np.zeros((self.max_people, self.num_joints, 4), dtype=np.int64) + inds = np.where(kpts[..., 2] > 0) + p, j = inds[0], inds[1] + visible = kpts[inds] + # tagmap is [p, j, 3], where last dim is j, y, x + tagmap[p, j, 0] = j + tagmap[p, j, 1] = visible[..., 1] # y + tagmap[p, j, 2] = visible[..., 0] # x + tagmap[p, j, 3] = 1 + records['tagmap'] = tagmap + del records['joints'] + return records + + +@register_keypointop +class ToHeatmaps(object): + """to generate the gaussin heatmaps of keypoint for heatmap loss + + Args: + num_joints (int): the keypoint numbers of dataset to train + hmsize (list[2]): output heatmap's shape list of different scale outputs of higherhrnet + sigma (float): the std of gaussin kernel genereted + records(dict): the dict contained the image, mask and coords + + Returns: + records(dict): contain the heatmaps used to heatmaploss + + """ + + def __init__(self, num_joints, hmsize, sigma=None): + super(ToHeatmaps, self).__init__() + self.num_joints = num_joints + self.hmsize = np.array(hmsize) + if sigma is None: + sigma = hmsize[0] // 64 + self.sigma = sigma + + r = 6 * sigma + 3 + x = np.arange(0, r, 1, np.float32) + y = x[:, None] + x0, y0 = 3 * sigma + 1, 3 * sigma + 1 + self.gaussian = np.exp(-((x - x0)**2 + (y - y0)**2) / (2 * sigma**2)) + + def __call__(self, records): + kpts_lst = records['joints'] + mask_lst = records['mask'] + for idx, hmsize in enumerate(self.hmsize): + mask = mask_lst[idx] + kpts = kpts_lst[idx] + heatmaps = np.zeros((self.num_joints, hmsize, hmsize)) + inds = np.where(kpts[..., 2] > 0) + visible = kpts[inds].astype(np.int64)[..., :2] + ul = np.round(visible - 3 * self.sigma - 1) + br = np.round(visible + 3 * self.sigma + 2) + sul = np.maximum(0, -ul) + sbr = np.minimum(hmsize, br) - ul + dul = np.clip(ul, 0, hmsize - 1) + dbr = np.clip(br, 0, hmsize) + for i in range(len(visible)): + if visible[i][0] < 0 or visible[i][1] < 0 or visible[i][ + 0] >= hmsize or visible[i][1] >= hmsize: + continue + dx1, dy1 = dul[i] + dx2, dy2 = dbr[i] + sx1, sy1 = sul[i] + sx2, sy2 = sbr[i] + heatmaps[inds[1][i], dy1:dy2, dx1:dx2] = np.maximum( + self.gaussian[sy1:sy2, sx1:sx2], + heatmaps[inds[1][i], dy1:dy2, dx1:dx2]) + records['heatmap_gt{}x'.format(idx + 1)] = heatmaps + records['mask_{}x'.format(idx + 1)] = mask + del records['mask'] + return records + + +@register_keypointop +class RandomFlipHalfBodyTransform(object): + """apply data augment to image and coords + to achieve the flip, scale, rotate and half body transform effect for training image + + Args: + trainsize (list):[w, h], Image target size + upper_body_ids (list): The upper body joint ids + flip_pairs (list): The left-right joints exchange order list + pixel_std (int): The pixel std of the scale + scale (float): The scale factor to transform the image + rot (int): The rotate factor to transform the image + num_joints_half_body (int): The joints threshold of the half body transform + prob_half_body (float): The threshold of the half body transform + flip (bool): Whether to flip the image + + Returns: + records(dict): contain the image and coords after tranformed + + """ + + def __init__(self, + trainsize, + upper_body_ids, + flip_pairs, + pixel_std, + scale=0.35, + rot=40, + num_joints_half_body=8, + prob_half_body=0.3, + flip=True, + rot_prob=0.6): + super(RandomFlipHalfBodyTransform, self).__init__() + self.trainsize = trainsize + self.upper_body_ids = upper_body_ids + self.flip_pairs = flip_pairs + self.pixel_std = pixel_std + self.scale = scale + self.rot = rot + self.num_joints_half_body = num_joints_half_body + self.prob_half_body = prob_half_body + self.flip = flip + self.aspect_ratio = trainsize[0] * 1.0 / trainsize[1] + self.rot_prob = rot_prob + + def halfbody_transform(self, joints, joints_vis): + upper_joints = [] + lower_joints = [] + for joint_id in range(joints.shape[0]): + if joints_vis[joint_id][0] > 0: + if joint_id in self.upper_body_ids: + upper_joints.append(joints[joint_id]) + else: + lower_joints.append(joints[joint_id]) + if np.random.randn() < 0.5 and len(upper_joints) > 2: + selected_joints = upper_joints + else: + selected_joints = lower_joints if len( + lower_joints) > 2 else upper_joints + if len(selected_joints) < 2: + return None, None + selected_joints = np.array(selected_joints, dtype=np.float32) + center = selected_joints.mean(axis=0)[:2] + left_top = np.amin(selected_joints, axis=0) + right_bottom = np.amax(selected_joints, axis=0) + w = right_bottom[0] - left_top[0] + h = right_bottom[1] - left_top[1] + if w > self.aspect_ratio * h: + h = w * 1.0 / self.aspect_ratio + elif w < self.aspect_ratio * h: + w = h * self.aspect_ratio + scale = np.array( + [w * 1.0 / self.pixel_std, h * 1.0 / self.pixel_std], + dtype=np.float32) + scale = scale * 1.5 + + return center, scale + + def flip_joints(self, joints, joints_vis, width, matched_parts): + joints[:, 0] = width - joints[:, 0] - 1 + for pair in matched_parts: + joints[pair[0], :], joints[pair[1], :] = \ + joints[pair[1], :], joints[pair[0], :].copy() + joints_vis[pair[0], :], joints_vis[pair[1], :] = \ + joints_vis[pair[1], :], joints_vis[pair[0], :].copy() + + return joints * joints_vis, joints_vis + + def __call__(self, records): + image = records['image'] + joints = records['joints'] + joints_vis = records['joints_vis'] + c = records['center'] + s = records['scale'] + r = 0 + if (np.sum(joints_vis[:, 0]) > self.num_joints_half_body and + np.random.rand() < self.prob_half_body): + c_half_body, s_half_body = self.halfbody_transform(joints, + joints_vis) + if c_half_body is not None and s_half_body is not None: + c, s = c_half_body, s_half_body + sf = self.scale + rf = self.rot + s = s * np.clip(np.random.randn() * sf + 1, 1 - sf, 1 + sf) + r = np.clip(np.random.randn() * rf, -rf * 2, + rf * 2) if np.random.random() <= self.rot_prob else 0 + + if self.flip and np.random.random() <= 0.5: + image = image[:, ::-1, :] + joints, joints_vis = self.flip_joints( + joints, joints_vis, image.shape[1], self.flip_pairs) + c[0] = image.shape[1] - c[0] - 1 + records['image'] = image + records['joints'] = joints + records['joints_vis'] = joints_vis + records['center'] = c + records['scale'] = s + records['rotate'] = r + + return records + + +@register_keypointop +class AugmentationbyInformantionDropping(object): + """AID: Augmentation by Informantion Dropping. Please refer + to https://arxiv.org/abs/2008.07139 + + Args: + prob_cutout (float): The probability of the Cutout augmentation. + offset_factor (float): Offset factor of cutout center. + num_patch (int): Number of patches to be cutout. + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the image and coords after tranformed + + """ + + def __init__(self, + trainsize, + prob_cutout=0.0, + offset_factor=0.2, + num_patch=1): + self.prob_cutout = prob_cutout + self.offset_factor = offset_factor + self.num_patch = num_patch + self.trainsize = trainsize + + def _cutout(self, img, joints, joints_vis): + height, width, _ = img.shape + img = img.reshape((height * width, -1)) + feat_x_int = np.arange(0, width) + feat_y_int = np.arange(0, height) + feat_x_int, feat_y_int = np.meshgrid(feat_x_int, feat_y_int) + feat_x_int = feat_x_int.reshape((-1, )) + feat_y_int = feat_y_int.reshape((-1, )) + for _ in range(self.num_patch): + vis_idx, _ = np.where(joints_vis > 0) + occlusion_joint_id = np.random.choice(vis_idx) + center = joints[occlusion_joint_id, 0:2] + offset = np.random.randn(2) * self.trainsize[0] * self.offset_factor + center = center + offset + radius = np.random.uniform(0.1, 0.2) * self.trainsize[0] + x_offset = (center[0] - feat_x_int) / radius + y_offset = (center[1] - feat_y_int) / radius + dis = x_offset**2 + y_offset**2 + keep_pos = np.where((dis <= 1) & (dis >= 0))[0] + img[keep_pos, :] = 0 + img = img.reshape((height, width, -1)) + return img + + def __call__(self, records): + img = records['image'] + joints = records['joints'] + joints_vis = records['joints_vis'] + if np.random.rand() < self.prob_cutout: + img = self._cutout(img, joints, joints_vis) + records['image'] = img + return records + + +@register_keypointop +class TopDownAffine(object): + """apply affine transform to image and coords + + Args: + trainsize (list): [w, h], the standard size used to train + use_udp (bool): whether to use Unbiased Data Processing. + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the image and coords after tranformed + + """ + + def __init__(self, trainsize, use_udp=False): + self.trainsize = trainsize + self.use_udp = use_udp + + def __call__(self, records): + image = records['image'] + joints = records['joints'] + joints_vis = records['joints_vis'] + rot = records['rotate'] if "rotate" in records else 0 + if self.use_udp: + trans = get_warp_matrix( + rot, records['center'] * 2.0, + [self.trainsize[0] - 1.0, self.trainsize[1] - 1.0], + records['scale'] * 200.0) + image = cv2.warpAffine( + image, + trans, (int(self.trainsize[0]), int(self.trainsize[1])), + flags=cv2.INTER_LINEAR) + joints[:, 0:2] = warp_affine_joints(joints[:, 0:2].copy(), trans) + else: + trans = get_affine_transform(records['center'], records['scale'] * + 200, rot, self.trainsize) + image = cv2.warpAffine( + image, + trans, (int(self.trainsize[0]), int(self.trainsize[1])), + flags=cv2.INTER_LINEAR) + for i in range(joints.shape[0]): + if joints_vis[i, 0] > 0.0: + joints[i, 0:2] = affine_transform(joints[i, 0:2], trans) + + records['image'] = image + records['joints'] = joints + + return records + + +@register_keypointop +class TopDownEvalAffine(object): + """apply affine transform to image and coords + + Args: + trainsize (list): [w, h], the standard size used to train + use_udp (bool): whether to use Unbiased Data Processing. + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the image and coords after tranformed + + """ + + def __init__(self, trainsize, use_udp=False): + self.trainsize = trainsize + self.use_udp = use_udp + + def __call__(self, records): + image = records['image'] + rot = 0 + imshape = records['im_shape'][::-1] + center = imshape / 2. + scale = imshape + + if self.use_udp: + trans = get_warp_matrix( + rot, center * 2.0, + [self.trainsize[0] - 1.0, self.trainsize[1] - 1.0], scale) + image = cv2.warpAffine( + image, + trans, (int(self.trainsize[0]), int(self.trainsize[1])), + flags=cv2.INTER_LINEAR) + else: + trans = get_affine_transform(center, scale, rot, self.trainsize) + image = cv2.warpAffine( + image, + trans, (int(self.trainsize[0]), int(self.trainsize[1])), + flags=cv2.INTER_LINEAR) + records['image'] = image + + return records + + +@register_keypointop +class ToHeatmapsTopDown(object): + """to generate the gaussin heatmaps of keypoint for heatmap loss + + Args: + hmsize (list): [w, h] output heatmap's size + sigma (float): the std of gaussin kernel genereted + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the heatmaps used to heatmaploss + + """ + + def __init__(self, hmsize, sigma): + super(ToHeatmapsTopDown, self).__init__() + self.hmsize = np.array(hmsize) + self.sigma = sigma + + def __call__(self, records): + joints = records['joints'] + joints_vis = records['joints_vis'] + num_joints = joints.shape[0] + image_size = np.array( + [records['image'].shape[1], records['image'].shape[0]]) + target_weight = np.ones((num_joints, 1), dtype=np.float32) + target_weight[:, 0] = joints_vis[:, 0] + target = np.zeros( + (num_joints, self.hmsize[1], self.hmsize[0]), dtype=np.float32) + tmp_size = self.sigma * 3 + feat_stride = image_size / self.hmsize + for joint_id in range(num_joints): + mu_x = int(joints[joint_id][0] + 0.5) / feat_stride[0] + mu_y = int(joints[joint_id][1] + 0.5) / feat_stride[1] + # Check that any part of the gaussian is in-bounds + ul = [int(mu_x - tmp_size), int(mu_y - tmp_size)] + br = [int(mu_x + tmp_size + 1), int(mu_y + tmp_size + 1)] + if ul[0] >= self.hmsize[0] or ul[1] >= self.hmsize[1] or br[ + 0] < 0 or br[1] < 0: + # If not, just return the image as is + target_weight[joint_id] = 0 + continue + # # Generate gaussian + size = 2 * tmp_size + 1 + x = np.arange(0, size, 1, np.float32) + y = x[:, np.newaxis] + x0 = y0 = size // 2 + # The gaussian is not normalized, we want the center value to equal 1 + g = np.exp(-((x - x0)**2 + (y - y0)**2) / (2 * self.sigma**2)) + + # Usable gaussian range + g_x = max(0, -ul[0]), min(br[0], self.hmsize[0]) - ul[0] + g_y = max(0, -ul[1]), min(br[1], self.hmsize[1]) - ul[1] + # Image range + img_x = max(0, ul[0]), min(br[0], self.hmsize[0]) + img_y = max(0, ul[1]), min(br[1], self.hmsize[1]) + + v = target_weight[joint_id] + if v > 0.5: + target[joint_id][img_y[0]:img_y[1], img_x[0]:img_x[1]] = g[g_y[ + 0]:g_y[1], g_x[0]:g_x[1]] + records['target'] = target + records['target_weight'] = target_weight + del records['joints'], records['joints_vis'] + + return records + + +@register_keypointop +class ToHeatmapsTopDown_DARK(object): + """to generate the gaussin heatmaps of keypoint for heatmap loss + + Args: + hmsize (list): [w, h] output heatmap's size + sigma (float): the std of gaussin kernel genereted + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the heatmaps used to heatmaploss + + """ + + def __init__(self, hmsize, sigma): + super(ToHeatmapsTopDown_DARK, self).__init__() + self.hmsize = np.array(hmsize) + self.sigma = sigma + + def __call__(self, records): + joints = records['joints'] + joints_vis = records['joints_vis'] + num_joints = joints.shape[0] + image_size = np.array( + [records['image'].shape[1], records['image'].shape[0]]) + target_weight = np.ones((num_joints, 1), dtype=np.float32) + target_weight[:, 0] = joints_vis[:, 0] + target = np.zeros( + (num_joints, self.hmsize[1], self.hmsize[0]), dtype=np.float32) + tmp_size = self.sigma * 3 + feat_stride = image_size / self.hmsize + for joint_id in range(num_joints): + mu_x = joints[joint_id][0] / feat_stride[0] + mu_y = joints[joint_id][1] / feat_stride[1] + # Check that any part of the gaussian is in-bounds + ul = [int(mu_x - tmp_size), int(mu_y - tmp_size)] + br = [int(mu_x + tmp_size + 1), int(mu_y + tmp_size + 1)] + if ul[0] >= self.hmsize[0] or ul[1] >= self.hmsize[1] or br[ + 0] < 0 or br[1] < 0: + # If not, just return the image as is + target_weight[joint_id] = 0 + continue + + x = np.arange(0, self.hmsize[0], 1, np.float32) + y = np.arange(0, self.hmsize[1], 1, np.float32) + y = y[:, np.newaxis] + + v = target_weight[joint_id] + if v > 0.5: + target[joint_id] = np.exp(-( + (x - mu_x)**2 + (y - mu_y)**2) / (2 * self.sigma**2)) + records['target'] = target + records['target_weight'] = target_weight + del records['joints'], records['joints_vis'] + + return records + + +@register_keypointop +class ToHeatmapsTopDown_UDP(object): + """to generate the gaussian heatmaps of keypoint for heatmap loss. + ref: Huang et al. The Devil is in the Details: Delving into Unbiased Data Processing + for Human Pose Estimation (CVPR 2020). + + Args: + hmsize (list): [w, h] output heatmap's size + sigma (float): the std of gaussin kernel genereted + records(dict): the dict contained the image and coords + + Returns: + records (dict): contain the heatmaps used to heatmaploss + """ + + def __init__(self, hmsize, sigma): + super(ToHeatmapsTopDown_UDP, self).__init__() + self.hmsize = np.array(hmsize) + self.sigma = sigma + + def __call__(self, records): + joints = records['joints'] + joints_vis = records['joints_vis'] + num_joints = joints.shape[0] + image_size = np.array( + [records['image'].shape[1], records['image'].shape[0]]) + target_weight = np.ones((num_joints, 1), dtype=np.float32) + target_weight[:, 0] = joints_vis[:, 0] + target = np.zeros( + (num_joints, self.hmsize[1], self.hmsize[0]), dtype=np.float32) + tmp_size = self.sigma * 3 + size = 2 * tmp_size + 1 + x = np.arange(0, size, 1, np.float32) + y = x[:, None] + feat_stride = (image_size - 1.0) / (self.hmsize - 1.0) + for joint_id in range(num_joints): + mu_x = int(joints[joint_id][0] / feat_stride[0] + 0.5) + mu_y = int(joints[joint_id][1] / feat_stride[1] + 0.5) + # Check that any part of the gaussian is in-bounds + ul = [int(mu_x - tmp_size), int(mu_y - tmp_size)] + br = [int(mu_x + tmp_size + 1), int(mu_y + tmp_size + 1)] + if ul[0] >= self.hmsize[0] or ul[1] >= self.hmsize[1] or br[ + 0] < 0 or br[1] < 0: + # If not, just return the image as is + target_weight[joint_id] = 0 + continue + + mu_x_ac = joints[joint_id][0] / feat_stride[0] + mu_y_ac = joints[joint_id][1] / feat_stride[1] + x0 = y0 = size // 2 + x0 += mu_x_ac - mu_x + y0 += mu_y_ac - mu_y + g = np.exp(-((x - x0)**2 + (y - y0)**2) / (2 * self.sigma**2)) + # Usable gaussian range + g_x = max(0, -ul[0]), min(br[0], self.hmsize[0]) - ul[0] + g_y = max(0, -ul[1]), min(br[1], self.hmsize[1]) - ul[1] + # Image range + img_x = max(0, ul[0]), min(br[0], self.hmsize[0]) + img_y = max(0, ul[1]), min(br[1], self.hmsize[1]) + + v = target_weight[joint_id] + if v > 0.5: + target[joint_id][img_y[0]:img_y[1], img_x[0]:img_x[1]] = g[g_y[ + 0]:g_y[1], g_x[0]:g_x[1]] + records['target'] = target + records['target_weight'] = target_weight + del records['joints'], records['joints_vis'] + + return records diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/mot_operators.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/mot_operators.py new file mode 100644 index 000000000..ef7d7be45 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/mot_operators.py @@ -0,0 +1,627 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence +from numbers import Integral + +import cv2 +import copy +import numpy as np +import random +import math + +from .operators import BaseOperator, register_op +from .batch_operators import Gt2TTFTarget +from ppdet.modeling.bbox_utils import bbox_iou_np_expand +from ppdet.utils.logger import setup_logger +from .op_helper import gaussian_radius +logger = setup_logger(__name__) + +__all__ = [ + 'RGBReverse', 'LetterBoxResize', 'MOTRandomAffine', 'Gt2JDETargetThres', + 'Gt2JDETargetMax', 'Gt2FairMOTTarget' +] + + +@register_op +class RGBReverse(BaseOperator): + """RGB to BGR, or BGR to RGB, sensitive to MOTRandomAffine + """ + + def __init__(self): + super(RGBReverse, self).__init__() + + def apply(self, sample, context=None): + im = sample['image'] + sample['image'] = np.ascontiguousarray(im[:, :, ::-1]) + return sample + + +@register_op +class LetterBoxResize(BaseOperator): + def __init__(self, target_size): + """ + Resize image to target size, convert normalized xywh to pixel xyxy + format ([x_center, y_center, width, height] -> [x0, y0, x1, y1]). + Args: + target_size (int|list): image target size. + """ + super(LetterBoxResize, self).__init__() + if not isinstance(target_size, (Integral, Sequence)): + raise TypeError( + "Type of target_size is invalid. Must be Integer or List or Tuple, now is {}". + format(type(target_size))) + if isinstance(target_size, Integral): + target_size = [target_size, target_size] + self.target_size = target_size + + def apply_image(self, img, height, width, color=(127.5, 127.5, 127.5)): + # letterbox: resize a rectangular image to a padded rectangular + shape = img.shape[:2] # [height, width] + ratio_h = float(height) / shape[0] + ratio_w = float(width) / shape[1] + ratio = min(ratio_h, ratio_w) + new_shape = (round(shape[1] * ratio), + round(shape[0] * ratio)) # [width, height] + padw = (width - new_shape[0]) / 2 + padh = (height - new_shape[1]) / 2 + top, bottom = round(padh - 0.1), round(padh + 0.1) + left, right = round(padw - 0.1), round(padw + 0.1) + + img = cv2.resize( + img, new_shape, interpolation=cv2.INTER_AREA) # resized, no border + img = cv2.copyMakeBorder( + img, top, bottom, left, right, cv2.BORDER_CONSTANT, + value=color) # padded rectangular + return img, ratio, padw, padh + + def apply_bbox(self, bbox0, h, w, ratio, padw, padh): + bboxes = bbox0.copy() + bboxes[:, 0] = ratio * w * (bbox0[:, 0] - bbox0[:, 2] / 2) + padw + bboxes[:, 1] = ratio * h * (bbox0[:, 1] - bbox0[:, 3] / 2) + padh + bboxes[:, 2] = ratio * w * (bbox0[:, 0] + bbox0[:, 2] / 2) + padw + bboxes[:, 3] = ratio * h * (bbox0[:, 1] + bbox0[:, 3] / 2) + padh + return bboxes + + def apply(self, sample, context=None): + """ Resize the image numpy. + """ + im = sample['image'] + h, w = sample['im_shape'] + if not isinstance(im, np.ndarray): + raise TypeError("{}: image type is not numpy.".format(self)) + if len(im.shape) != 3: + from PIL import UnidentifiedImageError + raise UnidentifiedImageError( + '{}: image is not 3-dimensional.'.format(self)) + + # apply image + height, width = self.target_size + img, ratio, padw, padh = self.apply_image( + im, height=height, width=width) + + sample['image'] = img + new_shape = (round(h * ratio), round(w * ratio)) + sample['im_shape'] = np.asarray(new_shape, dtype=np.float32) + sample['scale_factor'] = np.asarray([ratio, ratio], dtype=np.float32) + + # apply bbox + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], h, w, ratio, + padw, padh) + return sample + + +@register_op +class MOTRandomAffine(BaseOperator): + """ + Affine transform to image and coords to achieve the rotate, scale and + shift effect for training image. + + Args: + degrees (list[2]): the rotate range to apply, transform range is [min, max] + translate (list[2]): the translate range to apply, transform range is [min, max] + scale (list[2]): the scale range to apply, transform range is [min, max] + shear (list[2]): the shear range to apply, transform range is [min, max] + borderValue (list[3]): value used in case of a constant border when appling + the perspective transformation + reject_outside (bool): reject warped bounding bboxes outside of image + + Returns: + records(dict): contain the image and coords after tranformed + + """ + + def __init__(self, + degrees=(-5, 5), + translate=(0.10, 0.10), + scale=(0.50, 1.20), + shear=(-2, 2), + borderValue=(127.5, 127.5, 127.5), + reject_outside=True): + super(MOTRandomAffine, self).__init__() + self.degrees = degrees + self.translate = translate + self.scale = scale + self.shear = shear + self.borderValue = borderValue + self.reject_outside = reject_outside + + def apply(self, sample, context=None): + # https://medium.com/uruvideo/dataset-augmentation-with-random-homographies-a8f4b44830d4 + border = 0 # width of added border (optional) + + img = sample['image'] + height, width = img.shape[0], img.shape[1] + + # Rotation and Scale + R = np.eye(3) + a = random.random() * (self.degrees[1] - self.degrees[0] + ) + self.degrees[0] + s = random.random() * (self.scale[1] - self.scale[0]) + self.scale[0] + R[:2] = cv2.getRotationMatrix2D( + angle=a, center=(width / 2, height / 2), scale=s) + + # Translation + T = np.eye(3) + T[0, 2] = ( + random.random() * 2 - 1 + ) * self.translate[0] * height + border # x translation (pixels) + T[1, 2] = ( + random.random() * 2 - 1 + ) * self.translate[1] * width + border # y translation (pixels) + + # Shear + S = np.eye(3) + S[0, 1] = math.tan((random.random() * + (self.shear[1] - self.shear[0]) + self.shear[0]) * + math.pi / 180) # x shear (deg) + S[1, 0] = math.tan((random.random() * + (self.shear[1] - self.shear[0]) + self.shear[0]) * + math.pi / 180) # y shear (deg) + + M = S @T @R # Combined rotation matrix. ORDER IS IMPORTANT HERE!! + imw = cv2.warpPerspective( + img, + M, + dsize=(width, height), + flags=cv2.INTER_LINEAR, + borderValue=self.borderValue) # BGR order borderValue + + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + targets = sample['gt_bbox'] + n = targets.shape[0] + points = targets.copy() + area0 = (points[:, 2] - points[:, 0]) * ( + points[:, 3] - points[:, 1]) + + # warp points + xy = np.ones((n * 4, 3)) + xy[:, :2] = points[:, [0, 1, 2, 3, 0, 3, 2, 1]].reshape( + n * 4, 2) # x1y1, x2y2, x1y2, x2y1 + xy = (xy @M.T)[:, :2].reshape(n, 8) + + # create new boxes + x = xy[:, [0, 2, 4, 6]] + y = xy[:, [1, 3, 5, 7]] + xy = np.concatenate( + (x.min(1), y.min(1), x.max(1), y.max(1))).reshape(4, n).T + + # apply angle-based reduction + radians = a * math.pi / 180 + reduction = max(abs(math.sin(radians)), abs(math.cos(radians)))**0.5 + x = (xy[:, 2] + xy[:, 0]) / 2 + y = (xy[:, 3] + xy[:, 1]) / 2 + w = (xy[:, 2] - xy[:, 0]) * reduction + h = (xy[:, 3] - xy[:, 1]) * reduction + xy = np.concatenate( + (x - w / 2, y - h / 2, x + w / 2, y + h / 2)).reshape(4, n).T + + # reject warped points outside of image + if self.reject_outside: + np.clip(xy[:, 0], 0, width, out=xy[:, 0]) + np.clip(xy[:, 2], 0, width, out=xy[:, 2]) + np.clip(xy[:, 1], 0, height, out=xy[:, 1]) + np.clip(xy[:, 3], 0, height, out=xy[:, 3]) + w = xy[:, 2] - xy[:, 0] + h = xy[:, 3] - xy[:, 1] + area = w * h + ar = np.maximum(w / (h + 1e-16), h / (w + 1e-16)) + i = (w > 4) & (h > 4) & (area / (area0 + 1e-16) > 0.1) & (ar < 10) + + if sum(i) > 0: + sample['gt_bbox'] = xy[i].astype(sample['gt_bbox'].dtype) + sample['gt_class'] = sample['gt_class'][i] + if 'difficult' in sample: + sample['difficult'] = sample['difficult'][i] + if 'gt_ide' in sample: + sample['gt_ide'] = sample['gt_ide'][i] + if 'is_crowd' in sample: + sample['is_crowd'] = sample['is_crowd'][i] + sample['image'] = imw + return sample + else: + return sample + + +@register_op +class Gt2JDETargetThres(BaseOperator): + __shared__ = ['num_classes'] + """ + Generate JDE targets by groud truth data when training + Args: + anchors (list): anchors of JDE model + anchor_masks (list): anchor_masks of JDE model + downsample_ratios (list): downsample ratios of JDE model + ide_thresh (float): thresh of identity, higher is groud truth + fg_thresh (float): thresh of foreground, higher is foreground + bg_thresh (float): thresh of background, lower is background + num_classes (int): number of classes + """ + + def __init__(self, + anchors, + anchor_masks, + downsample_ratios, + ide_thresh=0.5, + fg_thresh=0.5, + bg_thresh=0.4, + num_classes=1): + super(Gt2JDETargetThres, self).__init__() + self.anchors = anchors + self.anchor_masks = anchor_masks + self.downsample_ratios = downsample_ratios + self.ide_thresh = ide_thresh + self.fg_thresh = fg_thresh + self.bg_thresh = bg_thresh + self.num_classes = num_classes + + def generate_anchor(self, nGh, nGw, anchor_hw): + nA = len(anchor_hw) + yy, xx = np.meshgrid(np.arange(nGh), np.arange(nGw)) + + mesh = np.stack([xx.T, yy.T], axis=0) # [2, nGh, nGw] + mesh = np.repeat(mesh[None, :], nA, axis=0) # [nA, 2, nGh, nGw] + + anchor_offset_mesh = anchor_hw[:, :, None][:, :, :, None] + anchor_offset_mesh = np.repeat(anchor_offset_mesh, nGh, axis=-2) + anchor_offset_mesh = np.repeat(anchor_offset_mesh, nGw, axis=-1) + + anchor_mesh = np.concatenate( + [mesh, anchor_offset_mesh], axis=1) # [nA, 4, nGh, nGw] + return anchor_mesh + + def encode_delta(self, gt_box_list, fg_anchor_list): + px, py, pw, ph = fg_anchor_list[:, 0], fg_anchor_list[:,1], \ + fg_anchor_list[:, 2], fg_anchor_list[:,3] + gx, gy, gw, gh = gt_box_list[:, 0], gt_box_list[:, 1], \ + gt_box_list[:, 2], gt_box_list[:, 3] + dx = (gx - px) / pw + dy = (gy - py) / ph + dw = np.log(gw / pw) + dh = np.log(gh / ph) + return np.stack([dx, dy, dw, dh], axis=1) + + def pad_box(self, sample, num_max): + assert 'gt_bbox' in sample + bbox = sample['gt_bbox'] + gt_num = len(bbox) + pad_bbox = np.zeros((num_max, 4), dtype=np.float32) + if gt_num > 0: + pad_bbox[:gt_num, :] = bbox[:gt_num, :] + sample['gt_bbox'] = pad_bbox + if 'gt_score' in sample: + pad_score = np.zeros((num_max, ), dtype=np.float32) + if gt_num > 0: + pad_score[:gt_num] = sample['gt_score'][:gt_num, 0] + sample['gt_score'] = pad_score + if 'difficult' in sample: + pad_diff = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_diff[:gt_num] = sample['difficult'][:gt_num, 0] + sample['difficult'] = pad_diff + if 'is_crowd' in sample: + pad_crowd = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_crowd[:gt_num] = sample['is_crowd'][:gt_num, 0] + sample['is_crowd'] = pad_crowd + if 'gt_ide' in sample: + pad_ide = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_ide[:gt_num] = sample['gt_ide'][:gt_num, 0] + sample['gt_ide'] = pad_ide + return sample + + def __call__(self, samples, context=None): + assert len(self.anchor_masks) == len(self.downsample_ratios), \ + "anchor_masks', and 'downsample_ratios' should have same length." + h, w = samples[0]['image'].shape[1:3] + + num_max = 0 + for sample in samples: + num_max = max(num_max, len(sample['gt_bbox'])) + + for sample in samples: + gt_bbox = sample['gt_bbox'] + gt_ide = sample['gt_ide'] + for i, (anchor_hw, downsample_ratio + ) in enumerate(zip(self.anchors, self.downsample_ratios)): + anchor_hw = np.array( + anchor_hw, dtype=np.float32) / downsample_ratio + nA = len(anchor_hw) + nGh, nGw = int(h / downsample_ratio), int(w / downsample_ratio) + tbox = np.zeros((nA, nGh, nGw, 4), dtype=np.float32) + tconf = np.zeros((nA, nGh, nGw), dtype=np.float32) + tid = -np.ones((nA, nGh, nGw, 1), dtype=np.float32) + + gxy, gwh = gt_bbox[:, 0:2].copy(), gt_bbox[:, 2:4].copy() + gxy[:, 0] = gxy[:, 0] * nGw + gxy[:, 1] = gxy[:, 1] * nGh + gwh[:, 0] = gwh[:, 0] * nGw + gwh[:, 1] = gwh[:, 1] * nGh + gxy[:, 0] = np.clip(gxy[:, 0], 0, nGw - 1) + gxy[:, 1] = np.clip(gxy[:, 1], 0, nGh - 1) + tboxes = np.concatenate([gxy, gwh], axis=1) + + anchor_mesh = self.generate_anchor(nGh, nGw, anchor_hw) + + anchor_list = np.transpose(anchor_mesh, + (0, 2, 3, 1)).reshape(-1, 4) + iou_pdist = bbox_iou_np_expand( + anchor_list, tboxes, x1y1x2y2=False) + + iou_max = np.max(iou_pdist, axis=1) + max_gt_index = np.argmax(iou_pdist, axis=1) + + iou_map = iou_max.reshape(nA, nGh, nGw) + gt_index_map = max_gt_index.reshape(nA, nGh, nGw) + + id_index = iou_map > self.ide_thresh + fg_index = iou_map > self.fg_thresh + bg_index = iou_map < self.bg_thresh + ign_index = (iou_map < self.fg_thresh) * ( + iou_map > self.bg_thresh) + tconf[fg_index] = 1 + tconf[bg_index] = 0 + tconf[ign_index] = -1 + + gt_index = gt_index_map[fg_index] + gt_box_list = tboxes[gt_index] + gt_id_list = gt_ide[gt_index_map[id_index]] + + if np.sum(fg_index) > 0: + tid[id_index] = gt_id_list + + fg_anchor_list = anchor_list.reshape(nA, nGh, nGw, + 4)[fg_index] + delta_target = self.encode_delta(gt_box_list, + fg_anchor_list) + tbox[fg_index] = delta_target + + sample['tbox{}'.format(i)] = tbox + sample['tconf{}'.format(i)] = tconf + sample['tide{}'.format(i)] = tid + sample.pop('gt_class') + sample = self.pad_box(sample, num_max) + return samples + + +@register_op +class Gt2JDETargetMax(BaseOperator): + __shared__ = ['num_classes'] + """ + Generate JDE targets by groud truth data when evaluating + Args: + anchors (list): anchors of JDE model + anchor_masks (list): anchor_masks of JDE model + downsample_ratios (list): downsample ratios of JDE model + max_iou_thresh (float): iou thresh for high quality anchor + num_classes (int): number of classes + """ + + def __init__(self, + anchors, + anchor_masks, + downsample_ratios, + max_iou_thresh=0.60, + num_classes=1): + super(Gt2JDETargetMax, self).__init__() + self.anchors = anchors + self.anchor_masks = anchor_masks + self.downsample_ratios = downsample_ratios + self.max_iou_thresh = max_iou_thresh + self.num_classes = num_classes + + def __call__(self, samples, context=None): + assert len(self.anchor_masks) == len(self.downsample_ratios), \ + "anchor_masks', and 'downsample_ratios' should have same length." + h, w = samples[0]['image'].shape[1:3] + for sample in samples: + gt_bbox = sample['gt_bbox'] + gt_ide = sample['gt_ide'] + for i, (anchor_hw, downsample_ratio + ) in enumerate(zip(self.anchors, self.downsample_ratios)): + anchor_hw = np.array( + anchor_hw, dtype=np.float32) / downsample_ratio + nA = len(anchor_hw) + nGh, nGw = int(h / downsample_ratio), int(w / downsample_ratio) + tbox = np.zeros((nA, nGh, nGw, 4), dtype=np.float32) + tconf = np.zeros((nA, nGh, nGw), dtype=np.float32) + tid = -np.ones((nA, nGh, nGw, 1), dtype=np.float32) + + gxy, gwh = gt_bbox[:, 0:2].copy(), gt_bbox[:, 2:4].copy() + gxy[:, 0] = gxy[:, 0] * nGw + gxy[:, 1] = gxy[:, 1] * nGh + gwh[:, 0] = gwh[:, 0] * nGw + gwh[:, 1] = gwh[:, 1] * nGh + gi = np.clip(gxy[:, 0], 0, nGw - 1).astype(int) + gj = np.clip(gxy[:, 1], 0, nGh - 1).astype(int) + + # iou of targets-anchors (using wh only) + box1 = gwh + box2 = anchor_hw[:, None, :] + inter_area = np.minimum(box1, box2).prod(2) + iou = inter_area / ( + box1.prod(1) + box2.prod(2) - inter_area + 1e-16) + + # Select best iou_pred and anchor + iou_best = iou.max(0) # best anchor [0-2] for each target + a = np.argmax(iou, axis=0) + + # Select best unique target-anchor combinations + iou_order = np.argsort(-iou_best) # best to worst + + # Unique anchor selection + u = np.stack((gi, gj, a), 0)[:, iou_order] + _, first_unique = np.unique(u, axis=1, return_index=True) + mask = iou_order[first_unique] + # best anchor must share significant commonality (iou) with target + # TODO: examine arbitrary threshold + idx = mask[iou_best[mask] > self.max_iou_thresh] + + if len(idx) > 0: + a_i, gj_i, gi_i = a[idx], gj[idx], gi[idx] + t_box = gt_bbox[idx] + t_id = gt_ide[idx] + if len(t_box.shape) == 1: + t_box = t_box.reshape(1, 4) + + gxy, gwh = t_box[:, 0:2].copy(), t_box[:, 2:4].copy() + gxy[:, 0] = gxy[:, 0] * nGw + gxy[:, 1] = gxy[:, 1] * nGh + gwh[:, 0] = gwh[:, 0] * nGw + gwh[:, 1] = gwh[:, 1] * nGh + + # XY coordinates + tbox[:, :, :, 0:2][a_i, gj_i, gi_i] = gxy - gxy.astype(int) + # Width and height in yolo method + tbox[:, :, :, 2:4][a_i, gj_i, gi_i] = np.log(gwh / + anchor_hw[a_i]) + tconf[a_i, gj_i, gi_i] = 1 + tid[a_i, gj_i, gi_i] = t_id + + sample['tbox{}'.format(i)] = tbox + sample['tconf{}'.format(i)] = tconf + sample['tide{}'.format(i)] = tid + + +class Gt2FairMOTTarget(Gt2TTFTarget): + __shared__ = ['num_classes'] + """ + Generate FairMOT targets by ground truth data. + Difference between Gt2FairMOTTarget and Gt2TTFTarget are: + 1. the gaussian kernal radius to generate a heatmap. + 2. the targets needed during traing. + + Args: + num_classes(int): the number of classes. + down_ratio(int): the down ratio from images to heatmap, 4 by default. + max_objs(int): the maximum number of ground truth objects in a image, 500 by default. + """ + + def __init__(self, num_classes=1, down_ratio=4, max_objs=500): + super(Gt2TTFTarget, self).__init__() + self.down_ratio = down_ratio + self.num_classes = num_classes + self.max_objs = max_objs + + def __call__(self, samples, context=None): + for b_id, sample in enumerate(samples): + output_h = sample['image'].shape[1] // self.down_ratio + output_w = sample['image'].shape[2] // self.down_ratio + + heatmap = np.zeros( + (self.num_classes, output_h, output_w), dtype='float32') + bbox_size = np.zeros((self.max_objs, 4), dtype=np.float32) + center_offset = np.zeros((self.max_objs, 2), dtype=np.float32) + index = np.zeros((self.max_objs, ), dtype=np.int64) + index_mask = np.zeros((self.max_objs, ), dtype=np.int32) + reid = np.zeros((self.max_objs, ), dtype=np.int64) + bbox_xys = np.zeros((self.max_objs, 4), dtype=np.float32) + if self.num_classes > 1: + # each category corresponds to a set of track ids + cls_tr_ids = np.zeros( + (self.num_classes, output_h, output_w), dtype=np.int64) + cls_id_map = np.full((output_h, output_w), -1, dtype=np.int64) + + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + gt_ide = sample['gt_ide'] + + for k in range(len(gt_bbox)): + cls_id = gt_class[k][0] + bbox = gt_bbox[k] + ide = gt_ide[k][0] + bbox[[0, 2]] = bbox[[0, 2]] * output_w + bbox[[1, 3]] = bbox[[1, 3]] * output_h + bbox_amodal = copy.deepcopy(bbox) + bbox_amodal[0] = bbox_amodal[0] - bbox_amodal[2] / 2. + bbox_amodal[1] = bbox_amodal[1] - bbox_amodal[3] / 2. + bbox_amodal[2] = bbox_amodal[0] + bbox_amodal[2] + bbox_amodal[3] = bbox_amodal[1] + bbox_amodal[3] + bbox[0] = np.clip(bbox[0], 0, output_w - 1) + bbox[1] = np.clip(bbox[1], 0, output_h - 1) + h = bbox[3] + w = bbox[2] + + bbox_xy = copy.deepcopy(bbox) + bbox_xy[0] = bbox_xy[0] - bbox_xy[2] / 2 + bbox_xy[1] = bbox_xy[1] - bbox_xy[3] / 2 + bbox_xy[2] = bbox_xy[0] + bbox_xy[2] + bbox_xy[3] = bbox_xy[1] + bbox_xy[3] + + if h > 0 and w > 0: + radius = gaussian_radius((math.ceil(h), math.ceil(w)), 0.7) + radius = max(0, int(radius)) + ct = np.array([bbox[0], bbox[1]], dtype=np.float32) + ct_int = ct.astype(np.int32) + self.draw_truncate_gaussian(heatmap[cls_id], ct_int, radius, + radius) + bbox_size[k] = ct[0] - bbox_amodal[0], ct[1] - bbox_amodal[1], \ + bbox_amodal[2] - ct[0], bbox_amodal[3] - ct[1] + + index[k] = ct_int[1] * output_w + ct_int[0] + center_offset[k] = ct - ct_int + index_mask[k] = 1 + reid[k] = ide + bbox_xys[k] = bbox_xy + if self.num_classes > 1: + cls_id_map[ct_int[1], ct_int[0]] = cls_id + cls_tr_ids[cls_id][ct_int[1]][ct_int[0]] = ide - 1 + # track id start from 0 + + sample['heatmap'] = heatmap + sample['index'] = index + sample['offset'] = center_offset + sample['size'] = bbox_size + sample['index_mask'] = index_mask + sample['reid'] = reid + if self.num_classes > 1: + sample['cls_id_map'] = cls_id_map + sample['cls_tr_ids'] = cls_tr_ids + sample['bbox_xys'] = bbox_xys + sample.pop('is_crowd', None) + sample.pop('difficult', None) + sample.pop('gt_class', None) + sample.pop('gt_bbox', None) + sample.pop('gt_score', None) + sample.pop('gt_ide', None) + return samples diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/op_helper.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/op_helper.py new file mode 100644 index 000000000..6c400306d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/op_helper.py @@ -0,0 +1,494 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# this file contains helper methods for BBOX processing + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import numpy as np +import random +import math +import cv2 + + +def meet_emit_constraint(src_bbox, sample_bbox): + center_x = (src_bbox[2] + src_bbox[0]) / 2 + center_y = (src_bbox[3] + src_bbox[1]) / 2 + if center_x >= sample_bbox[0] and \ + center_x <= sample_bbox[2] and \ + center_y >= sample_bbox[1] and \ + center_y <= sample_bbox[3]: + return True + return False + + +def clip_bbox(src_bbox): + src_bbox[0] = max(min(src_bbox[0], 1.0), 0.0) + src_bbox[1] = max(min(src_bbox[1], 1.0), 0.0) + src_bbox[2] = max(min(src_bbox[2], 1.0), 0.0) + src_bbox[3] = max(min(src_bbox[3], 1.0), 0.0) + return src_bbox + + +def bbox_area(src_bbox): + if src_bbox[2] < src_bbox[0] or src_bbox[3] < src_bbox[1]: + return 0. + else: + width = src_bbox[2] - src_bbox[0] + height = src_bbox[3] - src_bbox[1] + return width * height + + +def is_overlap(object_bbox, sample_bbox): + if object_bbox[0] >= sample_bbox[2] or \ + object_bbox[2] <= sample_bbox[0] or \ + object_bbox[1] >= sample_bbox[3] or \ + object_bbox[3] <= sample_bbox[1]: + return False + else: + return True + + +def filter_and_process(sample_bbox, bboxes, labels, scores=None, + keypoints=None): + new_bboxes = [] + new_labels = [] + new_scores = [] + new_keypoints = [] + new_kp_ignore = [] + for i in range(len(bboxes)): + new_bbox = [0, 0, 0, 0] + obj_bbox = [bboxes[i][0], bboxes[i][1], bboxes[i][2], bboxes[i][3]] + if not meet_emit_constraint(obj_bbox, sample_bbox): + continue + if not is_overlap(obj_bbox, sample_bbox): + continue + sample_width = sample_bbox[2] - sample_bbox[0] + sample_height = sample_bbox[3] - sample_bbox[1] + new_bbox[0] = (obj_bbox[0] - sample_bbox[0]) / sample_width + new_bbox[1] = (obj_bbox[1] - sample_bbox[1]) / sample_height + new_bbox[2] = (obj_bbox[2] - sample_bbox[0]) / sample_width + new_bbox[3] = (obj_bbox[3] - sample_bbox[1]) / sample_height + new_bbox = clip_bbox(new_bbox) + if bbox_area(new_bbox) > 0: + new_bboxes.append(new_bbox) + new_labels.append([labels[i][0]]) + if scores is not None: + new_scores.append([scores[i][0]]) + if keypoints is not None: + sample_keypoint = keypoints[0][i] + for j in range(len(sample_keypoint)): + kp_len = sample_height if j % 2 else sample_width + sample_coord = sample_bbox[1] if j % 2 else sample_bbox[0] + sample_keypoint[j] = ( + sample_keypoint[j] - sample_coord) / kp_len + sample_keypoint[j] = max(min(sample_keypoint[j], 1.0), 0.0) + new_keypoints.append(sample_keypoint) + new_kp_ignore.append(keypoints[1][i]) + + bboxes = np.array(new_bboxes) + labels = np.array(new_labels) + scores = np.array(new_scores) + if keypoints is not None: + keypoints = np.array(new_keypoints) + new_kp_ignore = np.array(new_kp_ignore) + return bboxes, labels, scores, (keypoints, new_kp_ignore) + return bboxes, labels, scores + + +def bbox_area_sampling(bboxes, labels, scores, target_size, min_size): + new_bboxes = [] + new_labels = [] + new_scores = [] + for i, bbox in enumerate(bboxes): + w = float((bbox[2] - bbox[0]) * target_size) + h = float((bbox[3] - bbox[1]) * target_size) + if w * h < float(min_size * min_size): + continue + else: + new_bboxes.append(bbox) + new_labels.append(labels[i]) + if scores is not None and scores.size != 0: + new_scores.append(scores[i]) + bboxes = np.array(new_bboxes) + labels = np.array(new_labels) + scores = np.array(new_scores) + return bboxes, labels, scores + + +def generate_sample_bbox(sampler): + scale = np.random.uniform(sampler[2], sampler[3]) + aspect_ratio = np.random.uniform(sampler[4], sampler[5]) + aspect_ratio = max(aspect_ratio, (scale**2.0)) + aspect_ratio = min(aspect_ratio, 1 / (scale**2.0)) + bbox_width = scale * (aspect_ratio**0.5) + bbox_height = scale / (aspect_ratio**0.5) + xmin_bound = 1 - bbox_width + ymin_bound = 1 - bbox_height + xmin = np.random.uniform(0, xmin_bound) + ymin = np.random.uniform(0, ymin_bound) + xmax = xmin + bbox_width + ymax = ymin + bbox_height + sampled_bbox = [xmin, ymin, xmax, ymax] + return sampled_bbox + + +def generate_sample_bbox_square(sampler, image_width, image_height): + scale = np.random.uniform(sampler[2], sampler[3]) + aspect_ratio = np.random.uniform(sampler[4], sampler[5]) + aspect_ratio = max(aspect_ratio, (scale**2.0)) + aspect_ratio = min(aspect_ratio, 1 / (scale**2.0)) + bbox_width = scale * (aspect_ratio**0.5) + bbox_height = scale / (aspect_ratio**0.5) + if image_height < image_width: + bbox_width = bbox_height * image_height / image_width + else: + bbox_height = bbox_width * image_width / image_height + xmin_bound = 1 - bbox_width + ymin_bound = 1 - bbox_height + xmin = np.random.uniform(0, xmin_bound) + ymin = np.random.uniform(0, ymin_bound) + xmax = xmin + bbox_width + ymax = ymin + bbox_height + sampled_bbox = [xmin, ymin, xmax, ymax] + return sampled_bbox + + +def data_anchor_sampling(bbox_labels, image_width, image_height, scale_array, + resize_width): + num_gt = len(bbox_labels) + # np.random.randint range: [low, high) + rand_idx = np.random.randint(0, num_gt) if num_gt != 0 else 0 + + if num_gt != 0: + norm_xmin = bbox_labels[rand_idx][0] + norm_ymin = bbox_labels[rand_idx][1] + norm_xmax = bbox_labels[rand_idx][2] + norm_ymax = bbox_labels[rand_idx][3] + + xmin = norm_xmin * image_width + ymin = norm_ymin * image_height + wid = image_width * (norm_xmax - norm_xmin) + hei = image_height * (norm_ymax - norm_ymin) + range_size = 0 + + area = wid * hei + for scale_ind in range(0, len(scale_array) - 1): + if area > scale_array[scale_ind] ** 2 and area < \ + scale_array[scale_ind + 1] ** 2: + range_size = scale_ind + 1 + break + + if area > scale_array[len(scale_array) - 2]**2: + range_size = len(scale_array) - 2 + + scale_choose = 0.0 + if range_size == 0: + rand_idx_size = 0 + else: + # np.random.randint range: [low, high) + rng_rand_size = np.random.randint(0, range_size + 1) + rand_idx_size = rng_rand_size % (range_size + 1) + + if rand_idx_size == range_size: + min_resize_val = scale_array[rand_idx_size] / 2.0 + max_resize_val = min(2.0 * scale_array[rand_idx_size], + 2 * math.sqrt(wid * hei)) + scale_choose = random.uniform(min_resize_val, max_resize_val) + else: + min_resize_val = scale_array[rand_idx_size] / 2.0 + max_resize_val = 2.0 * scale_array[rand_idx_size] + scale_choose = random.uniform(min_resize_val, max_resize_val) + + sample_bbox_size = wid * resize_width / scale_choose + + w_off_orig = 0.0 + h_off_orig = 0.0 + if sample_bbox_size < max(image_height, image_width): + if wid <= sample_bbox_size: + w_off_orig = np.random.uniform(xmin + wid - sample_bbox_size, + xmin) + else: + w_off_orig = np.random.uniform(xmin, + xmin + wid - sample_bbox_size) + + if hei <= sample_bbox_size: + h_off_orig = np.random.uniform(ymin + hei - sample_bbox_size, + ymin) + else: + h_off_orig = np.random.uniform(ymin, + ymin + hei - sample_bbox_size) + + else: + w_off_orig = np.random.uniform(image_width - sample_bbox_size, 0.0) + h_off_orig = np.random.uniform(image_height - sample_bbox_size, 0.0) + + w_off_orig = math.floor(w_off_orig) + h_off_orig = math.floor(h_off_orig) + + # Figure out top left coordinates. + w_off = float(w_off_orig / image_width) + h_off = float(h_off_orig / image_height) + + sampled_bbox = [ + w_off, h_off, w_off + float(sample_bbox_size / image_width), + h_off + float(sample_bbox_size / image_height) + ] + return sampled_bbox + else: + return 0 + + +def jaccard_overlap(sample_bbox, object_bbox): + if sample_bbox[0] >= object_bbox[2] or \ + sample_bbox[2] <= object_bbox[0] or \ + sample_bbox[1] >= object_bbox[3] or \ + sample_bbox[3] <= object_bbox[1]: + return 0 + intersect_xmin = max(sample_bbox[0], object_bbox[0]) + intersect_ymin = max(sample_bbox[1], object_bbox[1]) + intersect_xmax = min(sample_bbox[2], object_bbox[2]) + intersect_ymax = min(sample_bbox[3], object_bbox[3]) + intersect_size = (intersect_xmax - intersect_xmin) * ( + intersect_ymax - intersect_ymin) + sample_bbox_size = bbox_area(sample_bbox) + object_bbox_size = bbox_area(object_bbox) + overlap = intersect_size / ( + sample_bbox_size + object_bbox_size - intersect_size) + return overlap + + +def intersect_bbox(bbox1, bbox2): + if bbox2[0] > bbox1[2] or bbox2[2] < bbox1[0] or \ + bbox2[1] > bbox1[3] or bbox2[3] < bbox1[1]: + intersection_box = [0.0, 0.0, 0.0, 0.0] + else: + intersection_box = [ + max(bbox1[0], bbox2[0]), max(bbox1[1], bbox2[1]), + min(bbox1[2], bbox2[2]), min(bbox1[3], bbox2[3]) + ] + return intersection_box + + +def bbox_coverage(bbox1, bbox2): + inter_box = intersect_bbox(bbox1, bbox2) + intersect_size = bbox_area(inter_box) + + if intersect_size > 0: + bbox1_size = bbox_area(bbox1) + return intersect_size / bbox1_size + else: + return 0. + + +def satisfy_sample_constraint(sampler, + sample_bbox, + gt_bboxes, + satisfy_all=False): + if sampler[6] == 0 and sampler[7] == 0: + return True + satisfied = [] + for i in range(len(gt_bboxes)): + object_bbox = [ + gt_bboxes[i][0], gt_bboxes[i][1], gt_bboxes[i][2], gt_bboxes[i][3] + ] + overlap = jaccard_overlap(sample_bbox, object_bbox) + if sampler[6] != 0 and \ + overlap < sampler[6]: + satisfied.append(False) + continue + if sampler[7] != 0 and \ + overlap > sampler[7]: + satisfied.append(False) + continue + satisfied.append(True) + if not satisfy_all: + return True + + if satisfy_all: + return np.all(satisfied) + else: + return False + + +def satisfy_sample_constraint_coverage(sampler, sample_bbox, gt_bboxes): + if sampler[6] == 0 and sampler[7] == 0: + has_jaccard_overlap = False + else: + has_jaccard_overlap = True + if sampler[8] == 0 and sampler[9] == 0: + has_object_coverage = False + else: + has_object_coverage = True + + if not has_jaccard_overlap and not has_object_coverage: + return True + found = False + for i in range(len(gt_bboxes)): + object_bbox = [ + gt_bboxes[i][0], gt_bboxes[i][1], gt_bboxes[i][2], gt_bboxes[i][3] + ] + if has_jaccard_overlap: + overlap = jaccard_overlap(sample_bbox, object_bbox) + if sampler[6] != 0 and \ + overlap < sampler[6]: + continue + if sampler[7] != 0 and \ + overlap > sampler[7]: + continue + found = True + if has_object_coverage: + object_coverage = bbox_coverage(object_bbox, sample_bbox) + if sampler[8] != 0 and \ + object_coverage < sampler[8]: + continue + if sampler[9] != 0 and \ + object_coverage > sampler[9]: + continue + found = True + if found: + return True + return found + + +def crop_image_sampling(img, sample_bbox, image_width, image_height, + target_size): + # no clipping here + xmin = int(sample_bbox[0] * image_width) + xmax = int(sample_bbox[2] * image_width) + ymin = int(sample_bbox[1] * image_height) + ymax = int(sample_bbox[3] * image_height) + + w_off = xmin + h_off = ymin + width = xmax - xmin + height = ymax - ymin + cross_xmin = max(0.0, float(w_off)) + cross_ymin = max(0.0, float(h_off)) + cross_xmax = min(float(w_off + width - 1.0), float(image_width)) + cross_ymax = min(float(h_off + height - 1.0), float(image_height)) + cross_width = cross_xmax - cross_xmin + cross_height = cross_ymax - cross_ymin + + roi_xmin = 0 if w_off >= 0 else abs(w_off) + roi_ymin = 0 if h_off >= 0 else abs(h_off) + roi_width = cross_width + roi_height = cross_height + + roi_y1 = int(roi_ymin) + roi_y2 = int(roi_ymin + roi_height) + roi_x1 = int(roi_xmin) + roi_x2 = int(roi_xmin + roi_width) + + cross_y1 = int(cross_ymin) + cross_y2 = int(cross_ymin + cross_height) + cross_x1 = int(cross_xmin) + cross_x2 = int(cross_xmin + cross_width) + + sample_img = np.zeros((height, width, 3)) + sample_img[roi_y1: roi_y2, roi_x1: roi_x2] = \ + img[cross_y1: cross_y2, cross_x1: cross_x2] + + sample_img = cv2.resize( + sample_img, (target_size, target_size), interpolation=cv2.INTER_AREA) + + return sample_img + + +def is_poly(segm): + assert isinstance(segm, (list, dict)), \ + "Invalid segm type: {}".format(type(segm)) + return isinstance(segm, list) + + +def gaussian_radius(bbox_size, min_overlap): + height, width = bbox_size + + a1 = 1 + b1 = (height + width) + c1 = width * height * (1 - min_overlap) / (1 + min_overlap) + sq1 = np.sqrt(b1**2 - 4 * a1 * c1) + radius1 = (b1 + sq1) / (2 * a1) + + a2 = 4 + b2 = 2 * (height + width) + c2 = (1 - min_overlap) * width * height + sq2 = np.sqrt(b2**2 - 4 * a2 * c2) + radius2 = (b2 + sq2) / 2 + + a3 = 4 * min_overlap + b3 = -2 * min_overlap * (height + width) + c3 = (min_overlap - 1) * width * height + sq3 = np.sqrt(b3**2 - 4 * a3 * c3) + radius3 = (b3 + sq3) / 2 + return min(radius1, radius2, radius3) + + +def draw_gaussian(heatmap, center, radius, k=1, delte=6): + diameter = 2 * radius + 1 + sigma = diameter / delte + gaussian = gaussian2D((diameter, diameter), sigma_x=sigma, sigma_y=sigma) + + x, y = center + + height, width = heatmap.shape[0:2] + + left, right = min(x, radius), min(width - x, radius + 1) + top, bottom = min(y, radius), min(height - y, radius + 1) + + masked_heatmap = heatmap[y - top:y + bottom, x - left:x + right] + masked_gaussian = gaussian[radius - top:radius + bottom, radius - left: + radius + right] + np.maximum(masked_heatmap, masked_gaussian * k, out=masked_heatmap) + + +def gaussian2D(shape, sigma_x=1, sigma_y=1): + m, n = [(ss - 1.) / 2. for ss in shape] + y, x = np.ogrid[-m:m + 1, -n:n + 1] + + h = np.exp(-(x * x / (2 * sigma_x * sigma_x) + y * y / (2 * sigma_y * + sigma_y))) + h[h < np.finfo(h.dtype).eps * h.max()] = 0 + return h + + +def draw_umich_gaussian(heatmap, center, radius, k=1): + """ + draw_umich_gaussian, refer to https://github.com/xingyizhou/CenterNet/blob/master/src/lib/utils/image.py#L126 + """ + diameter = 2 * radius + 1 + gaussian = gaussian2D( + (diameter, diameter), sigma_x=diameter / 6, sigma_y=diameter / 6) + + x, y = int(center[0]), int(center[1]) + + height, width = heatmap.shape[0:2] + + left, right = min(x, radius), min(width - x, radius + 1) + top, bottom = min(y, radius), min(height - y, radius + 1) + + masked_heatmap = heatmap[y - top:y + bottom, x - left:x + right] + masked_gaussian = gaussian[radius - top:radius + bottom, radius - left: + radius + right] + if min(masked_gaussian.shape) > 0 and min(masked_heatmap.shape) > 0: + np.maximum(masked_heatmap, masked_gaussian * k, out=masked_heatmap) + return heatmap + + +def get_border(border, size): + i = 1 + while size - border // i <= border // i: + i *= 2 + return border // i diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/operators.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/operators.py new file mode 100644 index 000000000..5cc14a44d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/data/transform/operators.py @@ -0,0 +1,3015 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# function: +# operators to process sample, +# eg: decode/resize/crop image + +from __future__ import absolute_import +from __future__ import print_function +from __future__ import division + +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence + +from numbers import Number, Integral + +import uuid +import random +import math +import numpy as np +import os +import copy +import logging +import cv2 +from PIL import Image, ImageDraw +import pickle +import threading +MUTEX = threading.Lock() + +from ppdet.core.workspace import serializable +from ppdet.modeling import bbox_utils +from ..reader import Compose + +from .op_helper import (satisfy_sample_constraint, filter_and_process, + generate_sample_bbox, clip_bbox, data_anchor_sampling, + satisfy_sample_constraint_coverage, crop_image_sampling, + generate_sample_bbox_square, bbox_area_sampling, + is_poly, get_border) + +from ppdet.utils.logger import setup_logger +from ppdet.modeling.keypoint_utils import get_affine_transform, affine_transform +logger = setup_logger(__name__) + +registered_ops = [] + + +def register_op(cls): + registered_ops.append(cls.__name__) + if not hasattr(BaseOperator, cls.__name__): + setattr(BaseOperator, cls.__name__, cls) + else: + raise KeyError("The {} class has been registered.".format(cls.__name__)) + return serializable(cls) + + +class BboxError(ValueError): + pass + + +class ImageError(ValueError): + pass + + +class BaseOperator(object): + def __init__(self, name=None): + if name is None: + name = self.__class__.__name__ + self._id = name + '_' + str(uuid.uuid4())[-6:] + + def apply(self, sample, context=None): + """ Process a sample. + Args: + sample (dict): a dict of sample, eg: {'image':xx, 'label': xxx} + context (dict): info about this sample processing + Returns: + result (dict): a processed sample + """ + return sample + + def __call__(self, sample, context=None): + """ Process a sample. + Args: + sample (dict): a dict of sample, eg: {'image':xx, 'label': xxx} + context (dict): info about this sample processing + Returns: + result (dict): a processed sample + """ + if isinstance(sample, Sequence): + for i in range(len(sample)): + sample[i] = self.apply(sample[i], context) + else: + sample = self.apply(sample, context) + return sample + + def __str__(self): + return str(self._id) + + +@register_op +class Decode(BaseOperator): + def __init__(self): + """ Transform the image data to numpy format following the rgb format + """ + super(Decode, self).__init__() + + def apply(self, sample, context=None): + """ load image if 'im_file' field is not empty but 'image' is""" + if 'image' not in sample: + with open(sample['im_file'], 'rb') as f: + sample['image'] = f.read() + sample.pop('im_file') + + im = sample['image'] + data = np.frombuffer(im, dtype='uint8') + im = cv2.imdecode(data, 1) # BGR mode, but need RGB mode + if 'keep_ori_im' in sample and sample['keep_ori_im']: + sample['ori_image'] = im + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + + sample['image'] = im + if 'h' not in sample: + sample['h'] = im.shape[0] + elif sample['h'] != im.shape[0]: + logger.warning( + "The actual image height: {} is not equal to the " + "height: {} in annotation, and update sample['h'] by actual " + "image height.".format(im.shape[0], sample['h'])) + sample['h'] = im.shape[0] + if 'w' not in sample: + sample['w'] = im.shape[1] + elif sample['w'] != im.shape[1]: + logger.warning( + "The actual image width: {} is not equal to the " + "width: {} in annotation, and update sample['w'] by actual " + "image width.".format(im.shape[1], sample['w'])) + sample['w'] = im.shape[1] + + sample['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + sample['scale_factor'] = np.array([1., 1.], dtype=np.float32) + return sample + + +def _make_dirs(dirname): + try: + from pathlib import Path + except ImportError: + from pathlib2 import Path + Path(dirname).mkdir(exist_ok=True) + + +@register_op +class DecodeCache(BaseOperator): + def __init__(self, cache_root=None): + '''decode image and caching + ''' + super(DecodeCache, self).__init__() + + self.use_cache = False if cache_root is None else True + self.cache_root = cache_root + + if cache_root is not None: + _make_dirs(cache_root) + + def apply(self, sample, context=None): + + if self.use_cache and os.path.exists( + self.cache_path(self.cache_root, sample['im_file'])): + path = self.cache_path(self.cache_root, sample['im_file']) + im = self.load(path) + + else: + if 'image' not in sample: + with open(sample['im_file'], 'rb') as f: + sample['image'] = f.read() + + im = sample['image'] + data = np.frombuffer(im, dtype='uint8') + im = cv2.imdecode(data, 1) # BGR mode, but need RGB mode + if 'keep_ori_im' in sample and sample['keep_ori_im']: + sample['ori_image'] = im + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + + if self.use_cache and not os.path.exists( + self.cache_path(self.cache_root, sample['im_file'])): + path = self.cache_path(self.cache_root, sample['im_file']) + self.dump(im, path) + + sample['image'] = im + sample['h'] = im.shape[0] + sample['w'] = im.shape[1] + + sample['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + sample['scale_factor'] = np.array([1., 1.], dtype=np.float32) + + sample.pop('im_file') + + return sample + + @staticmethod + def cache_path(dir_oot, im_file): + return os.path.join(dir_oot, os.path.basename(im_file) + '.pkl') + + @staticmethod + def load(path): + with open(path, 'rb') as f: + im = pickle.load(f) + return im + + @staticmethod + def dump(obj, path): + MUTEX.acquire() + try: + with open(path, 'wb') as f: + pickle.dump(obj, f) + + except Exception as e: + logger.warning('dump {} occurs exception {}'.format(path, str(e))) + + finally: + MUTEX.release() + + +@register_op +class SniperDecodeCrop(BaseOperator): + def __init__(self): + super(SniperDecodeCrop, self).__init__() + + def __call__(self, sample, context=None): + if 'image' not in sample: + with open(sample['im_file'], 'rb') as f: + sample['image'] = f.read() + sample.pop('im_file') + + im = sample['image'] + data = np.frombuffer(im, dtype='uint8') + im = cv2.imdecode(data, cv2.IMREAD_COLOR) # BGR mode, but need RGB mode + if 'keep_ori_im' in sample and sample['keep_ori_im']: + sample['ori_image'] = im + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + + chip = sample['chip'] + x1, y1, x2, y2 = [int(xi) for xi in chip] + im = im[max(y1, 0):min(y2, im.shape[0]), max(x1, 0):min(x2, im.shape[ + 1]), :] + + sample['image'] = im + h = im.shape[0] + w = im.shape[1] + # sample['im_info'] = [h, w, 1.0] + sample['h'] = h + sample['w'] = w + + sample['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + sample['scale_factor'] = np.array([1., 1.], dtype=np.float32) + return sample + + +@register_op +class Permute(BaseOperator): + def __init__(self): + """ + Change the channel to be (C, H, W) + """ + super(Permute, self).__init__() + + def apply(self, sample, context=None): + im = sample['image'] + im = im.transpose((2, 0, 1)) + sample['image'] = im + return sample + + +@register_op +class Lighting(BaseOperator): + """ + Lighting the image by eigenvalues and eigenvectors + Args: + eigval (list): eigenvalues + eigvec (list): eigenvectors + alphastd (float): random weight of lighting, 0.1 by default + """ + + def __init__(self, eigval, eigvec, alphastd=0.1): + super(Lighting, self).__init__() + self.alphastd = alphastd + self.eigval = np.array(eigval).astype('float32') + self.eigvec = np.array(eigvec).astype('float32') + + def apply(self, sample, context=None): + alpha = np.random.normal(scale=self.alphastd, size=(3, )) + sample['image'] += np.dot(self.eigvec, self.eigval * alpha) + return sample + + +@register_op +class RandomErasingImage(BaseOperator): + def __init__(self, prob=0.5, lower=0.02, higher=0.4, aspect_ratio=0.3): + """ + Random Erasing Data Augmentation, see https://arxiv.org/abs/1708.04896 + Args: + prob (float): probability to carry out random erasing + lower (float): lower limit of the erasing area ratio + higher (float): upper limit of the erasing area ratio + aspect_ratio (float): aspect ratio of the erasing region + """ + super(RandomErasingImage, self).__init__() + self.prob = prob + self.lower = lower + self.higher = higher + self.aspect_ratio = aspect_ratio + + def apply(self, sample): + gt_bbox = sample['gt_bbox'] + im = sample['image'] + if not isinstance(im, np.ndarray): + raise TypeError("{}: image is not a numpy array.".format(self)) + if len(im.shape) != 3: + raise ImageError("{}: image is not 3-dimensional.".format(self)) + + for idx in range(gt_bbox.shape[0]): + if self.prob <= np.random.rand(): + continue + + x1, y1, x2, y2 = gt_bbox[idx, :] + w_bbox = x2 - x1 + h_bbox = y2 - y1 + area = w_bbox * h_bbox + + target_area = random.uniform(self.lower, self.higher) * area + aspect_ratio = random.uniform(self.aspect_ratio, + 1 / self.aspect_ratio) + + h = int(round(math.sqrt(target_area * aspect_ratio))) + w = int(round(math.sqrt(target_area / aspect_ratio))) + + if w < w_bbox and h < h_bbox: + off_y1 = random.randint(0, int(h_bbox - h)) + off_x1 = random.randint(0, int(w_bbox - w)) + im[int(y1 + off_y1):int(y1 + off_y1 + h), int(x1 + off_x1):int( + x1 + off_x1 + w), :] = 0 + sample['image'] = im + return sample + + +@register_op +class NormalizeImage(BaseOperator): + def __init__(self, mean=[0.485, 0.456, 0.406], std=[1, 1, 1], + is_scale=True): + """ + Args: + mean (list): the pixel mean + std (list): the pixel variance + """ + super(NormalizeImage, self).__init__() + self.mean = mean + self.std = std + self.is_scale = is_scale + if not (isinstance(self.mean, list) and isinstance(self.std, list) and + isinstance(self.is_scale, bool)): + raise TypeError("{}: input type is invalid.".format(self)) + from functools import reduce + if reduce(lambda x, y: x * y, self.std) == 0: + raise ValueError('{}: std is invalid!'.format(self)) + + def apply(self, sample, context=None): + """Normalize the image. + Operators: + 1.(optional) Scale the image to [0,1] + 2. Each pixel minus mean and is divided by std + """ + im = sample['image'] + im = im.astype(np.float32, copy=False) + mean = np.array(self.mean)[np.newaxis, np.newaxis, :] + std = np.array(self.std)[np.newaxis, np.newaxis, :] + + if self.is_scale: + im = im / 255.0 + + im -= mean + im /= std + + sample['image'] = im + return sample + + +@register_op +class GridMask(BaseOperator): + def __init__(self, + use_h=True, + use_w=True, + rotate=1, + offset=False, + ratio=0.5, + mode=1, + prob=0.7, + upper_iter=360000): + """ + GridMask Data Augmentation, see https://arxiv.org/abs/2001.04086 + Args: + use_h (bool): whether to mask vertically + use_w (boo;): whether to mask horizontally + rotate (float): angle for the mask to rotate + offset (float): mask offset + ratio (float): mask ratio + mode (int): gridmask mode + prob (float): max probability to carry out gridmask + upper_iter (int): suggested to be equal to global max_iter + """ + super(GridMask, self).__init__() + self.use_h = use_h + self.use_w = use_w + self.rotate = rotate + self.offset = offset + self.ratio = ratio + self.mode = mode + self.prob = prob + self.upper_iter = upper_iter + + from .gridmask_utils import Gridmask + self.gridmask_op = Gridmask( + use_h, + use_w, + rotate=rotate, + offset=offset, + ratio=ratio, + mode=mode, + prob=prob, + upper_iter=upper_iter) + + def apply(self, sample, context=None): + sample['image'] = self.gridmask_op(sample['image'], sample['curr_iter']) + return sample + + +@register_op +class RandomDistort(BaseOperator): + """Random color distortion. + Args: + hue (list): hue settings. in [lower, upper, probability] format. + saturation (list): saturation settings. in [lower, upper, probability] format. + contrast (list): contrast settings. in [lower, upper, probability] format. + brightness (list): brightness settings. in [lower, upper, probability] format. + random_apply (bool): whether to apply in random (yolo) or fixed (SSD) + order. + count (int): the number of doing distrot + random_channel (bool): whether to swap channels randomly + """ + + def __init__(self, + hue=[-18, 18, 0.5], + saturation=[0.5, 1.5, 0.5], + contrast=[0.5, 1.5, 0.5], + brightness=[0.5, 1.5, 0.5], + random_apply=True, + count=4, + random_channel=False): + super(RandomDistort, self).__init__() + self.hue = hue + self.saturation = saturation + self.contrast = contrast + self.brightness = brightness + self.random_apply = random_apply + self.count = count + self.random_channel = random_channel + + def apply_hue(self, img): + low, high, prob = self.hue + if np.random.uniform(0., 1.) < prob: + return img + + img = img.astype(np.float32) + # it works, but result differ from HSV version + delta = np.random.uniform(low, high) + u = np.cos(delta * np.pi) + w = np.sin(delta * np.pi) + bt = np.array([[1.0, 0.0, 0.0], [0.0, u, -w], [0.0, w, u]]) + tyiq = np.array([[0.299, 0.587, 0.114], [0.596, -0.274, -0.321], + [0.211, -0.523, 0.311]]) + ityiq = np.array([[1.0, 0.956, 0.621], [1.0, -0.272, -0.647], + [1.0, -1.107, 1.705]]) + t = np.dot(np.dot(ityiq, bt), tyiq).T + img = np.dot(img, t) + return img + + def apply_saturation(self, img): + low, high, prob = self.saturation + if np.random.uniform(0., 1.) < prob: + return img + delta = np.random.uniform(low, high) + img = img.astype(np.float32) + # it works, but result differ from HSV version + gray = img * np.array([[[0.299, 0.587, 0.114]]], dtype=np.float32) + gray = gray.sum(axis=2, keepdims=True) + gray *= (1.0 - delta) + img *= delta + img += gray + return img + + def apply_contrast(self, img): + low, high, prob = self.contrast + if np.random.uniform(0., 1.) < prob: + return img + delta = np.random.uniform(low, high) + img = img.astype(np.float32) + img *= delta + return img + + def apply_brightness(self, img): + low, high, prob = self.brightness + if np.random.uniform(0., 1.) < prob: + return img + delta = np.random.uniform(low, high) + img = img.astype(np.float32) + img += delta + return img + + def apply(self, sample, context=None): + img = sample['image'] + if self.random_apply: + functions = [ + self.apply_brightness, self.apply_contrast, + self.apply_saturation, self.apply_hue + ] + distortions = np.random.permutation(functions)[:self.count] + for func in distortions: + img = func(img) + sample['image'] = img + return sample + + img = self.apply_brightness(img) + mode = np.random.randint(0, 2) + + if mode: + img = self.apply_contrast(img) + + img = self.apply_saturation(img) + img = self.apply_hue(img) + + if not mode: + img = self.apply_contrast(img) + + if self.random_channel: + if np.random.randint(0, 2): + img = img[..., np.random.permutation(3)] + sample['image'] = img + return sample + + +@register_op +class AutoAugment(BaseOperator): + def __init__(self, autoaug_type="v1"): + """ + Args: + autoaug_type (str): autoaug type, support v0, v1, v2, v3, test + """ + super(AutoAugment, self).__init__() + self.autoaug_type = autoaug_type + + def apply(self, sample, context=None): + """ + Learning Data Augmentation Strategies for Object Detection, see https://arxiv.org/abs/1906.11172 + """ + im = sample['image'] + gt_bbox = sample['gt_bbox'] + if not isinstance(im, np.ndarray): + raise TypeError("{}: image is not a numpy array.".format(self)) + if len(im.shape) != 3: + raise ImageError("{}: image is not 3-dimensional.".format(self)) + if len(gt_bbox) == 0: + return sample + + height, width, _ = im.shape + norm_gt_bbox = np.ones_like(gt_bbox, dtype=np.float32) + norm_gt_bbox[:, 0] = gt_bbox[:, 1] / float(height) + norm_gt_bbox[:, 1] = gt_bbox[:, 0] / float(width) + norm_gt_bbox[:, 2] = gt_bbox[:, 3] / float(height) + norm_gt_bbox[:, 3] = gt_bbox[:, 2] / float(width) + + from .autoaugment_utils import distort_image_with_autoaugment + im, norm_gt_bbox = distort_image_with_autoaugment(im, norm_gt_bbox, + self.autoaug_type) + + gt_bbox[:, 0] = norm_gt_bbox[:, 1] * float(width) + gt_bbox[:, 1] = norm_gt_bbox[:, 0] * float(height) + gt_bbox[:, 2] = norm_gt_bbox[:, 3] * float(width) + gt_bbox[:, 3] = norm_gt_bbox[:, 2] * float(height) + + sample['image'] = im + sample['gt_bbox'] = gt_bbox + return sample + + +@register_op +class RandomFlip(BaseOperator): + def __init__(self, prob=0.5): + """ + Args: + prob (float): the probability of flipping image + """ + super(RandomFlip, self).__init__() + self.prob = prob + if not (isinstance(self.prob, float)): + raise TypeError("{}: input type is invalid.".format(self)) + + def apply_segm(self, segms, height, width): + def _flip_poly(poly, width): + flipped_poly = np.array(poly) + flipped_poly[0::2] = width - np.array(poly[0::2]) + return flipped_poly.tolist() + + def _flip_rle(rle, height, width): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, height, width) + mask = mask_util.decode(rle) + mask = mask[:, ::-1] + rle = mask_util.encode(np.array(mask, order='F', dtype=np.uint8)) + return rle + + flipped_segms = [] + for segm in segms: + if is_poly(segm): + # Polygon format + flipped_segms.append([_flip_poly(poly, width) for poly in segm]) + else: + # RLE format + import pycocotools.mask as mask_util + flipped_segms.append(_flip_rle(segm, height, width)) + return flipped_segms + + def apply_keypoint(self, gt_keypoint, width): + for i in range(gt_keypoint.shape[1]): + if i % 2 == 0: + old_x = gt_keypoint[:, i].copy() + gt_keypoint[:, i] = width - old_x + return gt_keypoint + + def apply_image(self, image): + return image[:, ::-1, :] + + def apply_bbox(self, bbox, width): + oldx1 = bbox[:, 0].copy() + oldx2 = bbox[:, 2].copy() + bbox[:, 0] = width - oldx2 + bbox[:, 2] = width - oldx1 + return bbox + + def apply_rbox(self, bbox, width): + oldx1 = bbox[:, 0].copy() + oldx2 = bbox[:, 2].copy() + oldx3 = bbox[:, 4].copy() + oldx4 = bbox[:, 6].copy() + bbox[:, 0] = width - oldx1 + bbox[:, 2] = width - oldx2 + bbox[:, 4] = width - oldx3 + bbox[:, 6] = width - oldx4 + bbox = [bbox_utils.get_best_begin_point_single(e) for e in bbox] + return bbox + + def apply(self, sample, context=None): + """Filp the image and bounding box. + Operators: + 1. Flip the image numpy. + 2. Transform the bboxes' x coordinates. + (Must judge whether the coordinates are normalized!) + 3. Transform the segmentations' x coordinates. + (Must judge whether the coordinates are normalized!) + Output: + sample: the image, bounding box and segmentation part + in sample are flipped. + """ + if np.random.uniform(0, 1) < self.prob: + im = sample['image'] + height, width = im.shape[:2] + im = self.apply_image(im) + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], width) + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], height, + width) + if 'gt_keypoint' in sample and len(sample['gt_keypoint']) > 0: + sample['gt_keypoint'] = self.apply_keypoint( + sample['gt_keypoint'], width) + + if 'semantic' in sample and sample['semantic']: + sample['semantic'] = sample['semantic'][:, ::-1] + + if 'gt_segm' in sample and sample['gt_segm'].any(): + sample['gt_segm'] = sample['gt_segm'][:, :, ::-1] + + if 'gt_rbox2poly' in sample and sample['gt_rbox2poly'].any(): + sample['gt_rbox2poly'] = self.apply_rbox(sample['gt_rbox2poly'], + width) + + sample['flipped'] = True + sample['image'] = im + return sample + + +@register_op +class Resize(BaseOperator): + def __init__(self, target_size, keep_ratio, interp=cv2.INTER_LINEAR): + """ + Resize image to target size. if keep_ratio is True, + resize the image's long side to the maximum of target_size + if keep_ratio is False, resize the image to target size(h, w) + Args: + target_size (int|list): image target size + keep_ratio (bool): whether keep_ratio or not, default true + interp (int): the interpolation method + """ + super(Resize, self).__init__() + self.keep_ratio = keep_ratio + self.interp = interp + if not isinstance(target_size, (Integral, Sequence)): + raise TypeError( + "Type of target_size is invalid. Must be Integer or List or Tuple, now is {}". + format(type(target_size))) + if isinstance(target_size, Integral): + target_size = [target_size, target_size] + self.target_size = target_size + + def apply_image(self, image, scale): + im_scale_x, im_scale_y = scale + + return cv2.resize( + image, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + + def apply_bbox(self, bbox, scale, size): + im_scale_x, im_scale_y = scale + resize_w, resize_h = size + bbox[:, 0::2] *= im_scale_x + bbox[:, 1::2] *= im_scale_y + bbox[:, 0::2] = np.clip(bbox[:, 0::2], 0, resize_w) + bbox[:, 1::2] = np.clip(bbox[:, 1::2], 0, resize_h) + return bbox + + def apply_segm(self, segms, im_size, scale): + def _resize_poly(poly, im_scale_x, im_scale_y): + resized_poly = np.array(poly).astype('float32') + resized_poly[0::2] *= im_scale_x + resized_poly[1::2] *= im_scale_y + return resized_poly.tolist() + + def _resize_rle(rle, im_h, im_w, im_scale_x, im_scale_y): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, im_h, im_w) + + mask = mask_util.decode(rle) + mask = cv2.resize( + mask, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + rle = mask_util.encode(np.array(mask, order='F', dtype=np.uint8)) + return rle + + im_h, im_w = im_size + im_scale_x, im_scale_y = scale + resized_segms = [] + for segm in segms: + if is_poly(segm): + # Polygon format + resized_segms.append([ + _resize_poly(poly, im_scale_x, im_scale_y) for poly in segm + ]) + else: + # RLE format + import pycocotools.mask as mask_util + resized_segms.append( + _resize_rle(segm, im_h, im_w, im_scale_x, im_scale_y)) + + return resized_segms + + def apply(self, sample, context=None): + """ Resize the image numpy. + """ + im = sample['image'] + if not isinstance(im, np.ndarray): + raise TypeError("{}: image type is not numpy.".format(self)) + if len(im.shape) != 3: + raise ImageError('{}: image is not 3-dimensional.'.format(self)) + + # apply image + im_shape = im.shape + if self.keep_ratio: + + im_size_min = np.min(im_shape[0:2]) + im_size_max = np.max(im_shape[0:2]) + + target_size_min = np.min(self.target_size) + target_size_max = np.max(self.target_size) + + im_scale = min(target_size_min / im_size_min, + target_size_max / im_size_max) + + resize_h = im_scale * float(im_shape[0]) + resize_w = im_scale * float(im_shape[1]) + + im_scale_x = im_scale + im_scale_y = im_scale + else: + resize_h, resize_w = self.target_size + im_scale_y = resize_h / im_shape[0] + im_scale_x = resize_w / im_shape[1] + + im = self.apply_image(sample['image'], [im_scale_x, im_scale_y]) + sample['image'] = im + sample['im_shape'] = np.asarray([resize_h, resize_w], dtype=np.float32) + if 'scale_factor' in sample: + scale_factor = sample['scale_factor'] + sample['scale_factor'] = np.asarray( + [scale_factor[0] * im_scale_y, scale_factor[1] * im_scale_x], + dtype=np.float32) + else: + sample['scale_factor'] = np.asarray( + [im_scale_y, im_scale_x], dtype=np.float32) + + # apply bbox + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], + [im_scale_x, im_scale_y], + [resize_w, resize_h]) + + # apply rbox + if 'gt_rbox2poly' in sample: + if np.array(sample['gt_rbox2poly']).shape[1] != 8: + logger.warning( + "gt_rbox2poly's length shoule be 8, but actually is {}". + format(len(sample['gt_rbox2poly']))) + sample['gt_rbox2poly'] = self.apply_bbox(sample['gt_rbox2poly'], + [im_scale_x, im_scale_y], + [resize_w, resize_h]) + + # apply polygon + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], im_shape[:2], + [im_scale_x, im_scale_y]) + + # apply semantic + if 'semantic' in sample and sample['semantic']: + semantic = sample['semantic'] + semantic = cv2.resize( + semantic.astype('float32'), + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + semantic = np.asarray(semantic).astype('int32') + semantic = np.expand_dims(semantic, 0) + sample['semantic'] = semantic + + # apply gt_segm + if 'gt_segm' in sample and len(sample['gt_segm']) > 0: + masks = [ + cv2.resize( + gt_segm, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=cv2.INTER_NEAREST) + for gt_segm in sample['gt_segm'] + ] + sample['gt_segm'] = np.asarray(masks).astype(np.uint8) + + return sample + + +@register_op +class MultiscaleTestResize(BaseOperator): + def __init__(self, + origin_target_size=[800, 1333], + target_size=[], + interp=cv2.INTER_LINEAR, + use_flip=True): + """ + Rescale image to the each size in target size, and capped at max_size. + Args: + origin_target_size (list): origin target size of image + target_size (list): A list of target sizes of image. + interp (int): the interpolation method. + use_flip (bool): whether use flip augmentation. + """ + super(MultiscaleTestResize, self).__init__() + self.interp = interp + self.use_flip = use_flip + + if not isinstance(target_size, Sequence): + raise TypeError( + "Type of target_size is invalid. Must be List or Tuple, now is {}". + format(type(target_size))) + self.target_size = target_size + + if not isinstance(origin_target_size, Sequence): + raise TypeError( + "Type of origin_target_size is invalid. Must be List or Tuple, now is {}". + format(type(origin_target_size))) + + self.origin_target_size = origin_target_size + + def apply(self, sample, context=None): + """ Resize the image numpy for multi-scale test. + """ + samples = [] + resizer = Resize( + self.origin_target_size, keep_ratio=True, interp=self.interp) + samples.append(resizer(sample.copy(), context)) + if self.use_flip: + flipper = RandomFlip(1.1) + samples.append(flipper(sample.copy(), context=context)) + + for size in self.target_size: + resizer = Resize(size, keep_ratio=True, interp=self.interp) + samples.append(resizer(sample.copy(), context)) + + return samples + + +@register_op +class RandomResize(BaseOperator): + def __init__(self, + target_size, + keep_ratio=True, + interp=cv2.INTER_LINEAR, + random_size=True, + random_interp=False): + """ + Resize image to target size randomly. random target_size and interpolation method + Args: + target_size (int, list, tuple): image target size, if random size is True, must be list or tuple + keep_ratio (bool): whether keep_raio or not, default true + interp (int): the interpolation method + random_size (bool): whether random select target size of image + random_interp (bool): whether random select interpolation method + """ + super(RandomResize, self).__init__() + self.keep_ratio = keep_ratio + self.interp = interp + self.interps = [ + cv2.INTER_NEAREST, + cv2.INTER_LINEAR, + cv2.INTER_AREA, + cv2.INTER_CUBIC, + cv2.INTER_LANCZOS4, + ] + assert isinstance(target_size, ( + Integral, Sequence)), "target_size must be Integer, List or Tuple" + if random_size and not isinstance(target_size, Sequence): + raise TypeError( + "Type of target_size is invalid when random_size is True. Must be List or Tuple, now is {}". + format(type(target_size))) + self.target_size = target_size + self.random_size = random_size + self.random_interp = random_interp + + def apply(self, sample, context=None): + """ Resize the image numpy. + """ + if self.random_size: + target_size = random.choice(self.target_size) + else: + target_size = self.target_size + + if self.random_interp: + interp = random.choice(self.interps) + else: + interp = self.interp + + resizer = Resize(target_size, self.keep_ratio, interp) + return resizer(sample, context=context) + + +@register_op +class RandomExpand(BaseOperator): + """Random expand the canvas. + Args: + ratio (float): maximum expansion ratio. + prob (float): probability to expand. + fill_value (list): color value used to fill the canvas. in RGB order. + """ + + def __init__(self, ratio=4., prob=0.5, fill_value=(127.5, 127.5, 127.5)): + super(RandomExpand, self).__init__() + assert ratio > 1.01, "expand ratio must be larger than 1.01" + self.ratio = ratio + self.prob = prob + assert isinstance(fill_value, (Number, Sequence)), \ + "fill value must be either float or sequence" + if isinstance(fill_value, Number): + fill_value = (fill_value, ) * 3 + if not isinstance(fill_value, tuple): + fill_value = tuple(fill_value) + self.fill_value = fill_value + + def apply(self, sample, context=None): + if np.random.uniform(0., 1.) < self.prob: + return sample + + im = sample['image'] + height, width = im.shape[:2] + ratio = np.random.uniform(1., self.ratio) + h = int(height * ratio) + w = int(width * ratio) + if not h > height or not w > width: + return sample + y = np.random.randint(0, h - height) + x = np.random.randint(0, w - width) + offsets, size = [x, y], [h, w] + + pad = Pad(size, + pad_mode=-1, + offsets=offsets, + fill_value=self.fill_value) + + return pad(sample, context=context) + + +@register_op +class CropWithSampling(BaseOperator): + def __init__(self, batch_sampler, satisfy_all=False, avoid_no_bbox=True): + """ + Args: + batch_sampler (list): Multiple sets of different + parameters for cropping. + satisfy_all (bool): whether all boxes must satisfy. + e.g.[[1, 1, 1.0, 1.0, 1.0, 1.0, 0.0, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.1, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.3, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.5, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.7, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.9, 1.0], + [1, 50, 0.3, 1.0, 0.5, 2.0, 0.0, 1.0]] + [max sample, max trial, min scale, max scale, + min aspect ratio, max aspect ratio, + min overlap, max overlap] + avoid_no_bbox (bool): whether to to avoid the + situation where the box does not appear. + """ + super(CropWithSampling, self).__init__() + self.batch_sampler = batch_sampler + self.satisfy_all = satisfy_all + self.avoid_no_bbox = avoid_no_bbox + + def apply(self, sample, context): + """ + Crop the image and modify bounding box. + Operators: + 1. Scale the image width and height. + 2. Crop the image according to a radom sample. + 3. Rescale the bounding box. + 4. Determine if the new bbox is satisfied in the new image. + Returns: + sample: the image, bounding box are replaced. + """ + assert 'image' in sample, "image data not found" + im = sample['image'] + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + im_height, im_width = im.shape[:2] + gt_score = None + if 'gt_score' in sample: + gt_score = sample['gt_score'] + sampled_bbox = [] + gt_bbox = gt_bbox.tolist() + for sampler in self.batch_sampler: + found = 0 + for i in range(sampler[1]): + if found >= sampler[0]: + break + sample_bbox = generate_sample_bbox(sampler) + if satisfy_sample_constraint(sampler, sample_bbox, gt_bbox, + self.satisfy_all): + sampled_bbox.append(sample_bbox) + found = found + 1 + im = np.array(im) + while sampled_bbox: + idx = int(np.random.uniform(0, len(sampled_bbox))) + sample_bbox = sampled_bbox.pop(idx) + sample_bbox = clip_bbox(sample_bbox) + crop_bbox, crop_class, crop_score = \ + filter_and_process(sample_bbox, gt_bbox, gt_class, scores=gt_score) + if self.avoid_no_bbox: + if len(crop_bbox) < 1: + continue + xmin = int(sample_bbox[0] * im_width) + xmax = int(sample_bbox[2] * im_width) + ymin = int(sample_bbox[1] * im_height) + ymax = int(sample_bbox[3] * im_height) + im = im[ymin:ymax, xmin:xmax] + sample['image'] = im + sample['gt_bbox'] = crop_bbox + sample['gt_class'] = crop_class + sample['gt_score'] = crop_score + return sample + return sample + + +@register_op +class CropWithDataAchorSampling(BaseOperator): + def __init__(self, + batch_sampler, + anchor_sampler=None, + target_size=None, + das_anchor_scales=[16, 32, 64, 128], + sampling_prob=0.5, + min_size=8., + avoid_no_bbox=True): + """ + Args: + anchor_sampler (list): anchor_sampling sets of different + parameters for cropping. + batch_sampler (list): Multiple sets of different + parameters for cropping. + e.g.[[1, 10, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 0.2, 0.0]] + [[1, 50, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0], + [1, 50, 0.3, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0], + [1, 50, 0.3, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0], + [1, 50, 0.3, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0], + [1, 50, 0.3, 1.0, 1.0, 1.0, 0.0, 0.0, 1.0, 0.0]] + [max sample, max trial, min scale, max scale, + min aspect ratio, max aspect ratio, + min overlap, max overlap, min coverage, max coverage] + target_size (int): target image size. + das_anchor_scales (list[float]): a list of anchor scales in data + anchor smapling. + min_size (float): minimum size of sampled bbox. + avoid_no_bbox (bool): whether to to avoid the + situation where the box does not appear. + """ + super(CropWithDataAchorSampling, self).__init__() + self.anchor_sampler = anchor_sampler + self.batch_sampler = batch_sampler + self.target_size = target_size + self.sampling_prob = sampling_prob + self.min_size = min_size + self.avoid_no_bbox = avoid_no_bbox + self.das_anchor_scales = np.array(das_anchor_scales) + + def apply(self, sample, context): + """ + Crop the image and modify bounding box. + Operators: + 1. Scale the image width and height. + 2. Crop the image according to a radom sample. + 3. Rescale the bounding box. + 4. Determine if the new bbox is satisfied in the new image. + Returns: + sample: the image, bounding box are replaced. + """ + assert 'image' in sample, "image data not found" + im = sample['image'] + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + image_height, image_width = im.shape[:2] + gt_bbox[:, 0] /= image_width + gt_bbox[:, 1] /= image_height + gt_bbox[:, 2] /= image_width + gt_bbox[:, 3] /= image_height + gt_score = None + if 'gt_score' in sample: + gt_score = sample['gt_score'] + sampled_bbox = [] + gt_bbox = gt_bbox.tolist() + + prob = np.random.uniform(0., 1.) + if prob > self.sampling_prob: # anchor sampling + assert self.anchor_sampler + for sampler in self.anchor_sampler: + found = 0 + for i in range(sampler[1]): + if found >= sampler[0]: + break + sample_bbox = data_anchor_sampling( + gt_bbox, image_width, image_height, + self.das_anchor_scales, self.target_size) + if sample_bbox == 0: + break + if satisfy_sample_constraint_coverage(sampler, sample_bbox, + gt_bbox): + sampled_bbox.append(sample_bbox) + found = found + 1 + im = np.array(im) + while sampled_bbox: + idx = int(np.random.uniform(0, len(sampled_bbox))) + sample_bbox = sampled_bbox.pop(idx) + + if 'gt_keypoint' in sample.keys(): + keypoints = (sample['gt_keypoint'], + sample['keypoint_ignore']) + crop_bbox, crop_class, crop_score, gt_keypoints = \ + filter_and_process(sample_bbox, gt_bbox, gt_class, + scores=gt_score, + keypoints=keypoints) + else: + crop_bbox, crop_class, crop_score = filter_and_process( + sample_bbox, gt_bbox, gt_class, scores=gt_score) + crop_bbox, crop_class, crop_score = bbox_area_sampling( + crop_bbox, crop_class, crop_score, self.target_size, + self.min_size) + + if self.avoid_no_bbox: + if len(crop_bbox) < 1: + continue + im = crop_image_sampling(im, sample_bbox, image_width, + image_height, self.target_size) + height, width = im.shape[:2] + crop_bbox[:, 0] *= width + crop_bbox[:, 1] *= height + crop_bbox[:, 2] *= width + crop_bbox[:, 3] *= height + sample['image'] = im + sample['gt_bbox'] = crop_bbox + sample['gt_class'] = crop_class + if 'gt_score' in sample: + sample['gt_score'] = crop_score + if 'gt_keypoint' in sample.keys(): + sample['gt_keypoint'] = gt_keypoints[0] + sample['keypoint_ignore'] = gt_keypoints[1] + return sample + return sample + + else: + for sampler in self.batch_sampler: + found = 0 + for i in range(sampler[1]): + if found >= sampler[0]: + break + sample_bbox = generate_sample_bbox_square( + sampler, image_width, image_height) + if satisfy_sample_constraint_coverage(sampler, sample_bbox, + gt_bbox): + sampled_bbox.append(sample_bbox) + found = found + 1 + im = np.array(im) + while sampled_bbox: + idx = int(np.random.uniform(0, len(sampled_bbox))) + sample_bbox = sampled_bbox.pop(idx) + sample_bbox = clip_bbox(sample_bbox) + + if 'gt_keypoint' in sample.keys(): + keypoints = (sample['gt_keypoint'], + sample['keypoint_ignore']) + crop_bbox, crop_class, crop_score, gt_keypoints = \ + filter_and_process(sample_bbox, gt_bbox, gt_class, + scores=gt_score, + keypoints=keypoints) + else: + crop_bbox, crop_class, crop_score = filter_and_process( + sample_bbox, gt_bbox, gt_class, scores=gt_score) + # sampling bbox according the bbox area + crop_bbox, crop_class, crop_score = bbox_area_sampling( + crop_bbox, crop_class, crop_score, self.target_size, + self.min_size) + + if self.avoid_no_bbox: + if len(crop_bbox) < 1: + continue + xmin = int(sample_bbox[0] * image_width) + xmax = int(sample_bbox[2] * image_width) + ymin = int(sample_bbox[1] * image_height) + ymax = int(sample_bbox[3] * image_height) + im = im[ymin:ymax, xmin:xmax] + height, width = im.shape[:2] + crop_bbox[:, 0] *= width + crop_bbox[:, 1] *= height + crop_bbox[:, 2] *= width + crop_bbox[:, 3] *= height + sample['image'] = im + sample['gt_bbox'] = crop_bbox + sample['gt_class'] = crop_class + if 'gt_score' in sample: + sample['gt_score'] = crop_score + if 'gt_keypoint' in sample.keys(): + sample['gt_keypoint'] = gt_keypoints[0] + sample['keypoint_ignore'] = gt_keypoints[1] + return sample + return sample + + +@register_op +class RandomCrop(BaseOperator): + """Random crop image and bboxes. + Args: + aspect_ratio (list): aspect ratio of cropped region. + in [min, max] format. + thresholds (list): iou thresholds for decide a valid bbox crop. + scaling (list): ratio between a cropped region and the original image. + in [min, max] format. + num_attempts (int): number of tries before giving up. + allow_no_crop (bool): allow return without actually cropping them. + cover_all_box (bool): ensure all bboxes are covered in the final crop. + is_mask_crop(bool): whether crop the segmentation. + """ + + def __init__(self, + aspect_ratio=[.5, 2.], + thresholds=[.0, .1, .3, .5, .7, .9], + scaling=[.3, 1.], + num_attempts=50, + allow_no_crop=True, + cover_all_box=False, + is_mask_crop=False): + super(RandomCrop, self).__init__() + self.aspect_ratio = aspect_ratio + self.thresholds = thresholds + self.scaling = scaling + self.num_attempts = num_attempts + self.allow_no_crop = allow_no_crop + self.cover_all_box = cover_all_box + self.is_mask_crop = is_mask_crop + + def crop_segms(self, segms, valid_ids, crop, height, width): + def _crop_poly(segm, crop): + xmin, ymin, xmax, ymax = crop + crop_coord = [xmin, ymin, xmin, ymax, xmax, ymax, xmax, ymin] + crop_p = np.array(crop_coord).reshape(4, 2) + crop_p = Polygon(crop_p) + + crop_segm = list() + for poly in segm: + poly = np.array(poly).reshape(len(poly) // 2, 2) + polygon = Polygon(poly) + if not polygon.is_valid: + exterior = polygon.exterior + multi_lines = exterior.intersection(exterior) + polygons = shapely.ops.polygonize(multi_lines) + polygon = MultiPolygon(polygons) + multi_polygon = list() + if isinstance(polygon, MultiPolygon): + multi_polygon = copy.deepcopy(polygon) + else: + multi_polygon.append(copy.deepcopy(polygon)) + for per_polygon in multi_polygon: + inter = per_polygon.intersection(crop_p) + if not inter: + continue + if isinstance(inter, (MultiPolygon, GeometryCollection)): + for part in inter: + if not isinstance(part, Polygon): + continue + part = np.squeeze( + np.array(part.exterior.coords[:-1]).reshape(1, + -1)) + part[0::2] -= xmin + part[1::2] -= ymin + crop_segm.append(part.tolist()) + elif isinstance(inter, Polygon): + crop_poly = np.squeeze( + np.array(inter.exterior.coords[:-1]).reshape(1, -1)) + crop_poly[0::2] -= xmin + crop_poly[1::2] -= ymin + crop_segm.append(crop_poly.tolist()) + else: + continue + return crop_segm + + def _crop_rle(rle, crop, height, width): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, height, width) + mask = mask_util.decode(rle) + mask = mask[crop[1]:crop[3], crop[0]:crop[2]] + rle = mask_util.encode(np.array(mask, order='F', dtype=np.uint8)) + return rle + + crop_segms = [] + for id in valid_ids: + segm = segms[id] + if is_poly(segm): + import copy + import shapely.ops + from shapely.geometry import Polygon, MultiPolygon, GeometryCollection + logging.getLogger("shapely").setLevel(logging.WARNING) + # Polygon format + crop_segms.append(_crop_poly(segm, crop)) + else: + # RLE format + import pycocotools.mask as mask_util + crop_segms.append(_crop_rle(segm, crop, height, width)) + return crop_segms + + def apply(self, sample, context=None): + if 'gt_bbox' in sample and len(sample['gt_bbox']) == 0: + return sample + + h, w = sample['image'].shape[:2] + gt_bbox = sample['gt_bbox'] + + # NOTE Original method attempts to generate one candidate for each + # threshold then randomly sample one from the resulting list. + # Here a short circuit approach is taken, i.e., randomly choose a + # threshold and attempt to find a valid crop, and simply return the + # first one found. + # The probability is not exactly the same, kinda resembling the + # "Monty Hall" problem. Actually carrying out the attempts will affect + # observability (just like opening doors in the "Monty Hall" game). + thresholds = list(self.thresholds) + if self.allow_no_crop: + thresholds.append('no_crop') + np.random.shuffle(thresholds) + + for thresh in thresholds: + if thresh == 'no_crop': + return sample + + found = False + for i in range(self.num_attempts): + scale = np.random.uniform(*self.scaling) + if self.aspect_ratio is not None: + min_ar, max_ar = self.aspect_ratio + aspect_ratio = np.random.uniform( + max(min_ar, scale**2), min(max_ar, scale**-2)) + h_scale = scale / np.sqrt(aspect_ratio) + w_scale = scale * np.sqrt(aspect_ratio) + else: + h_scale = np.random.uniform(*self.scaling) + w_scale = np.random.uniform(*self.scaling) + crop_h = h * h_scale + crop_w = w * w_scale + if self.aspect_ratio is None: + if crop_h / crop_w < 0.5 or crop_h / crop_w > 2.0: + continue + + crop_h = int(crop_h) + crop_w = int(crop_w) + crop_y = np.random.randint(0, h - crop_h) + crop_x = np.random.randint(0, w - crop_w) + crop_box = [crop_x, crop_y, crop_x + crop_w, crop_y + crop_h] + iou = self._iou_matrix( + gt_bbox, np.array( + [crop_box], dtype=np.float32)) + if iou.max() < thresh: + continue + + if self.cover_all_box and iou.min() < thresh: + continue + + cropped_box, valid_ids = self._crop_box_with_center_constraint( + gt_bbox, np.array( + crop_box, dtype=np.float32)) + if valid_ids.size > 0: + found = True + break + + if found: + if self.is_mask_crop and 'gt_poly' in sample and len(sample[ + 'gt_poly']) > 0: + crop_polys = self.crop_segms( + sample['gt_poly'], + valid_ids, + np.array( + crop_box, dtype=np.int64), + h, + w) + if [] in crop_polys: + delete_id = list() + valid_polys = list() + for id, crop_poly in enumerate(crop_polys): + if crop_poly == []: + delete_id.append(id) + else: + valid_polys.append(crop_poly) + valid_ids = np.delete(valid_ids, delete_id) + if len(valid_polys) == 0: + return sample + sample['gt_poly'] = valid_polys + else: + sample['gt_poly'] = crop_polys + + if 'gt_segm' in sample: + sample['gt_segm'] = self._crop_segm(sample['gt_segm'], + crop_box) + sample['gt_segm'] = np.take( + sample['gt_segm'], valid_ids, axis=0) + + sample['image'] = self._crop_image(sample['image'], crop_box) + sample['gt_bbox'] = np.take(cropped_box, valid_ids, axis=0) + sample['gt_class'] = np.take( + sample['gt_class'], valid_ids, axis=0) + if 'gt_score' in sample: + sample['gt_score'] = np.take( + sample['gt_score'], valid_ids, axis=0) + + if 'is_crowd' in sample: + sample['is_crowd'] = np.take( + sample['is_crowd'], valid_ids, axis=0) + return sample + + return sample + + def _iou_matrix(self, a, b): + tl_i = np.maximum(a[:, np.newaxis, :2], b[:, :2]) + br_i = np.minimum(a[:, np.newaxis, 2:], b[:, 2:]) + + area_i = np.prod(br_i - tl_i, axis=2) * (tl_i < br_i).all(axis=2) + area_a = np.prod(a[:, 2:] - a[:, :2], axis=1) + area_b = np.prod(b[:, 2:] - b[:, :2], axis=1) + area_o = (area_a[:, np.newaxis] + area_b - area_i) + return area_i / (area_o + 1e-10) + + def _crop_box_with_center_constraint(self, box, crop): + cropped_box = box.copy() + + cropped_box[:, :2] = np.maximum(box[:, :2], crop[:2]) + cropped_box[:, 2:] = np.minimum(box[:, 2:], crop[2:]) + cropped_box[:, :2] -= crop[:2] + cropped_box[:, 2:] -= crop[:2] + + centers = (box[:, :2] + box[:, 2:]) / 2 + valid = np.logical_and(crop[:2] <= centers, + centers < crop[2:]).all(axis=1) + valid = np.logical_and( + valid, (cropped_box[:, :2] < cropped_box[:, 2:]).all(axis=1)) + + return cropped_box, np.where(valid)[0] + + def _crop_image(self, img, crop): + x1, y1, x2, y2 = crop + return img[y1:y2, x1:x2, :] + + def _crop_segm(self, segm, crop): + x1, y1, x2, y2 = crop + return segm[:, y1:y2, x1:x2] + + +@register_op +class RandomScaledCrop(BaseOperator): + """Resize image and bbox based on long side (with optional random scaling), + then crop or pad image to target size. + Args: + target_dim (int): target size. + scale_range (list): random scale range. + interp (int): interpolation method, default to `cv2.INTER_LINEAR`. + """ + + def __init__(self, + target_dim=512, + scale_range=[.1, 2.], + interp=cv2.INTER_LINEAR): + super(RandomScaledCrop, self).__init__() + self.target_dim = target_dim + self.scale_range = scale_range + self.interp = interp + + def apply(self, sample, context=None): + img = sample['image'] + h, w = img.shape[:2] + random_scale = np.random.uniform(*self.scale_range) + dim = self.target_dim + random_dim = int(dim * random_scale) + dim_max = max(h, w) + scale = random_dim / dim_max + resize_w = w * scale + resize_h = h * scale + offset_x = int(max(0, np.random.uniform(0., resize_w - dim))) + offset_y = int(max(0, np.random.uniform(0., resize_h - dim))) + + img = cv2.resize(img, (resize_w, resize_h), interpolation=self.interp) + img = np.array(img) + canvas = np.zeros((dim, dim, 3), dtype=img.dtype) + canvas[:min(dim, resize_h), :min(dim, resize_w), :] = img[ + offset_y:offset_y + dim, offset_x:offset_x + dim, :] + sample['image'] = canvas + sample['im_shape'] = np.asarray([resize_h, resize_w], dtype=np.float32) + scale_factor = sample['sacle_factor'] + sample['scale_factor'] = np.asarray( + [scale_factor[0] * scale, scale_factor[1] * scale], + dtype=np.float32) + + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + scale_array = np.array([scale, scale] * 2, dtype=np.float32) + shift_array = np.array([offset_x, offset_y] * 2, dtype=np.float32) + boxes = sample['gt_bbox'] * scale_array - shift_array + boxes = np.clip(boxes, 0, dim - 1) + # filter boxes with no area + area = np.prod(boxes[..., 2:] - boxes[..., :2], axis=1) + valid = (area > 1.).nonzero()[0] + sample['gt_bbox'] = boxes[valid] + sample['gt_class'] = sample['gt_class'][valid] + + return sample + + +@register_op +class Cutmix(BaseOperator): + def __init__(self, alpha=1.5, beta=1.5): + """ + CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features, see https://arxiv.org/abs/1905.04899 + Cutmix image and gt_bbbox/gt_score + Args: + alpha (float): alpha parameter of beta distribute + beta (float): beta parameter of beta distribute + """ + super(Cutmix, self).__init__() + self.alpha = alpha + self.beta = beta + if self.alpha <= 0.0: + raise ValueError("alpha shold be positive in {}".format(self)) + if self.beta <= 0.0: + raise ValueError("beta shold be positive in {}".format(self)) + + def apply_image(self, img1, img2, factor): + """ _rand_bbox """ + h = max(img1.shape[0], img2.shape[0]) + w = max(img1.shape[1], img2.shape[1]) + cut_rat = np.sqrt(1. - factor) + + cut_w = np.int32(w * cut_rat) + cut_h = np.int32(h * cut_rat) + + # uniform + cx = np.random.randint(w) + cy = np.random.randint(h) + + bbx1 = np.clip(cx - cut_w // 2, 0, w - 1) + bby1 = np.clip(cy - cut_h // 2, 0, h - 1) + bbx2 = np.clip(cx + cut_w // 2, 0, w - 1) + bby2 = np.clip(cy + cut_h // 2, 0, h - 1) + + img_1_pad = np.zeros((h, w, img1.shape[2]), 'float32') + img_1_pad[:img1.shape[0], :img1.shape[1], :] = \ + img1.astype('float32') + img_2_pad = np.zeros((h, w, img2.shape[2]), 'float32') + img_2_pad[:img2.shape[0], :img2.shape[1], :] = \ + img2.astype('float32') + img_1_pad[bby1:bby2, bbx1:bbx2, :] = img_2_pad[bby1:bby2, bbx1:bbx2, :] + return img_1_pad + + def __call__(self, sample, context=None): + if not isinstance(sample, Sequence): + return sample + + assert len(sample) == 2, 'cutmix need two samples' + + factor = np.random.beta(self.alpha, self.beta) + factor = max(0.0, min(1.0, factor)) + if factor >= 1.0: + return sample[0] + if factor <= 0.0: + return sample[1] + img1 = sample[0]['image'] + img2 = sample[1]['image'] + img = self.apply_image(img1, img2, factor) + gt_bbox1 = sample[0]['gt_bbox'] + gt_bbox2 = sample[1]['gt_bbox'] + gt_bbox = np.concatenate((gt_bbox1, gt_bbox2), axis=0) + gt_class1 = sample[0]['gt_class'] + gt_class2 = sample[1]['gt_class'] + gt_class = np.concatenate((gt_class1, gt_class2), axis=0) + gt_score1 = np.ones_like(sample[0]['gt_class']) + gt_score2 = np.ones_like(sample[1]['gt_class']) + gt_score = np.concatenate( + (gt_score1 * factor, gt_score2 * (1. - factor)), axis=0) + result = copy.deepcopy(sample[0]) + result['image'] = img + result['gt_bbox'] = gt_bbox + result['gt_score'] = gt_score + result['gt_class'] = gt_class + if 'is_crowd' in sample[0]: + is_crowd1 = sample[0]['is_crowd'] + is_crowd2 = sample[1]['is_crowd'] + is_crowd = np.concatenate((is_crowd1, is_crowd2), axis=0) + result['is_crowd'] = is_crowd + if 'difficult' in sample[0]: + is_difficult1 = sample[0]['difficult'] + is_difficult2 = sample[1]['difficult'] + is_difficult = np.concatenate( + (is_difficult1, is_difficult2), axis=0) + result['difficult'] = is_difficult + return result + + +@register_op +class Mixup(BaseOperator): + def __init__(self, alpha=1.5, beta=1.5): + """ Mixup image and gt_bbbox/gt_score + Args: + alpha (float): alpha parameter of beta distribute + beta (float): beta parameter of beta distribute + """ + super(Mixup, self).__init__() + self.alpha = alpha + self.beta = beta + if self.alpha <= 0.0: + raise ValueError("alpha shold be positive in {}".format(self)) + if self.beta <= 0.0: + raise ValueError("beta shold be positive in {}".format(self)) + + def apply_image(self, img1, img2, factor): + h = max(img1.shape[0], img2.shape[0]) + w = max(img1.shape[1], img2.shape[1]) + img = np.zeros((h, w, img1.shape[2]), 'float32') + img[:img1.shape[0], :img1.shape[1], :] = \ + img1.astype('float32') * factor + img[:img2.shape[0], :img2.shape[1], :] += \ + img2.astype('float32') * (1.0 - factor) + return img.astype('uint8') + + def __call__(self, sample, context=None): + if not isinstance(sample, Sequence): + return sample + + assert len(sample) == 2, 'mixup need two samples' + + factor = np.random.beta(self.alpha, self.beta) + factor = max(0.0, min(1.0, factor)) + if factor >= 1.0: + return sample[0] + if factor <= 0.0: + return sample[1] + im = self.apply_image(sample[0]['image'], sample[1]['image'], factor) + result = copy.deepcopy(sample[0]) + result['image'] = im + # apply bbox and score + if 'gt_bbox' in sample[0]: + gt_bbox1 = sample[0]['gt_bbox'] + gt_bbox2 = sample[1]['gt_bbox'] + gt_bbox = np.concatenate((gt_bbox1, gt_bbox2), axis=0) + result['gt_bbox'] = gt_bbox + if 'gt_class' in sample[0]: + gt_class1 = sample[0]['gt_class'] + gt_class2 = sample[1]['gt_class'] + gt_class = np.concatenate((gt_class1, gt_class2), axis=0) + result['gt_class'] = gt_class + + gt_score1 = np.ones_like(sample[0]['gt_class']) + gt_score2 = np.ones_like(sample[1]['gt_class']) + gt_score = np.concatenate( + (gt_score1 * factor, gt_score2 * (1. - factor)), axis=0) + result['gt_score'] = gt_score + if 'is_crowd' in sample[0]: + is_crowd1 = sample[0]['is_crowd'] + is_crowd2 = sample[1]['is_crowd'] + is_crowd = np.concatenate((is_crowd1, is_crowd2), axis=0) + result['is_crowd'] = is_crowd + if 'difficult' in sample[0]: + is_difficult1 = sample[0]['difficult'] + is_difficult2 = sample[1]['difficult'] + is_difficult = np.concatenate( + (is_difficult1, is_difficult2), axis=0) + result['difficult'] = is_difficult + + if 'gt_ide' in sample[0]: + gt_ide1 = sample[0]['gt_ide'] + gt_ide2 = sample[1]['gt_ide'] + gt_ide = np.concatenate((gt_ide1, gt_ide2), axis=0) + result['gt_ide'] = gt_ide + return result + + +@register_op +class NormalizeBox(BaseOperator): + """Transform the bounding box's coornidates to [0,1].""" + + def __init__(self): + super(NormalizeBox, self).__init__() + + def apply(self, sample, context): + im = sample['image'] + gt_bbox = sample['gt_bbox'] + height, width, _ = im.shape + for i in range(gt_bbox.shape[0]): + gt_bbox[i][0] = gt_bbox[i][0] / width + gt_bbox[i][1] = gt_bbox[i][1] / height + gt_bbox[i][2] = gt_bbox[i][2] / width + gt_bbox[i][3] = gt_bbox[i][3] / height + sample['gt_bbox'] = gt_bbox + + if 'gt_keypoint' in sample.keys(): + gt_keypoint = sample['gt_keypoint'] + + for i in range(gt_keypoint.shape[1]): + if i % 2: + gt_keypoint[:, i] = gt_keypoint[:, i] / height + else: + gt_keypoint[:, i] = gt_keypoint[:, i] / width + sample['gt_keypoint'] = gt_keypoint + + return sample + + +@register_op +class BboxXYXY2XYWH(BaseOperator): + """ + Convert bbox XYXY format to XYWH format. + """ + + def __init__(self): + super(BboxXYXY2XYWH, self).__init__() + + def apply(self, sample, context=None): + assert 'gt_bbox' in sample + bbox = sample['gt_bbox'] + bbox[:, 2:4] = bbox[:, 2:4] - bbox[:, :2] + bbox[:, :2] = bbox[:, :2] + bbox[:, 2:4] / 2. + sample['gt_bbox'] = bbox + return sample + + +@register_op +class PadBox(BaseOperator): + def __init__(self, num_max_boxes=50): + """ + Pad zeros to bboxes if number of bboxes is less than num_max_boxes. + Args: + num_max_boxes (int): the max number of bboxes + """ + self.num_max_boxes = num_max_boxes + super(PadBox, self).__init__() + + def apply(self, sample, context=None): + assert 'gt_bbox' in sample + bbox = sample['gt_bbox'] + gt_num = min(self.num_max_boxes, len(bbox)) + num_max = self.num_max_boxes + # fields = context['fields'] if context else [] + pad_bbox = np.zeros((num_max, 4), dtype=np.float32) + if gt_num > 0: + pad_bbox[:gt_num, :] = bbox[:gt_num, :] + sample['gt_bbox'] = pad_bbox + if 'gt_class' in sample: + pad_class = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_class[:gt_num] = sample['gt_class'][:gt_num, 0] + sample['gt_class'] = pad_class + if 'gt_score' in sample: + pad_score = np.zeros((num_max, ), dtype=np.float32) + if gt_num > 0: + pad_score[:gt_num] = sample['gt_score'][:gt_num, 0] + sample['gt_score'] = pad_score + # in training, for example in op ExpandImage, + # the bbox and gt_class is expandded, but the difficult is not, + # so, judging by it's length + if 'difficult' in sample: + pad_diff = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_diff[:gt_num] = sample['difficult'][:gt_num, 0] + sample['difficult'] = pad_diff + if 'is_crowd' in sample: + pad_crowd = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_crowd[:gt_num] = sample['is_crowd'][:gt_num, 0] + sample['is_crowd'] = pad_crowd + if 'gt_ide' in sample: + pad_ide = np.zeros((num_max, ), dtype=np.int32) + if gt_num > 0: + pad_ide[:gt_num] = sample['gt_ide'][:gt_num, 0] + sample['gt_ide'] = pad_ide + return sample + + +@register_op +class DebugVisibleImage(BaseOperator): + """ + In debug mode, visualize images according to `gt_box`. + (Currently only supported when not cropping and flipping image.) + """ + + def __init__(self, output_dir='output/debug', is_normalized=False): + super(DebugVisibleImage, self).__init__() + self.is_normalized = is_normalized + self.output_dir = output_dir + if not os.path.isdir(output_dir): + os.makedirs(output_dir) + if not isinstance(self.is_normalized, bool): + raise TypeError("{}: input type is invalid.".format(self)) + + def apply(self, sample, context=None): + image = Image.fromarray(sample['image'].astype(np.uint8)) + out_file_name = '{:012d}.jpg'.format(sample['im_id'][0]) + width = sample['w'] + height = sample['h'] + gt_bbox = sample['gt_bbox'] + gt_class = sample['gt_class'] + draw = ImageDraw.Draw(image) + for i in range(gt_bbox.shape[0]): + if self.is_normalized: + gt_bbox[i][0] = gt_bbox[i][0] * width + gt_bbox[i][1] = gt_bbox[i][1] * height + gt_bbox[i][2] = gt_bbox[i][2] * width + gt_bbox[i][3] = gt_bbox[i][3] * height + + xmin, ymin, xmax, ymax = gt_bbox[i] + draw.line( + [(xmin, ymin), (xmin, ymax), (xmax, ymax), (xmax, ymin), + (xmin, ymin)], + width=2, + fill='green') + # draw label + text = str(gt_class[i][0]) + tw, th = draw.textsize(text) + draw.rectangle( + [(xmin + 1, ymin - th), (xmin + tw + 1, ymin)], fill='green') + draw.text((xmin + 1, ymin - th), text, fill=(255, 255, 255)) + + if 'gt_keypoint' in sample.keys(): + gt_keypoint = sample['gt_keypoint'] + if self.is_normalized: + for i in range(gt_keypoint.shape[1]): + if i % 2: + gt_keypoint[:, i] = gt_keypoint[:, i] * height + else: + gt_keypoint[:, i] = gt_keypoint[:, i] * width + for i in range(gt_keypoint.shape[0]): + keypoint = gt_keypoint[i] + for j in range(int(keypoint.shape[0] / 2)): + x1 = round(keypoint[2 * j]).astype(np.int32) + y1 = round(keypoint[2 * j + 1]).astype(np.int32) + draw.ellipse( + (x1, y1, x1 + 5, y1 + 5), fill='green', outline='green') + save_path = os.path.join(self.output_dir, out_file_name) + image.save(save_path, quality=95) + return sample + + +@register_op +class Pad(BaseOperator): + def __init__(self, + size=None, + size_divisor=32, + pad_mode=0, + offsets=None, + fill_value=(127.5, 127.5, 127.5)): + """ + Pad image to a specified size or multiple of size_divisor. + Args: + size (int, Sequence): image target size, if None, pad to multiple of size_divisor, default None + size_divisor (int): size divisor, default 32 + pad_mode (int): pad mode, currently only supports four modes [-1, 0, 1, 2]. if -1, use specified offsets + if 0, only pad to right and bottom. if 1, pad according to center. if 2, only pad left and top + offsets (list): [offset_x, offset_y], specify offset while padding, only supported pad_mode=-1 + fill_value (bool): rgb value of pad area, default (127.5, 127.5, 127.5) + """ + super(Pad, self).__init__() + + if not isinstance(size, (int, Sequence)): + raise TypeError( + "Type of target_size is invalid when random_size is True. \ + Must be List, now is {}".format(type(size))) + + if isinstance(size, int): + size = [size, size] + + assert pad_mode in [ + -1, 0, 1, 2 + ], 'currently only supports four modes [-1, 0, 1, 2]' + if pad_mode == -1: + assert offsets, 'if pad_mode is -1, offsets should not be None' + + self.size = size + self.size_divisor = size_divisor + self.pad_mode = pad_mode + self.fill_value = fill_value + self.offsets = offsets + + def apply_segm(self, segms, offsets, im_size, size): + def _expand_poly(poly, x, y): + expanded_poly = np.array(poly) + expanded_poly[0::2] += x + expanded_poly[1::2] += y + return expanded_poly.tolist() + + def _expand_rle(rle, x, y, height, width, h, w): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, height, width) + mask = mask_util.decode(rle) + expanded_mask = np.full((h, w), 0).astype(mask.dtype) + expanded_mask[y:y + height, x:x + width] = mask + rle = mask_util.encode( + np.array( + expanded_mask, order='F', dtype=np.uint8)) + return rle + + x, y = offsets + height, width = im_size + h, w = size + expanded_segms = [] + for segm in segms: + if is_poly(segm): + # Polygon format + expanded_segms.append( + [_expand_poly(poly, x, y) for poly in segm]) + else: + # RLE format + import pycocotools.mask as mask_util + expanded_segms.append( + _expand_rle(segm, x, y, height, width, h, w)) + return expanded_segms + + def apply_bbox(self, bbox, offsets): + return bbox + np.array(offsets * 2, dtype=np.float32) + + def apply_keypoint(self, keypoints, offsets): + n = len(keypoints[0]) // 2 + return keypoints + np.array(offsets * n, dtype=np.float32) + + def apply_image(self, image, offsets, im_size, size): + x, y = offsets + im_h, im_w = im_size + h, w = size + canvas = np.ones((h, w, 3), dtype=np.float32) + canvas *= np.array(self.fill_value, dtype=np.float32) + canvas[y:y + im_h, x:x + im_w, :] = image.astype(np.float32) + return canvas + + def apply(self, sample, context=None): + im = sample['image'] + im_h, im_w = im.shape[:2] + if self.size: + h, w = self.size + assert ( + im_h < h and im_w < w + ), '(h, w) of target size should be greater than (im_h, im_w)' + else: + h = np.ceil(im_h / self.size_divisor) * self.size_divisor + w = np.ceil(im_w / self.size_divisor) * self.size_divisor + + if h == im_h and w == im_w: + return sample + + if self.pad_mode == -1: + offset_x, offset_y = self.offsets + elif self.pad_mode == 0: + offset_y, offset_x = 0, 0 + elif self.pad_mode == 1: + offset_y, offset_x = (h - im_h) // 2, (w - im_w) // 2 + else: + offset_y, offset_x = h - im_h, w - im_w + + offsets, im_size, size = [offset_x, offset_y], [im_h, im_w], [h, w] + + sample['image'] = self.apply_image(im, offsets, im_size, size) + + if self.pad_mode == 0: + return sample + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], offsets) + + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], offsets, + im_size, size) + + if 'gt_keypoint' in sample and len(sample['gt_keypoint']) > 0: + sample['gt_keypoint'] = self.apply_keypoint(sample['gt_keypoint'], + offsets) + + return sample + + +@register_op +class Poly2Mask(BaseOperator): + """ + gt poly to mask annotations + """ + + def __init__(self): + super(Poly2Mask, self).__init__() + import pycocotools.mask as maskUtils + self.maskutils = maskUtils + + def _poly2mask(self, mask_ann, img_h, img_w): + if isinstance(mask_ann, list): + # polygon -- a single object might consist of multiple parts + # we merge all parts into one mask rle code + rles = self.maskutils.frPyObjects(mask_ann, img_h, img_w) + rle = self.maskutils.merge(rles) + elif isinstance(mask_ann['counts'], list): + # uncompressed RLE + rle = self.maskutils.frPyObjects(mask_ann, img_h, img_w) + else: + # rle + rle = mask_ann + mask = self.maskutils.decode(rle) + return mask + + def apply(self, sample, context=None): + assert 'gt_poly' in sample + im_h = sample['h'] + im_w = sample['w'] + masks = [ + self._poly2mask(gt_poly, im_h, im_w) + for gt_poly in sample['gt_poly'] + ] + sample['gt_segm'] = np.asarray(masks).astype(np.uint8) + return sample + + +@register_op +class Rbox2Poly(BaseOperator): + """ + Convert rbbox format to poly format. + """ + + def __init__(self): + super(Rbox2Poly, self).__init__() + + def apply(self, sample, context=None): + assert 'gt_rbox' in sample + assert sample['gt_rbox'].shape[1] == 5 + rrects = sample['gt_rbox'] + x_ctr = rrects[:, 0] + y_ctr = rrects[:, 1] + width = rrects[:, 2] + height = rrects[:, 3] + x1 = x_ctr - width / 2.0 + y1 = y_ctr - height / 2.0 + x2 = x_ctr + width / 2.0 + y2 = y_ctr + height / 2.0 + sample['gt_bbox'] = np.stack([x1, y1, x2, y2], axis=1) + polys = bbox_utils.rbox2poly_np(rrects) + sample['gt_rbox2poly'] = polys + return sample + + +@register_op +class AugmentHSV(BaseOperator): + def __init__(self, fraction=0.50, is_bgr=True): + """ + Augment the SV channel of image data. + Args: + fraction (float): the fraction for augment. Default: 0.5. + is_bgr (bool): whether the image is BGR mode. Default: True. + """ + super(AugmentHSV, self).__init__() + self.fraction = fraction + self.is_bgr = is_bgr + + def apply(self, sample, context=None): + img = sample['image'] + if self.is_bgr: + img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV) + else: + img_hsv = cv2.cvtColor(img, cv2.COLOR_RGB2HSV) + S = img_hsv[:, :, 1].astype(np.float32) + V = img_hsv[:, :, 2].astype(np.float32) + + a = (random.random() * 2 - 1) * self.fraction + 1 + S *= a + if a > 1: + np.clip(S, a_min=0, a_max=255, out=S) + + a = (random.random() * 2 - 1) * self.fraction + 1 + V *= a + if a > 1: + np.clip(V, a_min=0, a_max=255, out=V) + + img_hsv[:, :, 1] = S.astype(np.uint8) + img_hsv[:, :, 2] = V.astype(np.uint8) + if self.is_bgr: + cv2.cvtColor(img_hsv, cv2.COLOR_HSV2BGR, dst=img) + else: + cv2.cvtColor(img_hsv, cv2.COLOR_HSV2RGB, dst=img) + + sample['image'] = img + return sample + + +@register_op +class Norm2PixelBbox(BaseOperator): + """ + Transform the bounding box's coornidates which is in [0,1] to pixels. + """ + + def __init__(self): + super(Norm2PixelBbox, self).__init__() + + def apply(self, sample, context=None): + assert 'gt_bbox' in sample + bbox = sample['gt_bbox'] + height, width = sample['image'].shape[:2] + bbox[:, 0::2] = bbox[:, 0::2] * width + bbox[:, 1::2] = bbox[:, 1::2] * height + sample['gt_bbox'] = bbox + return sample + + +@register_op +class BboxCXCYWH2XYXY(BaseOperator): + """ + Convert bbox CXCYWH format to XYXY format. + [center_x, center_y, width, height] -> [x0, y0, x1, y1] + """ + + def __init__(self): + super(BboxCXCYWH2XYXY, self).__init__() + + def apply(self, sample, context=None): + assert 'gt_bbox' in sample + bbox0 = sample['gt_bbox'] + bbox = bbox0.copy() + + bbox[:, :2] = bbox0[:, :2] - bbox0[:, 2:4] / 2. + bbox[:, 2:4] = bbox0[:, :2] + bbox0[:, 2:4] / 2. + sample['gt_bbox'] = bbox + return sample + + +@register_op +class RandomResizeCrop(BaseOperator): + """Random resize and crop image and bboxes. + Args: + resizes (list): resize image to one of resizes. if keep_ratio is True and mode is + 'long', resize the image's long side to the maximum of target_size, if keep_ratio is + True and mode is 'short', resize the image's short side to the minimum of target_size. + cropsizes (list): crop sizes after resize, [(min_crop_1, max_crop_1), ...] + mode (str): resize mode, `long` or `short`. Details see resizes. + prob (float): probability of this op. + keep_ratio (bool): whether keep_ratio or not, default true + interp (int): the interpolation method + thresholds (list): iou thresholds for decide a valid bbox crop. + num_attempts (int): number of tries before giving up. + allow_no_crop (bool): allow return without actually cropping them. + cover_all_box (bool): ensure all bboxes are covered in the final crop. + is_mask_crop(bool): whether crop the segmentation. + """ + + def __init__( + self, + resizes, + cropsizes, + prob=0.5, + mode='short', + keep_ratio=True, + interp=cv2.INTER_LINEAR, + num_attempts=3, + cover_all_box=False, + allow_no_crop=False, + thresholds=[0.3, 0.5, 0.7], + is_mask_crop=False, ): + super(RandomResizeCrop, self).__init__() + + self.resizes = resizes + self.cropsizes = cropsizes + self.prob = prob + self.mode = mode + + self.resizer = Resize(0, keep_ratio=keep_ratio, interp=interp) + self.croper = RandomCrop( + num_attempts=num_attempts, + cover_all_box=cover_all_box, + thresholds=thresholds, + allow_no_crop=allow_no_crop, + is_mask_crop=is_mask_crop) + + def _format_size(self, size): + if isinstance(size, Integral): + size = (size, size) + return size + + def apply(self, sample, context=None): + if random.random() < self.prob: + _resize = self._format_size(random.choice(self.resizes)) + _cropsize = self._format_size(random.choice(self.cropsizes)) + sample = self._resize( + self.resizer, + sample, + size=_resize, + mode=self.mode, + context=context) + sample = self._random_crop( + self.croper, sample, size=_cropsize, context=context) + return sample + + @staticmethod + def _random_crop(croper, sample, size, context=None): + if 'gt_bbox' in sample and len(sample['gt_bbox']) == 0: + return sample + + self = croper + h, w = sample['image'].shape[:2] + gt_bbox = sample['gt_bbox'] + cropsize = size + min_crop = min(cropsize) + max_crop = max(cropsize) + + thresholds = list(self.thresholds) + np.random.shuffle(thresholds) + + for thresh in thresholds: + found = False + for _ in range(self.num_attempts): + + crop_h = random.randint(min_crop, min(h, max_crop)) + crop_w = random.randint(min_crop, min(w, max_crop)) + + crop_y = random.randint(0, h - crop_h) + crop_x = random.randint(0, w - crop_w) + + crop_box = [crop_x, crop_y, crop_x + crop_w, crop_y + crop_h] + iou = self._iou_matrix( + gt_bbox, np.array( + [crop_box], dtype=np.float32)) + if iou.max() < thresh: + continue + + if self.cover_all_box and iou.min() < thresh: + continue + + cropped_box, valid_ids = self._crop_box_with_center_constraint( + gt_bbox, np.array( + crop_box, dtype=np.float32)) + if valid_ids.size > 0: + found = True + break + + if found: + if self.is_mask_crop and 'gt_poly' in sample and len(sample[ + 'gt_poly']) > 0: + crop_polys = self.crop_segms( + sample['gt_poly'], + valid_ids, + np.array( + crop_box, dtype=np.int64), + h, + w) + if [] in crop_polys: + delete_id = list() + valid_polys = list() + for id, crop_poly in enumerate(crop_polys): + if crop_poly == []: + delete_id.append(id) + else: + valid_polys.append(crop_poly) + valid_ids = np.delete(valid_ids, delete_id) + if len(valid_polys) == 0: + return sample + sample['gt_poly'] = valid_polys + else: + sample['gt_poly'] = crop_polys + + if 'gt_segm' in sample: + sample['gt_segm'] = self._crop_segm(sample['gt_segm'], + crop_box) + sample['gt_segm'] = np.take( + sample['gt_segm'], valid_ids, axis=0) + + sample['image'] = self._crop_image(sample['image'], crop_box) + sample['gt_bbox'] = np.take(cropped_box, valid_ids, axis=0) + sample['gt_class'] = np.take( + sample['gt_class'], valid_ids, axis=0) + if 'gt_score' in sample: + sample['gt_score'] = np.take( + sample['gt_score'], valid_ids, axis=0) + + if 'is_crowd' in sample: + sample['is_crowd'] = np.take( + sample['is_crowd'], valid_ids, axis=0) + return sample + + return sample + + @staticmethod + def _resize(resizer, sample, size, mode='short', context=None): + self = resizer + im = sample['image'] + target_size = size + + if not isinstance(im, np.ndarray): + raise TypeError("{}: image type is not numpy.".format(self)) + if len(im.shape) != 3: + raise ImageError('{}: image is not 3-dimensional.'.format(self)) + + # apply image + im_shape = im.shape + if self.keep_ratio: + + im_size_min = np.min(im_shape[0:2]) + im_size_max = np.max(im_shape[0:2]) + + target_size_min = np.min(target_size) + target_size_max = np.max(target_size) + + if mode == 'long': + im_scale = min(target_size_min / im_size_min, + target_size_max / im_size_max) + else: + im_scale = max(target_size_min / im_size_min, + target_size_max / im_size_max) + + resize_h = im_scale * float(im_shape[0]) + resize_w = im_scale * float(im_shape[1]) + + im_scale_x = im_scale + im_scale_y = im_scale + else: + resize_h, resize_w = target_size + im_scale_y = resize_h / im_shape[0] + im_scale_x = resize_w / im_shape[1] + + im = self.apply_image(sample['image'], [im_scale_x, im_scale_y]) + sample['image'] = im + sample['im_shape'] = np.asarray([resize_h, resize_w], dtype=np.float32) + if 'scale_factor' in sample: + scale_factor = sample['scale_factor'] + sample['scale_factor'] = np.asarray( + [scale_factor[0] * im_scale_y, scale_factor[1] * im_scale_x], + dtype=np.float32) + else: + sample['scale_factor'] = np.asarray( + [im_scale_y, im_scale_x], dtype=np.float32) + + # apply bbox + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], + [im_scale_x, im_scale_y], + [resize_w, resize_h]) + + # apply rbox + if 'gt_rbox2poly' in sample: + if np.array(sample['gt_rbox2poly']).shape[1] != 8: + logger.warn( + "gt_rbox2poly's length shoule be 8, but actually is {}". + format(len(sample['gt_rbox2poly']))) + sample['gt_rbox2poly'] = self.apply_bbox(sample['gt_rbox2poly'], + [im_scale_x, im_scale_y], + [resize_w, resize_h]) + + # apply polygon + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], im_shape[:2], + [im_scale_x, im_scale_y]) + + # apply semantic + if 'semantic' in sample and sample['semantic']: + semantic = sample['semantic'] + semantic = cv2.resize( + semantic.astype('float32'), + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + semantic = np.asarray(semantic).astype('int32') + semantic = np.expand_dims(semantic, 0) + sample['semantic'] = semantic + + # apply gt_segm + if 'gt_segm' in sample and len(sample['gt_segm']) > 0: + masks = [ + cv2.resize( + gt_segm, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=cv2.INTER_NEAREST) + for gt_segm in sample['gt_segm'] + ] + sample['gt_segm'] = np.asarray(masks).astype(np.uint8) + + return sample + + +@register_op +class RandomSelect(BaseOperator): + """ + Randomly choose a transformation between transforms1 and transforms2, + and the probability of choosing transforms1 is p. + + The code is based on https://github.com/facebookresearch/detr/blob/main/datasets/transforms.py + + """ + + def __init__(self, transforms1, transforms2, p=0.5): + super(RandomSelect, self).__init__() + self.transforms1 = Compose(transforms1) + self.transforms2 = Compose(transforms2) + self.p = p + + def apply(self, sample, context=None): + if random.random() < self.p: + return self.transforms1(sample) + return self.transforms2(sample) + + +@register_op +class RandomShortSideResize(BaseOperator): + def __init__(self, + short_side_sizes, + max_size=None, + interp=cv2.INTER_LINEAR, + random_interp=False): + """ + Resize the image randomly according to the short side. If max_size is not None, + the long side is scaled according to max_size. The whole process will be keep ratio. + Args: + short_side_sizes (list|tuple): Image target short side size. + max_size (int): The size of the longest side of image after resize. + interp (int): The interpolation method. + random_interp (bool): Whether random select interpolation method. + """ + super(RandomShortSideResize, self).__init__() + + assert isinstance(short_side_sizes, + Sequence), "short_side_sizes must be List or Tuple" + + self.short_side_sizes = short_side_sizes + self.max_size = max_size + self.interp = interp + self.random_interp = random_interp + self.interps = [ + cv2.INTER_NEAREST, + cv2.INTER_LINEAR, + cv2.INTER_AREA, + cv2.INTER_CUBIC, + cv2.INTER_LANCZOS4, + ] + + def get_size_with_aspect_ratio(self, image_shape, size, max_size=None): + h, w = image_shape + if max_size is not None: + min_original_size = float(min((w, h))) + max_original_size = float(max((w, h))) + if max_original_size / min_original_size * size > max_size: + size = int( + round(max_size * min_original_size / max_original_size)) + + if (w <= h and w == size) or (h <= w and h == size): + return (w, h) + + if w < h: + ow = size + oh = int(size * h / w) + else: + oh = size + ow = int(size * w / h) + + return (ow, oh) + + def resize(self, + sample, + target_size, + max_size=None, + interp=cv2.INTER_LINEAR): + im = sample['image'] + if not isinstance(im, np.ndarray): + raise TypeError("{}: image type is not numpy.".format(self)) + if len(im.shape) != 3: + raise ImageError('{}: image is not 3-dimensional.'.format(self)) + + target_size = self.get_size_with_aspect_ratio(im.shape[:2], target_size, + max_size) + im_scale_y, im_scale_x = target_size[1] / im.shape[0], target_size[ + 0] / im.shape[1] + + sample['image'] = cv2.resize(im, target_size, interpolation=interp) + sample['im_shape'] = np.asarray(target_size[::-1], dtype=np.float32) + if 'scale_factor' in sample: + scale_factor = sample['scale_factor'] + sample['scale_factor'] = np.asarray( + [scale_factor[0] * im_scale_y, scale_factor[1] * im_scale_x], + dtype=np.float32) + else: + sample['scale_factor'] = np.asarray( + [im_scale_y, im_scale_x], dtype=np.float32) + + # apply bbox + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox( + sample['gt_bbox'], [im_scale_x, im_scale_y], target_size) + # apply polygon + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], im.shape[:2], + [im_scale_x, im_scale_y]) + # apply semantic + if 'semantic' in sample and sample['semantic']: + semantic = sample['semantic'] + semantic = cv2.resize( + semantic.astype('float32'), + target_size, + interpolation=self.interp) + semantic = np.asarray(semantic).astype('int32') + semantic = np.expand_dims(semantic, 0) + sample['semantic'] = semantic + # apply gt_segm + if 'gt_segm' in sample and len(sample['gt_segm']) > 0: + masks = [ + cv2.resize( + gt_segm, target_size, interpolation=cv2.INTER_NEAREST) + for gt_segm in sample['gt_segm'] + ] + sample['gt_segm'] = np.asarray(masks).astype(np.uint8) + return sample + + def apply_bbox(self, bbox, scale, size): + im_scale_x, im_scale_y = scale + resize_w, resize_h = size + bbox[:, 0::2] *= im_scale_x + bbox[:, 1::2] *= im_scale_y + bbox[:, 0::2] = np.clip(bbox[:, 0::2], 0, resize_w) + bbox[:, 1::2] = np.clip(bbox[:, 1::2], 0, resize_h) + return bbox.astype('float32') + + def apply_segm(self, segms, im_size, scale): + def _resize_poly(poly, im_scale_x, im_scale_y): + resized_poly = np.array(poly).astype('float32') + resized_poly[0::2] *= im_scale_x + resized_poly[1::2] *= im_scale_y + return resized_poly.tolist() + + def _resize_rle(rle, im_h, im_w, im_scale_x, im_scale_y): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, im_h, im_w) + + mask = mask_util.decode(rle) + mask = cv2.resize( + mask, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + rle = mask_util.encode(np.array(mask, order='F', dtype=np.uint8)) + return rle + + im_h, im_w = im_size + im_scale_x, im_scale_y = scale + resized_segms = [] + for segm in segms: + if is_poly(segm): + # Polygon format + resized_segms.append([ + _resize_poly(poly, im_scale_x, im_scale_y) for poly in segm + ]) + else: + # RLE format + import pycocotools.mask as mask_util + resized_segms.append( + _resize_rle(segm, im_h, im_w, im_scale_x, im_scale_y)) + + return resized_segms + + def apply(self, sample, context=None): + target_size = random.choice(self.short_side_sizes) + interp = random.choice( + self.interps) if self.random_interp else self.interp + + return self.resize(sample, target_size, self.max_size, interp) + + +@register_op +class RandomSizeCrop(BaseOperator): + """ + Cut the image randomly according to `min_size` and `max_size` + """ + + def __init__(self, min_size, max_size): + super(RandomSizeCrop, self).__init__() + self.min_size = min_size + self.max_size = max_size + + from paddle.vision.transforms.functional import crop as paddle_crop + self.paddle_crop = paddle_crop + + @staticmethod + def get_crop_params(img_shape, output_size): + """Get parameters for ``crop`` for a random crop. + Args: + img_shape (list|tuple): Image's height and width. + output_size (list|tuple): Expected output size of the crop. + Returns: + tuple: params (i, j, h, w) to be passed to ``crop`` for random crop. + """ + h, w = img_shape + th, tw = output_size + + if h + 1 < th or w + 1 < tw: + raise ValueError( + "Required crop size {} is larger then input image size {}". + format((th, tw), (h, w))) + + if w == tw and h == th: + return 0, 0, h, w + + i = random.randint(0, h - th + 1) + j = random.randint(0, w - tw + 1) + return i, j, th, tw + + def crop(self, sample, region): + image_shape = sample['image'].shape[:2] + sample['image'] = self.paddle_crop(sample['image'], *region) + + keep_index = None + # apply bbox + if 'gt_bbox' in sample and len(sample['gt_bbox']) > 0: + sample['gt_bbox'] = self.apply_bbox(sample['gt_bbox'], region) + bbox = sample['gt_bbox'].reshape([-1, 2, 2]) + area = (bbox[:, 1, :] - bbox[:, 0, :]).prod(axis=1) + keep_index = np.where(area > 0)[0] + sample['gt_bbox'] = sample['gt_bbox'][keep_index] if len( + keep_index) > 0 else np.zeros( + [0, 4], dtype=np.float32) + sample['gt_class'] = sample['gt_class'][keep_index] if len( + keep_index) > 0 else np.zeros( + [0, 1], dtype=np.float32) + if 'gt_score' in sample: + sample['gt_score'] = sample['gt_score'][keep_index] if len( + keep_index) > 0 else np.zeros( + [0, 1], dtype=np.float32) + if 'is_crowd' in sample: + sample['is_crowd'] = sample['is_crowd'][keep_index] if len( + keep_index) > 0 else np.zeros( + [0, 1], dtype=np.float32) + + # apply polygon + if 'gt_poly' in sample and len(sample['gt_poly']) > 0: + sample['gt_poly'] = self.apply_segm(sample['gt_poly'], region, + image_shape) + if keep_index is not None: + sample['gt_poly'] = sample['gt_poly'][keep_index] + # apply gt_segm + if 'gt_segm' in sample and len(sample['gt_segm']) > 0: + i, j, h, w = region + sample['gt_segm'] = sample['gt_segm'][:, i:i + h, j:j + w] + if keep_index is not None: + sample['gt_segm'] = sample['gt_segm'][keep_index] + + return sample + + def apply_bbox(self, bbox, region): + i, j, h, w = region + region_size = np.asarray([w, h]) + crop_bbox = bbox - np.asarray([j, i, j, i]) + crop_bbox = np.minimum(crop_bbox.reshape([-1, 2, 2]), region_size) + crop_bbox = crop_bbox.clip(min=0) + return crop_bbox.reshape([-1, 4]).astype('float32') + + def apply_segm(self, segms, region, image_shape): + def _crop_poly(segm, crop): + xmin, ymin, xmax, ymax = crop + crop_coord = [xmin, ymin, xmin, ymax, xmax, ymax, xmax, ymin] + crop_p = np.array(crop_coord).reshape(4, 2) + crop_p = Polygon(crop_p) + + crop_segm = list() + for poly in segm: + poly = np.array(poly).reshape(len(poly) // 2, 2) + polygon = Polygon(poly) + if not polygon.is_valid: + exterior = polygon.exterior + multi_lines = exterior.intersection(exterior) + polygons = shapely.ops.polygonize(multi_lines) + polygon = MultiPolygon(polygons) + multi_polygon = list() + if isinstance(polygon, MultiPolygon): + multi_polygon = copy.deepcopy(polygon) + else: + multi_polygon.append(copy.deepcopy(polygon)) + for per_polygon in multi_polygon: + inter = per_polygon.intersection(crop_p) + if not inter: + continue + if isinstance(inter, (MultiPolygon, GeometryCollection)): + for part in inter: + if not isinstance(part, Polygon): + continue + part = np.squeeze( + np.array(part.exterior.coords[:-1]).reshape(1, + -1)) + part[0::2] -= xmin + part[1::2] -= ymin + crop_segm.append(part.tolist()) + elif isinstance(inter, Polygon): + crop_poly = np.squeeze( + np.array(inter.exterior.coords[:-1]).reshape(1, -1)) + crop_poly[0::2] -= xmin + crop_poly[1::2] -= ymin + crop_segm.append(crop_poly.tolist()) + else: + continue + return crop_segm + + def _crop_rle(rle, crop, height, width): + if 'counts' in rle and type(rle['counts']) == list: + rle = mask_util.frPyObjects(rle, height, width) + mask = mask_util.decode(rle) + mask = mask[crop[1]:crop[3], crop[0]:crop[2]] + rle = mask_util.encode(np.array(mask, order='F', dtype=np.uint8)) + return rle + + i, j, h, w = region + crop = [j, i, j + w, i + h] + height, width = image_shape + crop_segms = [] + for segm in segms: + if is_poly(segm): + import copy + import shapely.ops + from shapely.geometry import Polygon, MultiPolygon, GeometryCollection + # Polygon format + crop_segms.append(_crop_poly(segm, crop)) + else: + # RLE format + import pycocotools.mask as mask_util + crop_segms.append(_crop_rle(segm, crop, height, width)) + return crop_segms + + def apply(self, sample, context=None): + h = random.randint(self.min_size, + min(sample['image'].shape[0], self.max_size)) + w = random.randint(self.min_size, + min(sample['image'].shape[1], self.max_size)) + + region = self.get_crop_params(sample['image'].shape[:2], [h, w]) + return self.crop(sample, region) + + +@register_op +class WarpAffine(BaseOperator): + def __init__(self, + keep_res=False, + pad=31, + input_h=512, + input_w=512, + scale=0.4, + shift=0.1): + """WarpAffine + Warp affine the image + + The code is based on https://github.com/xingyizhou/CenterNet/blob/master/src/lib/datasets/sample/ctdet.py + + + """ + super(WarpAffine, self).__init__() + self.keep_res = keep_res + self.pad = pad + self.input_h = input_h + self.input_w = input_w + self.scale = scale + self.shift = shift + + def apply(self, sample, context=None): + img = sample['image'] + img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR) + if 'gt_bbox' in sample and len(sample['gt_bbox']) == 0: + return sample + + h, w = img.shape[:2] + + if self.keep_res: + input_h = (h | self.pad) + 1 + input_w = (w | self.pad) + 1 + s = np.array([input_w, input_h], dtype=np.float32) + c = np.array([w // 2, h // 2], dtype=np.float32) + + else: + s = max(h, w) * 1.0 + input_h, input_w = self.input_h, self.input_w + c = np.array([w / 2., h / 2.], dtype=np.float32) + + trans_input = get_affine_transform(c, s, 0, [input_w, input_h]) + img = cv2.resize(img, (w, h)) + inp = cv2.warpAffine( + img, trans_input, (input_w, input_h), flags=cv2.INTER_LINEAR) + sample['image'] = inp + return sample + + +@register_op +class FlipWarpAffine(BaseOperator): + def __init__(self, + keep_res=False, + pad=31, + input_h=512, + input_w=512, + not_rand_crop=False, + scale=0.4, + shift=0.1, + flip=0.5, + is_scale=True, + use_random=True): + """FlipWarpAffine + 1. Random Crop + 2. Flip the image horizontal + 3. Warp affine the image + """ + super(FlipWarpAffine, self).__init__() + self.keep_res = keep_res + self.pad = pad + self.input_h = input_h + self.input_w = input_w + self.not_rand_crop = not_rand_crop + self.scale = scale + self.shift = shift + self.flip = flip + self.is_scale = is_scale + self.use_random = use_random + + def apply(self, sample, context=None): + img = sample['image'] + img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR) + if 'gt_bbox' in sample and len(sample['gt_bbox']) == 0: + return sample + + h, w = img.shape[:2] + + if self.keep_res: + input_h = (h | self.pad) + 1 + input_w = (w | self.pad) + 1 + s = np.array([input_w, input_h], dtype=np.float32) + c = np.array([w // 2, h // 2], dtype=np.float32) + + else: + s = max(h, w) * 1.0 + input_h, input_w = self.input_h, self.input_w + c = np.array([w / 2., h / 2.], dtype=np.float32) + + if self.use_random: + gt_bbox = sample['gt_bbox'] + if not self.not_rand_crop: + s = s * np.random.choice(np.arange(0.6, 1.4, 0.1)) + w_border = get_border(128, w) + h_border = get_border(128, h) + c[0] = np.random.randint(low=w_border, high=w - w_border) + c[1] = np.random.randint(low=h_border, high=h - h_border) + else: + sf = self.scale + cf = self.shift + c[0] += s * np.clip(np.random.randn() * cf, -2 * cf, 2 * cf) + c[1] += s * np.clip(np.random.randn() * cf, -2 * cf, 2 * cf) + s = s * np.clip(np.random.randn() * sf + 1, 1 - sf, 1 + sf) + + if np.random.random() < self.flip: + img = img[:, ::-1, :] + c[0] = w - c[0] - 1 + oldx1 = gt_bbox[:, 0].copy() + oldx2 = gt_bbox[:, 2].copy() + gt_bbox[:, 0] = w - oldx2 - 1 + gt_bbox[:, 2] = w - oldx1 - 1 + sample['gt_bbox'] = gt_bbox + + trans_input = get_affine_transform(c, s, 0, [input_w, input_h]) + if not self.use_random: + img = cv2.resize(img, (w, h)) + inp = cv2.warpAffine( + img, trans_input, (input_w, input_h), flags=cv2.INTER_LINEAR) + if self.is_scale: + inp = (inp.astype(np.float32) / 255.) + sample['image'] = inp + sample['center'] = c + sample['scale'] = s + return sample + + +@register_op +class CenterRandColor(BaseOperator): + """Random color for CenterNet series models. + Args: + saturation (float): saturation settings. + contrast (float): contrast settings. + brightness (float): brightness settings. + """ + + def __init__(self, saturation=0.4, contrast=0.4, brightness=0.4): + super(CenterRandColor, self).__init__() + self.saturation = saturation + self.contrast = contrast + self.brightness = brightness + + def apply_saturation(self, img, img_gray): + alpha = 1. + np.random.uniform( + low=-self.saturation, high=self.saturation) + self._blend(alpha, img, img_gray[:, :, None]) + return img + + def apply_contrast(self, img, img_gray): + alpha = 1. + np.random.uniform(low=-self.contrast, high=self.contrast) + img_mean = img_gray.mean() + self._blend(alpha, img, img_mean) + return img + + def apply_brightness(self, img, img_gray): + alpha = 1 + np.random.uniform( + low=-self.brightness, high=self.brightness) + img *= alpha + return img + + def _blend(self, alpha, img, img_mean): + img *= alpha + img_mean *= (1 - alpha) + img += img_mean + + def __call__(self, sample, context=None): + img = sample['image'] + img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) + functions = [ + self.apply_brightness, + self.apply_contrast, + self.apply_saturation, + ] + distortions = np.random.permutation(functions) + for func in distortions: + img = func(img, img_gray) + sample['image'] = img + return sample diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__init__.py new file mode 100644 index 000000000..9d14ee634 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__init__.py @@ -0,0 +1,30 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import trainer +from .trainer import * + +from . import callbacks +from .callbacks import * + +from . import env +from .env import * + +__all__ = trainer.__all__ \ + + callbacks.__all__ \ + + env.__all__ + +from . import tracker +from .tracker import * +__all__ = __all__ + tracker.__all__ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..3720322a3 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/callbacks.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/callbacks.cpython-37.pyc new file mode 100644 index 000000000..fa174b4c0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/callbacks.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/env.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/env.cpython-37.pyc new file mode 100644 index 000000000..c8937c83a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/env.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/export_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/export_utils.cpython-37.pyc new file mode 100644 index 000000000..0ec950e26 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/export_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/tracker.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/tracker.cpython-37.pyc new file mode 100644 index 000000000..179fd496d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/tracker.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/trainer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/trainer.cpython-37.pyc new file mode 100644 index 000000000..e3294bdaa Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/__pycache__/trainer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/callbacks.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/callbacks.py new file mode 100644 index 000000000..df42a687c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/callbacks.py @@ -0,0 +1,335 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys +import datetime +import six +import copy +import json + +import paddle +import paddle.distributed as dist + +from ppdet.utils.checkpoint import save_model +from ppdet.metrics import get_infer_results + +from ppdet.utils.logger import setup_logger +logger = setup_logger('ppdet.engine') + +__all__ = ['Callback', 'ComposeCallback', 'LogPrinter', 'Checkpointer', 'VisualDLWriter', 'SniperProposalsGenerator'] + + +class Callback(object): + def __init__(self, model): + self.model = model + + def on_step_begin(self, status): + pass + + def on_step_end(self, status): + pass + + def on_epoch_begin(self, status): + pass + + def on_epoch_end(self, status): + pass + + def on_train_begin(self, status): + pass + + def on_train_end(self, status): + pass + + +class ComposeCallback(object): + def __init__(self, callbacks): + callbacks = [c for c in list(callbacks) if c is not None] + for c in callbacks: + assert isinstance( + c, Callback), "callback should be subclass of Callback" + self._callbacks = callbacks + + def on_step_begin(self, status): + for c in self._callbacks: + c.on_step_begin(status) + + def on_step_end(self, status): + for c in self._callbacks: + c.on_step_end(status) + + def on_epoch_begin(self, status): + for c in self._callbacks: + c.on_epoch_begin(status) + + def on_epoch_end(self, status): + for c in self._callbacks: + c.on_epoch_end(status) + + def on_train_begin(self, status): + for c in self._callbacks: + c.on_train_begin(status) + + def on_train_end(self, status): + for c in self._callbacks: + c.on_train_end(status) + + +class LogPrinter(Callback): + def __init__(self, model): + super(LogPrinter, self).__init__(model) + + def on_step_end(self, status): + if dist.get_world_size() < 2 or dist.get_rank() == 0: + mode = status['mode'] + if mode == 'train': + epoch_id = status['epoch_id'] + step_id = status['step_id'] + steps_per_epoch = status['steps_per_epoch'] + training_staus = status['training_staus'] + batch_time = status['batch_time'] + data_time = status['data_time'] + + epoches = self.model.cfg.epoch + batch_size = self.model.cfg['{}Reader'.format(mode.capitalize( + ))]['batch_size'] + + logs = training_staus.log() + space_fmt = ':' + str(len(str(steps_per_epoch))) + 'd' + if step_id % self.model.cfg.log_iter == 0: + eta_steps = (epoches - epoch_id) * steps_per_epoch - step_id + eta_sec = eta_steps * batch_time.global_avg + eta_str = str(datetime.timedelta(seconds=int(eta_sec))) + ips = float(batch_size) / batch_time.avg + fmt = ' '.join([ + 'Epoch: [{}]', + '[{' + space_fmt + '}/{}]', + 'learning_rate: {lr:.6f}', + '{meters}', + 'eta: {eta}', + 'batch_cost: {btime}', + 'data_cost: {dtime}', + 'ips: {ips:.4f} images/s', + ]) + fmt = fmt.format( + epoch_id, + step_id, + steps_per_epoch, + lr=status['learning_rate'], + meters=logs, + eta=eta_str, + btime=str(batch_time), + dtime=str(data_time), + ips=ips) + logger.info(fmt) + if mode == 'eval': + step_id = status['step_id'] + if step_id % 100 == 0: + logger.info("Eval iter: {}".format(step_id)) + + def on_epoch_end(self, status): + if dist.get_world_size() < 2 or dist.get_rank() == 0: + mode = status['mode'] + if mode == 'eval': + sample_num = status['sample_num'] + cost_time = status['cost_time'] + logger.info('Total sample number: {}, averge FPS: {}'.format( + sample_num, sample_num / cost_time)) + + +class Checkpointer(Callback): + def __init__(self, model): + super(Checkpointer, self).__init__(model) + cfg = self.model.cfg + self.best_ap = 0. + self.save_dir = os.path.join(self.model.cfg.save_dir, + self.model.cfg.filename) + if hasattr(self.model.model, 'student_model'): + self.weight = self.model.model.student_model + else: + self.weight = self.model.model + + def on_epoch_end(self, status): + # Checkpointer only performed during training + mode = status['mode'] + epoch_id = status['epoch_id'] + weight = None + save_name = None + if dist.get_world_size() < 2 or dist.get_rank() == 0: + if mode == 'train': + end_epoch = self.model.cfg.epoch + if ( + epoch_id + 1 + ) % self.model.cfg.snapshot_epoch == 0 or epoch_id == end_epoch - 1: + save_name = str( + epoch_id) if epoch_id != end_epoch - 1 else "model_final" + weight = self.weight + elif mode == 'eval': + if 'save_best_model' in status and status['save_best_model']: + for metric in self.model._metrics: + map_res = metric.get_results() + if 'bbox' in map_res: + key = 'bbox' + elif 'keypoint' in map_res: + key = 'keypoint' + else: + key = 'mask' + if key not in map_res: + logger.warning("Evaluation results empty, this may be due to " \ + "training iterations being too few or not " \ + "loading the correct weights.") + return + if map_res[key][0] > self.best_ap: + self.best_ap = map_res[key][0] + save_name = 'best_model' + weight = self.weight + logger.info("Best test {} ap is {:0.3f}.".format( + key, self.best_ap)) + if weight: + save_model(weight, self.model.optimizer, self.save_dir, + save_name, epoch_id + 1) + + +class WiferFaceEval(Callback): + def __init__(self, model): + super(WiferFaceEval, self).__init__(model) + + def on_epoch_begin(self, status): + assert self.model.mode == 'eval', \ + "WiferFaceEval can only be set during evaluation" + for metric in self.model._metrics: + metric.update(self.model.model) + sys.exit() + + +class VisualDLWriter(Callback): + """ + Use VisualDL to log data or image + """ + + def __init__(self, model): + super(VisualDLWriter, self).__init__(model) + + assert six.PY3, "VisualDL requires Python >= 3.5" + try: + from visualdl import LogWriter + except Exception as e: + logger.error('visualdl not found, plaese install visualdl. ' + 'for example: `pip install visualdl`.') + raise e + self.vdl_writer = LogWriter( + model.cfg.get('vdl_log_dir', 'vdl_log_dir/scalar')) + self.vdl_loss_step = 0 + self.vdl_mAP_step = 0 + self.vdl_image_step = 0 + self.vdl_image_frame = 0 + + def on_step_end(self, status): + mode = status['mode'] + if dist.get_world_size() < 2 or dist.get_rank() == 0: + if mode == 'train': + training_staus = status['training_staus'] + for loss_name, loss_value in training_staus.get().items(): + self.vdl_writer.add_scalar(loss_name, loss_value, + self.vdl_loss_step) + self.vdl_loss_step += 1 + elif mode == 'test': + ori_image = status['original_image'] + result_image = status['result_image'] + self.vdl_writer.add_image( + "original/frame_{}".format(self.vdl_image_frame), ori_image, + self.vdl_image_step) + self.vdl_writer.add_image( + "result/frame_{}".format(self.vdl_image_frame), + result_image, self.vdl_image_step) + self.vdl_image_step += 1 + # each frame can display ten pictures at most. + if self.vdl_image_step % 10 == 0: + self.vdl_image_step = 0 + self.vdl_image_frame += 1 + + def on_epoch_end(self, status): + mode = status['mode'] + if dist.get_world_size() < 2 or dist.get_rank() == 0: + if mode == 'eval': + for metric in self.model._metrics: + for key, map_value in metric.get_results().items(): + self.vdl_writer.add_scalar("{}-mAP".format(key), + map_value[0], + self.vdl_mAP_step) + self.vdl_mAP_step += 1 + + +class SniperProposalsGenerator(Callback): + def __init__(self, model): + super(SniperProposalsGenerator, self).__init__(model) + ori_dataset = self.model.dataset + self.dataset = self._create_new_dataset(ori_dataset) + self.loader = self.model.loader + self.cfg = self.model.cfg + self.infer_model = self.model.model + + def _create_new_dataset(self, ori_dataset): + dataset = copy.deepcopy(ori_dataset) + # init anno_cropper + dataset.init_anno_cropper() + # generate infer roidbs + ori_roidbs = dataset.get_ori_roidbs() + roidbs = dataset.anno_cropper.crop_infer_anno_records(ori_roidbs) + # set new roidbs + dataset.set_roidbs(roidbs) + + return dataset + + def _eval_with_loader(self, loader): + results = [] + with paddle.no_grad(): + self.infer_model.eval() + for step_id, data in enumerate(loader): + outs = self.infer_model(data) + for key in ['im_shape', 'scale_factor', 'im_id']: + outs[key] = data[key] + for key, value in outs.items(): + if hasattr(value, 'numpy'): + outs[key] = value.numpy() + + results.append(outs) + + return results + + def on_train_end(self, status): + self.loader.dataset = self.dataset + results = self._eval_with_loader(self.loader) + results = self.dataset.anno_cropper.aggregate_chips_detections(results) + # sniper + proposals = [] + clsid2catid = {v: k for k, v in self.dataset.catid2clsid.items()} + for outs in results: + batch_res = get_infer_results(outs, clsid2catid) + start = 0 + for i, im_id in enumerate(outs['im_id']): + bbox_num = outs['bbox_num'] + end = start + bbox_num[i] + bbox_res = batch_res['bbox'][start:end] \ + if 'bbox' in batch_res else None + if bbox_res: + proposals += bbox_res + logger.info("save proposals in {}".format(self.cfg.proposals_path)) + with open(self.cfg.proposals_path, 'w') as f: + json.dump(proposals, f) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/env.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/env.py new file mode 100644 index 000000000..0a896571d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/env.py @@ -0,0 +1,50 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import random +import numpy as np + +import paddle +from paddle.distributed import fleet + +__all__ = ['init_parallel_env', 'set_random_seed', 'init_fleet_env'] + + +def init_fleet_env(find_unused_parameters=False): + strategy = fleet.DistributedStrategy() + strategy.find_unused_parameters = find_unused_parameters + fleet.init(is_collective=True, strategy=strategy) + + +def init_parallel_env(): + env = os.environ + dist = 'PADDLE_TRAINER_ID' in env and 'PADDLE_TRAINERS_NUM' in env + if dist: + trainer_id = int(env['PADDLE_TRAINER_ID']) + local_seed = (99 + trainer_id) + random.seed(local_seed) + np.random.seed(local_seed) + + paddle.distributed.init_parallel_env() + + +def set_random_seed(seed): + paddle.seed(seed) + random.seed(seed) + np.random.seed(seed) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/export_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/export_utils.py new file mode 100644 index 000000000..e1cf64638 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/export_utils.py @@ -0,0 +1,175 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import yaml +from collections import OrderedDict + +import paddle +from ppdet.data.source.category import get_categories + +from ppdet.utils.logger import setup_logger +logger = setup_logger('ppdet.engine') + +# Global dictionary +TRT_MIN_SUBGRAPH = { + 'YOLO': 3, + 'SSD': 60, + 'RCNN': 40, + 'RetinaNet': 40, + 'S2ANet': 80, + 'EfficientDet': 40, + 'Face': 3, + 'TTFNet': 60, + 'FCOS': 16, + 'SOLOv2': 60, + 'HigherHRNet': 3, + 'HRNet': 3, + 'DeepSORT': 3, + 'JDE': 10, + 'FairMOT': 5, + 'GFL': 16, + 'PicoDet': 3, + 'CenterNet': 5, +} + +KEYPOINT_ARCH = ['HigherHRNet', 'TopDownHRNet'] +MOT_ARCH = ['DeepSORT', 'JDE', 'FairMOT'] + + +def _prune_input_spec(input_spec, program, targets): + # try to prune static program to figure out pruned input spec + # so we perform following operations in static mode + paddle.enable_static() + pruned_input_spec = [{}] + program = program.clone() + program = program._prune(targets=targets) + global_block = program.global_block() + for name, spec in input_spec[0].items(): + try: + v = global_block.var(name) + pruned_input_spec[0][name] = spec + except Exception: + pass + paddle.disable_static() + return pruned_input_spec + + +def _parse_reader(reader_cfg, dataset_cfg, metric, arch, image_shape): + preprocess_list = [] + + anno_file = dataset_cfg.get_anno() + + clsid2catid, catid2name = get_categories(metric, anno_file, arch) + + label_list = [str(cat) for cat in catid2name.values()] + + fuse_normalize = reader_cfg.get('fuse_normalize', False) + sample_transforms = reader_cfg['sample_transforms'] + for st in sample_transforms[1:]: + for key, value in st.items(): + p = {'type': key} + if key == 'Resize': + if int(image_shape[1]) != -1: + value['target_size'] = image_shape[1:] + if fuse_normalize and key == 'NormalizeImage': + continue + p.update(value) + preprocess_list.append(p) + batch_transforms = reader_cfg.get('batch_transforms', None) + if batch_transforms: + for bt in batch_transforms: + for key, value in bt.items(): + # for deploy/infer, use PadStride(stride) instead PadBatch(pad_to_stride) + if key == 'PadBatch': + preprocess_list.append({ + 'type': 'PadStride', + 'stride': value['pad_to_stride'] + }) + break + + return preprocess_list, label_list + + +def _parse_tracker(tracker_cfg): + tracker_params = {} + for k, v in tracker_cfg.items(): + tracker_params.update({k: v}) + return tracker_params + + +def _dump_infer_config(config, path, image_shape, model): + arch_state = False + from ppdet.core.config.yaml_helpers import setup_orderdict + setup_orderdict() + use_dynamic_shape = True if image_shape[2] == -1 else False + infer_cfg = OrderedDict({ + 'mode': 'fluid', + 'draw_threshold': 0.5, + 'metric': config['metric'], + 'use_dynamic_shape': use_dynamic_shape + }) + infer_arch = config['architecture'] + + if infer_arch in MOT_ARCH: + if infer_arch == 'DeepSORT': + tracker_cfg = config['DeepSORTTracker'] + else: + tracker_cfg = config['JDETracker'] + infer_cfg['tracker'] = _parse_tracker(tracker_cfg) + + for arch, min_subgraph_size in TRT_MIN_SUBGRAPH.items(): + if arch in infer_arch: + infer_cfg['arch'] = arch + infer_cfg['min_subgraph_size'] = min_subgraph_size + arch_state = True + break + if not arch_state: + logger.error( + 'Architecture: {} is not supported for exporting model now.\n'. + format(infer_arch) + + 'Please set TRT_MIN_SUBGRAPH in ppdet/engine/export_utils.py') + os._exit(0) + if 'mask_head' in config[config['architecture']] and config[config[ + 'architecture']]['mask_head']: + infer_cfg['mask'] = True + label_arch = 'detection_arch' + if infer_arch in KEYPOINT_ARCH: + label_arch = 'keypoint_arch' + + if infer_arch in MOT_ARCH: + label_arch = 'mot_arch' + reader_cfg = config['TestMOTReader'] + dataset_cfg = config['TestMOTDataset'] + else: + reader_cfg = config['TestReader'] + dataset_cfg = config['TestDataset'] + + infer_cfg['Preprocess'], infer_cfg['label_list'] = _parse_reader( + reader_cfg, dataset_cfg, config['metric'], label_arch, image_shape[1:]) + + if infer_arch == 'PicoDet': + infer_cfg['NMS'] = config['PicoHead']['nms'] + # In order to speed up the prediction, the threshold of nms + # is adjusted here, which can be changed in infer_cfg.yml + config['PicoHead']['nms']["score_threshold"] = 0.3 + config['PicoHead']['nms']["nms_threshold"] = 0.5 + infer_cfg['fpn_stride'] = config['PicoHead']['fpn_stride'] + + yaml.dump(infer_cfg, open(path, 'w')) + logger.info("Export inference config file to {}".format(os.path.join(path))) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/tracker.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/tracker.py new file mode 100644 index 000000000..75602cb64 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/tracker.py @@ -0,0 +1,536 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import cv2 +import glob +import paddle +import numpy as np +from collections import defaultdict + +from ppdet.core.workspace import create +from ppdet.utils.checkpoint import load_weight, load_pretrain_weight +from ppdet.modeling.mot.utils import Detection, get_crops, scale_coords, clip_box +from ppdet.modeling.mot.utils import MOTTimer, load_det_results, write_mot_results, save_vis_results + +from ppdet.metrics import Metric, MOTMetric, KITTIMOTMetric +from ppdet.metrics import MCMOTMetric +import ppdet.utils.stats as stats + +from .callbacks import Callback, ComposeCallback + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['Tracker'] + + +class Tracker(object): + def __init__(self, cfg, mode='eval'): + self.cfg = cfg + assert mode.lower() in ['test', 'eval'], \ + "mode should be 'test' or 'eval'" + self.mode = mode.lower() + self.optimizer = None + + # build MOT data loader + self.dataset = cfg['{}MOTDataset'.format(self.mode.capitalize())] + + # build model + self.model = create(cfg.architecture) + + self.status = {} + self.start_epoch = 0 + + # initial default callbacks + self._init_callbacks() + + # initial default metrics + self._init_metrics() + self._reset_metrics() + + def _init_callbacks(self): + self._callbacks = [] + self._compose_callback = None + + def _init_metrics(self): + if self.mode in ['test']: + self._metrics = [] + return + + if self.cfg.metric == 'MOT': + self._metrics = [MOTMetric(), ] + elif self.cfg.metric == 'MCMOT': + self._metrics = [MCMOTMetric(self.cfg.num_classes), ] + elif self.cfg.metric == 'KITTI': + self._metrics = [KITTIMOTMetric(), ] + else: + logger.warning("Metric not support for metric type {}".format( + self.cfg.metric)) + self._metrics = [] + + def _reset_metrics(self): + for metric in self._metrics: + metric.reset() + + def register_callbacks(self, callbacks): + callbacks = [h for h in list(callbacks) if h is not None] + for c in callbacks: + assert isinstance(c, Callback), \ + "metrics shoule be instances of subclass of Metric" + self._callbacks.extend(callbacks) + self._compose_callback = ComposeCallback(self._callbacks) + + def register_metrics(self, metrics): + metrics = [m for m in list(metrics) if m is not None] + for m in metrics: + assert isinstance(m, Metric), \ + "metrics shoule be instances of subclass of Metric" + self._metrics.extend(metrics) + + def load_weights_jde(self, weights): + load_weight(self.model, weights, self.optimizer) + + def load_weights_sde(self, det_weights, reid_weights): + if self.model.detector: + load_weight(self.model.detector, det_weights) + load_weight(self.model.reid, reid_weights) + else: + load_weight(self.model.reid, reid_weights, self.optimizer) + + def _eval_seq_jde(self, + dataloader, + save_dir=None, + show_image=False, + frame_rate=30, + draw_threshold=0): + if save_dir: + if not os.path.exists(save_dir): os.makedirs(save_dir) + tracker = self.model.tracker + tracker.max_time_lost = int(frame_rate / 30.0 * tracker.track_buffer) + + timer = MOTTimer() + frame_id = 0 + self.status['mode'] = 'track' + self.model.eval() + results = defaultdict(list) # support single class and multi classes + + for step_id, data in enumerate(dataloader): + self.status['step_id'] = step_id + if frame_id % 40 == 0: + logger.info('Processing frame {} ({:.2f} fps)'.format( + frame_id, 1. / max(1e-5, timer.average_time))) + # forward + timer.tic() + pred_dets, pred_embs = self.model(data) + + pred_dets, pred_embs = pred_dets.numpy(), pred_embs.numpy() + online_targets_dict = self.model.tracker.update(pred_dets, + pred_embs) + online_tlwhs = defaultdict(list) + online_scores = defaultdict(list) + online_ids = defaultdict(list) + for cls_id in range(self.cfg.num_classes): + online_targets = online_targets_dict[cls_id] + for t in online_targets: + tlwh = t.tlwh + tid = t.track_id + tscore = t.score + if tlwh[2] * tlwh[3] <= tracker.min_box_area: continue + if tracker.vertical_ratio > 0 and tlwh[2] / tlwh[ + 3] > tracker.vertical_ratio: + continue + online_tlwhs[cls_id].append(tlwh) + online_ids[cls_id].append(tid) + online_scores[cls_id].append(tscore) + # save results + results[cls_id].append( + (frame_id + 1, online_tlwhs[cls_id], online_scores[cls_id], + online_ids[cls_id])) + + timer.toc() + save_vis_results(data, frame_id, online_ids, online_tlwhs, + online_scores, timer.average_time, show_image, + save_dir, self.cfg.num_classes) + frame_id += 1 + + return results, frame_id, timer.average_time, timer.calls + + def _eval_seq_sde(self, + dataloader, + save_dir=None, + show_image=False, + frame_rate=30, + seq_name='', + scaled=False, + det_file='', + draw_threshold=0): + if save_dir: + if not os.path.exists(save_dir): os.makedirs(save_dir) + use_detector = False if not self.model.detector else True + + timer = MOTTimer() + results = defaultdict(list) + frame_id = 0 + self.status['mode'] = 'track' + self.model.eval() + self.model.reid.eval() + if not use_detector: + dets_list = load_det_results(det_file, len(dataloader)) + logger.info('Finish loading detection results file {}.'.format( + det_file)) + + for step_id, data in enumerate(dataloader): + self.status['step_id'] = step_id + if frame_id % 40 == 0: + logger.info('Processing frame {} ({:.2f} fps)'.format( + frame_id, 1. / max(1e-5, timer.average_time))) + + ori_image = data['ori_image'] # [bs, H, W, 3] + ori_image_shape = data['ori_image'].shape[1:3] + # ori_image_shape: [H, W] + + input_shape = data['image'].shape[2:] + # input_shape: [h, w], before data transforms, set in model config + + im_shape = data['im_shape'][0].numpy() + # im_shape: [new_h, new_w], after data transforms + scale_factor = data['scale_factor'][0].numpy() + + empty_detections = False + # when it has no detected bboxes, will not inference reid model + # and if visualize, use original image instead + + # forward + timer.tic() + if not use_detector: + dets = dets_list[frame_id] + bbox_tlwh = np.array(dets['bbox'], dtype='float32') + if bbox_tlwh.shape[0] > 0: + # detector outputs: pred_cls_ids, pred_scores, pred_bboxes + pred_cls_ids = np.array(dets['cls_id'], dtype='float32') + pred_scores = np.array(dets['score'], dtype='float32') + pred_bboxes = np.concatenate( + (bbox_tlwh[:, 0:2], + bbox_tlwh[:, 2:4] + bbox_tlwh[:, 0:2]), + axis=1) + else: + logger.warning( + 'Frame {} has not object, try to modify score threshold.'. + format(frame_id)) + empty_detections = True + else: + outs = self.model.detector(data) + outs['bbox'] = outs['bbox'].numpy() + outs['bbox_num'] = outs['bbox_num'].numpy() + + if outs['bbox_num'] > 0 and empty_detections == False: + # detector outputs: pred_cls_ids, pred_scores, pred_bboxes + pred_cls_ids = outs['bbox'][:, 0:1] + pred_scores = outs['bbox'][:, 1:2] + if not scaled: + # Note: scaled=False only in JDE YOLOv3 or other detectors + # with LetterBoxResize and JDEBBoxPostProcess. + # + # 'scaled' means whether the coords after detector outputs + # have been scaled back to the original image, set True + # in general detector, set False in JDE YOLOv3. + pred_bboxes = scale_coords(outs['bbox'][:, 2:], + input_shape, im_shape, + scale_factor) + else: + pred_bboxes = outs['bbox'][:, 2:] + else: + logger.warning( + 'Frame {} has not detected object, try to modify score threshold.'. + format(frame_id)) + empty_detections = True + + if not empty_detections: + pred_xyxys, keep_idx = clip_box(pred_bboxes, ori_image_shape) + if len(keep_idx[0]) == 0: + logger.warning( + 'Frame {} has not detected object left after clip_box.'. + format(frame_id)) + empty_detections = True + + if empty_detections: + timer.toc() + # if visualize, use original image instead + online_ids, online_tlwhs, online_scores = None, None, None + save_vis_results(data, frame_id, online_ids, online_tlwhs, + online_scores, timer.average_time, show_image, + save_dir, self.cfg.num_classes) + frame_id += 1 + # thus will not inference reid model + continue + + pred_scores = pred_scores[keep_idx[0]] + pred_cls_ids = pred_cls_ids[keep_idx[0]] + pred_tlwhs = np.concatenate( + (pred_xyxys[:, 0:2], + pred_xyxys[:, 2:4] - pred_xyxys[:, 0:2] + 1), + axis=1) + pred_dets = np.concatenate( + (pred_tlwhs, pred_scores, pred_cls_ids), axis=1) + + tracker = self.model.tracker + crops = get_crops( + pred_xyxys, + ori_image, + w=tracker.input_size[0], + h=tracker.input_size[1]) + crops = paddle.to_tensor(crops) + + data.update({'crops': crops}) + pred_embs = self.model(data).numpy() + + tracker.predict() + online_targets = tracker.update(pred_dets, pred_embs) + + online_tlwhs, online_scores, online_ids = [], [], [] + for t in online_targets: + if not t.is_confirmed() or t.time_since_update > 1: + continue + tlwh = t.to_tlwh() + tscore = t.score + tid = t.track_id + if tscore < draw_threshold: continue + if tlwh[2] * tlwh[3] <= tracker.min_box_area: continue + if tracker.vertical_ratio > 0 and tlwh[2] / tlwh[ + 3] > tracker.vertical_ratio: + continue + online_tlwhs.append(tlwh) + online_scores.append(tscore) + online_ids.append(tid) + timer.toc() + + # save results + results[0].append( + (frame_id + 1, online_tlwhs, online_scores, online_ids)) + save_vis_results(data, frame_id, online_ids, online_tlwhs, + online_scores, timer.average_time, show_image, + save_dir, self.cfg.num_classes) + frame_id += 1 + + return results, frame_id, timer.average_time, timer.calls + + def mot_evaluate(self, + data_root, + seqs, + output_dir, + data_type='mot', + model_type='JDE', + save_images=False, + save_videos=False, + show_image=False, + scaled=False, + det_results_dir=''): + if not os.path.exists(output_dir): os.makedirs(output_dir) + result_root = os.path.join(output_dir, 'mot_results') + if not os.path.exists(result_root): os.makedirs(result_root) + assert data_type in ['mot', 'mcmot', 'kitti'], \ + "data_type should be 'mot', 'mcmot' or 'kitti'" + assert model_type in ['JDE', 'DeepSORT', 'FairMOT'], \ + "model_type should be 'JDE', 'DeepSORT' or 'FairMOT'" + + # run tracking + n_frame = 0 + timer_avgs, timer_calls = [], [] + for seq in seqs: + infer_dir = os.path.join(data_root, seq) + if not os.path.exists(infer_dir) or not os.path.isdir(infer_dir): + logger.warning("Seq {} error, {} has no images.".format( + seq, infer_dir)) + continue + if os.path.exists(os.path.join(infer_dir, 'img1')): + infer_dir = os.path.join(infer_dir, 'img1') + + frame_rate = 30 + seqinfo = os.path.join(data_root, seq, 'seqinfo.ini') + if os.path.exists(seqinfo): + meta_info = open(seqinfo).read() + frame_rate = int(meta_info[meta_info.find('frameRate') + 10: + meta_info.find('\nseqLength')]) + + save_dir = os.path.join(output_dir, 'mot_outputs', + seq) if save_images or save_videos else None + logger.info('start seq: {}'.format(seq)) + + self.dataset.set_images(self.get_infer_images(infer_dir)) + dataloader = create('EvalMOTReader')(self.dataset, 0) + + result_filename = os.path.join(result_root, '{}.txt'.format(seq)) + + with paddle.no_grad(): + if model_type in ['JDE', 'FairMOT']: + results, nf, ta, tc = self._eval_seq_jde( + dataloader, + save_dir=save_dir, + show_image=show_image, + frame_rate=frame_rate) + elif model_type in ['DeepSORT']: + results, nf, ta, tc = self._eval_seq_sde( + dataloader, + save_dir=save_dir, + show_image=show_image, + frame_rate=frame_rate, + seq_name=seq, + scaled=scaled, + det_file=os.path.join(det_results_dir, + '{}.txt'.format(seq))) + else: + raise ValueError(model_type) + + write_mot_results(result_filename, results, data_type, + self.cfg.num_classes) + n_frame += nf + timer_avgs.append(ta) + timer_calls.append(tc) + + if save_videos: + output_video_path = os.path.join(save_dir, '..', + '{}_vis.mp4'.format(seq)) + cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg {}'.format( + save_dir, output_video_path) + os.system(cmd_str) + logger.info('Save video in {}.'.format(output_video_path)) + + logger.info('Evaluate seq: {}'.format(seq)) + # update metrics + for metric in self._metrics: + metric.update(data_root, seq, data_type, result_root, + result_filename) + + timer_avgs = np.asarray(timer_avgs) + timer_calls = np.asarray(timer_calls) + all_time = np.dot(timer_avgs, timer_calls) + avg_time = all_time / np.sum(timer_calls) + logger.info('Time elapsed: {:.2f} seconds, FPS: {:.2f}'.format( + all_time, 1.0 / avg_time)) + + # accumulate metric to log out + for metric in self._metrics: + metric.accumulate() + metric.log() + # reset metric states for metric may performed multiple times + self._reset_metrics() + + def get_infer_images(self, infer_dir): + assert infer_dir is None or os.path.isdir(infer_dir), \ + "{} is not a directory".format(infer_dir) + images = set() + assert os.path.isdir(infer_dir), \ + "infer_dir {} is not a directory".format(infer_dir) + exts = ['jpg', 'jpeg', 'png', 'bmp'] + exts += [ext.upper() for ext in exts] + for ext in exts: + images.update(glob.glob('{}/*.{}'.format(infer_dir, ext))) + images = list(images) + images.sort() + assert len(images) > 0, "no image found in {}".format(infer_dir) + logger.info("Found {} inference images in total.".format(len(images))) + return images + + def mot_predict_seq(self, + video_file, + frame_rate, + image_dir, + output_dir, + data_type='mot', + model_type='JDE', + save_images=False, + save_videos=True, + show_image=False, + scaled=False, + det_results_dir='', + draw_threshold=0.5): + assert video_file is not None or image_dir is not None, \ + "--video_file or --image_dir should be set." + assert video_file is None or os.path.isfile(video_file), \ + "{} is not a file".format(video_file) + assert image_dir is None or os.path.isdir(image_dir), \ + "{} is not a directory".format(image_dir) + + if not os.path.exists(output_dir): os.makedirs(output_dir) + result_root = os.path.join(output_dir, 'mot_results') + if not os.path.exists(result_root): os.makedirs(result_root) + assert data_type in ['mot', 'mcmot', 'kitti'], \ + "data_type should be 'mot', 'mcmot' or 'kitti'" + assert model_type in ['JDE', 'DeepSORT', 'FairMOT'], \ + "model_type should be 'JDE', 'DeepSORT' or 'FairMOT'" + + # run tracking + if video_file: + seq = video_file.split('/')[-1].split('.')[0] + self.dataset.set_video(video_file, frame_rate) + logger.info('Starting tracking video {}'.format(video_file)) + elif image_dir: + seq = image_dir.split('/')[-1].split('.')[0] + if os.path.exists(os.path.join(image_dir, 'img1')): + image_dir = os.path.join(image_dir, 'img1') + images = [ + '{}/{}'.format(image_dir, x) for x in os.listdir(image_dir) + ] + images.sort() + self.dataset.set_images(images) + logger.info('Starting tracking folder {}, found {} images'.format( + image_dir, len(images))) + else: + raise ValueError('--video_file or --image_dir should be set.') + + save_dir = os.path.join(output_dir, 'mot_outputs', + seq) if save_images or save_videos else None + + dataloader = create('TestMOTReader')(self.dataset, 0) + result_filename = os.path.join(result_root, '{}.txt'.format(seq)) + if frame_rate == -1: + frame_rate = self.dataset.frame_rate + + with paddle.no_grad(): + if model_type in ['JDE', 'FairMOT']: + results, nf, ta, tc = self._eval_seq_jde( + dataloader, + save_dir=save_dir, + show_image=show_image, + frame_rate=frame_rate, + draw_threshold=draw_threshold) + elif model_type in ['DeepSORT']: + results, nf, ta, tc = self._eval_seq_sde( + dataloader, + save_dir=save_dir, + show_image=show_image, + frame_rate=frame_rate, + seq_name=seq, + scaled=scaled, + det_file=os.path.join(det_results_dir, + '{}.txt'.format(seq)), + draw_threshold=draw_threshold) + else: + raise ValueError(model_type) + + if save_videos: + output_video_path = os.path.join(save_dir, '..', + '{}_vis.mp4'.format(seq)) + cmd_str = 'ffmpeg -f image2 -i {}/%05d.jpg {}'.format( + save_dir, output_video_path) + os.system(cmd_str) + logger.info('Save video in {}'.format(output_video_path)) + + write_mot_results(result_filename, results, data_type, + self.cfg.num_classes) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/trainer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/trainer.py new file mode 100644 index 000000000..dc739ff62 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/engine/trainer.py @@ -0,0 +1,715 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys +import copy +import time + +import numpy as np +from PIL import Image, ImageOps + +import paddle +import paddle.distributed as dist +from paddle.distributed import fleet +from paddle import amp +from paddle.static import InputSpec +from ppdet.optimizer import ModelEMA + +from ppdet.core.workspace import create +from ppdet.utils.checkpoint import load_weight, load_pretrain_weight +from ppdet.utils.visualizer import visualize_results, save_result +from ppdet.metrics import Metric, COCOMetric, VOCMetric, WiderFaceMetric, get_infer_results, KeyPointTopDownCOCOEval, KeyPointTopDownMPIIEval +from ppdet.metrics import RBoxMetric, JDEDetMetric, SNIPERCOCOMetric +from ppdet.data.source.sniper_coco import SniperCOCODataSet +from ppdet.data.source.category import get_categories +import ppdet.utils.stats as stats +from ppdet.utils import profiler + +from .callbacks import Callback, ComposeCallback, LogPrinter, Checkpointer, WiferFaceEval, VisualDLWriter, SniperProposalsGenerator +from .export_utils import _dump_infer_config, _prune_input_spec + +from ppdet.utils.logger import setup_logger +logger = setup_logger('ppdet.engine') + +__all__ = ['Trainer'] + +MOT_ARCH = ['DeepSORT', 'JDE', 'FairMOT'] + + +class Trainer(object): + def __init__(self, cfg, mode='train'): + self.cfg = cfg + assert mode.lower() in ['train', 'eval', 'test'], \ + "mode should be 'train', 'eval' or 'test'" + self.mode = mode.lower() + self.optimizer = None + self.is_loaded_weights = False + + # build data loader + if cfg.architecture in MOT_ARCH and self.mode in ['eval', 'test']: + self.dataset = cfg['{}MOTDataset'.format(self.mode.capitalize())] + else: + self.dataset = cfg['{}Dataset'.format(self.mode.capitalize())] + + if cfg.architecture == 'DeepSORT' and self.mode == 'train': + logger.error('DeepSORT has no need of training on mot dataset.') + sys.exit(1) + + if self.mode == 'train': + self.loader = create('{}Reader'.format(self.mode.capitalize()))( + self.dataset, cfg.worker_num) + + if cfg.architecture == 'JDE' and self.mode == 'train': + cfg['JDEEmbeddingHead'][ + 'num_identities'] = self.dataset.num_identities_dict[0] + # JDE only support single class MOT now. + + if cfg.architecture == 'FairMOT' and self.mode == 'train': + cfg['FairMOTEmbeddingHead'][ + 'num_identities_dict'] = self.dataset.num_identities_dict + # FairMOT support single class and multi-class MOT now. + + # build model + if 'model' not in self.cfg: + self.model = create(cfg.architecture) + else: + self.model = self.cfg.model + self.is_loaded_weights = True + + #normalize params for deploy + self.model.load_meanstd(cfg['TestReader']['sample_transforms']) + + self.use_ema = ('use_ema' in cfg and cfg['use_ema']) + if self.use_ema: + ema_decay = self.cfg.get('ema_decay', 0.9998) + cycle_epoch = self.cfg.get('cycle_epoch', -1) + self.ema = ModelEMA( + self.model, + decay=ema_decay, + use_thres_step=True, + cycle_epoch=cycle_epoch) + + # EvalDataset build with BatchSampler to evaluate in single device + # TODO: multi-device evaluate + if self.mode == 'eval': + self._eval_batch_sampler = paddle.io.BatchSampler( + self.dataset, batch_size=self.cfg.EvalReader['batch_size']) + self.loader = create('{}Reader'.format(self.mode.capitalize()))( + self.dataset, cfg.worker_num, self._eval_batch_sampler) + # TestDataset build after user set images, skip loader creation here + + # build optimizer in train mode + if self.mode == 'train': + steps_per_epoch = len(self.loader) + self.lr = create('LearningRate')(steps_per_epoch) + self.optimizer = create('OptimizerBuilder')(self.lr, self.model) + + if self.cfg.get('unstructured_prune'): + self.pruner = create('UnstructuredPruner')(self.model, + steps_per_epoch) + + self._nranks = dist.get_world_size() + self._local_rank = dist.get_rank() + + self.status = {} + + self.start_epoch = 0 + self.end_epoch = 0 if 'epoch' not in cfg else cfg.epoch + + # initial default callbacks + self._init_callbacks() + + # initial default metrics + self._init_metrics() + self._reset_metrics() + + def _init_callbacks(self): + if self.mode == 'train': + self._callbacks = [LogPrinter(self), Checkpointer(self)] + if self.cfg.get('use_vdl', False): + self._callbacks.append(VisualDLWriter(self)) + if self.cfg.get('save_proposals', False): + self._callbacks.append(SniperProposalsGenerator(self)) + self._compose_callback = ComposeCallback(self._callbacks) + elif self.mode == 'eval': + self._callbacks = [LogPrinter(self)] + if self.cfg.metric == 'WiderFace': + self._callbacks.append(WiferFaceEval(self)) + self._compose_callback = ComposeCallback(self._callbacks) + elif self.mode == 'test' and self.cfg.get('use_vdl', False): + self._callbacks = [VisualDLWriter(self)] + self._compose_callback = ComposeCallback(self._callbacks) + else: + self._callbacks = [] + self._compose_callback = None + + def _init_metrics(self, validate=False): + if self.mode == 'test' or (self.mode == 'train' and not validate): + self._metrics = [] + return + classwise = self.cfg['classwise'] if 'classwise' in self.cfg else False + if self.cfg.metric == 'COCO' or self.cfg.metric == "SNIPERCOCO": + # TODO: bias should be unified + bias = self.cfg['bias'] if 'bias' in self.cfg else 0 + output_eval = self.cfg['output_eval'] \ + if 'output_eval' in self.cfg else None + save_prediction_only = self.cfg.get('save_prediction_only', False) + + # pass clsid2catid info to metric instance to avoid multiple loading + # annotation file + clsid2catid = {v: k for k, v in self.dataset.catid2clsid.items()} \ + if self.mode == 'eval' else None + + # when do validation in train, annotation file should be get from + # EvalReader instead of self.dataset(which is TrainReader) + anno_file = self.dataset.get_anno() + dataset = self.dataset + if self.mode == 'train' and validate: + eval_dataset = self.cfg['EvalDataset'] + eval_dataset.check_or_download_dataset() + anno_file = eval_dataset.get_anno() + dataset = eval_dataset + + IouType = self.cfg['IouType'] if 'IouType' in self.cfg else 'bbox' + if self.cfg.metric == "COCO": + self._metrics = [ + COCOMetric( + anno_file=anno_file, + clsid2catid=clsid2catid, + classwise=classwise, + output_eval=output_eval, + bias=bias, + IouType=IouType, + save_prediction_only=save_prediction_only) + ] + elif self.cfg.metric == "SNIPERCOCO": # sniper + self._metrics = [ + SNIPERCOCOMetric( + anno_file=anno_file, + dataset=dataset, + clsid2catid=clsid2catid, + classwise=classwise, + output_eval=output_eval, + bias=bias, + IouType=IouType, + save_prediction_only=save_prediction_only) + ] + elif self.cfg.metric == 'RBOX': + # TODO: bias should be unified + bias = self.cfg['bias'] if 'bias' in self.cfg else 0 + output_eval = self.cfg['output_eval'] \ + if 'output_eval' in self.cfg else None + save_prediction_only = self.cfg.get('save_prediction_only', False) + + # pass clsid2catid info to metric instance to avoid multiple loading + # annotation file + clsid2catid = {v: k for k, v in self.dataset.catid2clsid.items()} \ + if self.mode == 'eval' else None + + # when do validation in train, annotation file should be get from + # EvalReader instead of self.dataset(which is TrainReader) + anno_file = self.dataset.get_anno() + if self.mode == 'train' and validate: + eval_dataset = self.cfg['EvalDataset'] + eval_dataset.check_or_download_dataset() + anno_file = eval_dataset.get_anno() + + self._metrics = [ + RBoxMetric( + anno_file=anno_file, + clsid2catid=clsid2catid, + classwise=classwise, + output_eval=output_eval, + bias=bias, + save_prediction_only=save_prediction_only) + ] + elif self.cfg.metric == 'VOC': + self._metrics = [ + VOCMetric( + label_list=self.dataset.get_label_list(), + class_num=self.cfg.num_classes, + map_type=self.cfg.map_type, + classwise=classwise) + ] + elif self.cfg.metric == 'WiderFace': + multi_scale = self.cfg.multi_scale_eval if 'multi_scale_eval' in self.cfg else True + self._metrics = [ + WiderFaceMetric( + image_dir=os.path.join(self.dataset.dataset_dir, + self.dataset.image_dir), + anno_file=self.dataset.get_anno(), + multi_scale=multi_scale) + ] + elif self.cfg.metric == 'KeyPointTopDownCOCOEval': + eval_dataset = self.cfg['EvalDataset'] + eval_dataset.check_or_download_dataset() + anno_file = eval_dataset.get_anno() + save_prediction_only = self.cfg.get('save_prediction_only', False) + self._metrics = [ + KeyPointTopDownCOCOEval( + anno_file, + len(eval_dataset), + self.cfg.num_joints, + self.cfg.save_dir, + save_prediction_only=save_prediction_only) + ] + elif self.cfg.metric == 'KeyPointTopDownMPIIEval': + eval_dataset = self.cfg['EvalDataset'] + eval_dataset.check_or_download_dataset() + anno_file = eval_dataset.get_anno() + save_prediction_only = self.cfg.get('save_prediction_only', False) + self._metrics = [ + KeyPointTopDownMPIIEval( + anno_file, + len(eval_dataset), + self.cfg.num_joints, + self.cfg.save_dir, + save_prediction_only=save_prediction_only) + ] + elif self.cfg.metric == 'MOTDet': + self._metrics = [JDEDetMetric(), ] + else: + logger.warning("Metric not support for metric type {}".format( + self.cfg.metric)) + self._metrics = [] + + def _reset_metrics(self): + for metric in self._metrics: + metric.reset() + + def register_callbacks(self, callbacks): + callbacks = [c for c in list(callbacks) if c is not None] + for c in callbacks: + assert isinstance(c, Callback), \ + "metrics shoule be instances of subclass of Metric" + self._callbacks.extend(callbacks) + self._compose_callback = ComposeCallback(self._callbacks) + + def register_metrics(self, metrics): + metrics = [m for m in list(metrics) if m is not None] + for m in metrics: + assert isinstance(m, Metric), \ + "metrics shoule be instances of subclass of Metric" + self._metrics.extend(metrics) + + def load_weights(self, weights): + if self.is_loaded_weights: + return + self.start_epoch = 0 + load_pretrain_weight(self.model, weights) + logger.debug("Load weights {} to start training".format(weights)) + + def load_weights_sde(self, det_weights, reid_weights): + if self.model.detector: + load_weight(self.model.detector, det_weights) + load_weight(self.model.reid, reid_weights) + else: + load_weight(self.model.reid, reid_weights) + + def resume_weights(self, weights): + # support Distill resume weights + if hasattr(self.model, 'student_model'): + self.start_epoch = load_weight(self.model.student_model, weights, + self.optimizer) + else: + self.start_epoch = load_weight(self.model, weights, self.optimizer) + logger.debug("Resume weights of epoch {}".format(self.start_epoch)) + + def train(self, validate=False): + assert self.mode == 'train', "Model not in 'train' mode" + Init_mark = False + + model = self.model + if self.cfg.get('fleet', False): + model = fleet.distributed_model(model) + self.optimizer = fleet.distributed_optimizer(self.optimizer) + elif self._nranks > 1: + find_unused_parameters = self.cfg[ + 'find_unused_parameters'] if 'find_unused_parameters' in self.cfg else False + model = paddle.DataParallel( + self.model, find_unused_parameters=find_unused_parameters) + + # initial fp16 + if self.cfg.get('fp16', False): + scaler = amp.GradScaler( + enable=self.cfg.use_gpu, init_loss_scaling=1024) + + self.status.update({ + 'epoch_id': self.start_epoch, + 'step_id': 0, + 'steps_per_epoch': len(self.loader) + }) + + self.status['batch_time'] = stats.SmoothedValue( + self.cfg.log_iter, fmt='{avg:.4f}') + self.status['data_time'] = stats.SmoothedValue( + self.cfg.log_iter, fmt='{avg:.4f}') + self.status['training_staus'] = stats.TrainingStats(self.cfg.log_iter) + + if self.cfg.get('print_flops', False): + self._flops(self.loader) + profiler_options = self.cfg.get('profiler_options', None) + + self._compose_callback.on_train_begin(self.status) + + for epoch_id in range(self.start_epoch, self.cfg.epoch): + self.status['mode'] = 'train' + self.status['epoch_id'] = epoch_id + self._compose_callback.on_epoch_begin(self.status) + self.loader.dataset.set_epoch(epoch_id) + model.train() + iter_tic = time.time() + for step_id, data in enumerate(self.loader): + self.status['data_time'].update(time.time() - iter_tic) + self.status['step_id'] = step_id + profiler.add_profiler_step(profiler_options) + self._compose_callback.on_step_begin(self.status) + data['epoch_id'] = epoch_id + + if self.cfg.get('fp16', False): + with amp.auto_cast(enable=self.cfg.use_gpu): + # model forward + outputs = model(data) + loss = outputs['loss'] + + # model backward + scaled_loss = scaler.scale(loss) + scaled_loss.backward() + # in dygraph mode, optimizer.minimize is equal to optimizer.step + scaler.minimize(self.optimizer, scaled_loss) + else: + # model forward + outputs = model(data) + loss = outputs['loss'] + # model backward + loss.backward() + self.optimizer.step() + curr_lr = self.optimizer.get_lr() + self.lr.step() + if self.cfg.get('unstructured_prune'): + self.pruner.step() + self.optimizer.clear_grad() + self.status['learning_rate'] = curr_lr + + if self._nranks < 2 or self._local_rank == 0: + self.status['training_staus'].update(outputs) + + self.status['batch_time'].update(time.time() - iter_tic) + self._compose_callback.on_step_end(self.status) + if self.use_ema: + self.ema.update(self.model) + iter_tic = time.time() + + # apply ema weight on model + if self.use_ema: + weight = copy.deepcopy(self.model.state_dict()) + self.model.set_dict(self.ema.apply()) + if self.cfg.get('unstructured_prune'): + self.pruner.update_params() + + self._compose_callback.on_epoch_end(self.status) + + if validate and (self._nranks < 2 or self._local_rank == 0) \ + and ((epoch_id + 1) % self.cfg.snapshot_epoch == 0 \ + or epoch_id == self.end_epoch - 1): + if not hasattr(self, '_eval_loader'): + # build evaluation dataset and loader + self._eval_dataset = self.cfg.EvalDataset + self._eval_batch_sampler = \ + paddle.io.BatchSampler( + self._eval_dataset, + batch_size=self.cfg.EvalReader['batch_size']) + self._eval_loader = create('EvalReader')( + self._eval_dataset, + self.cfg.worker_num, + batch_sampler=self._eval_batch_sampler) + # if validation in training is enabled, metrics should be re-init + # Init_mark makes sure this code will only execute once + if validate and Init_mark == False: + Init_mark = True + self._init_metrics(validate=validate) + self._reset_metrics() + with paddle.no_grad(): + self.status['save_best_model'] = True + self._eval_with_loader(self._eval_loader) + + # restore origin weight on model + if self.use_ema: + self.model.set_dict(weight) + + self._compose_callback.on_train_end(self.status) + + def _eval_with_loader(self, loader): + sample_num = 0 + tic = time.time() + self._compose_callback.on_epoch_begin(self.status) + self.status['mode'] = 'eval' + self.model.eval() + if self.cfg.get('print_flops', False): + self._flops(loader) + for step_id, data in enumerate(loader): + self.status['step_id'] = step_id + self._compose_callback.on_step_begin(self.status) + # forward + outs = self.model(data) + + # update metrics + for metric in self._metrics: + metric.update(data, outs) + + sample_num += data['im_id'].numpy().shape[0] + self._compose_callback.on_step_end(self.status) + + self.status['sample_num'] = sample_num + self.status['cost_time'] = time.time() - tic + + # accumulate metric to log out + for metric in self._metrics: + metric.accumulate() + metric.log() + self._compose_callback.on_epoch_end(self.status) + # reset metric states for metric may performed multiple times + self._reset_metrics() + + def evaluate(self): + with paddle.no_grad(): + self._eval_with_loader(self.loader) + + def predict(self, + images, + draw_threshold=0.5, + output_dir='output', + save_txt=False): + self.dataset.set_images(images) + loader = create('TestReader')(self.dataset, 0) + + imid2path = self.dataset.get_imid2path() + + anno_file = self.dataset.get_anno() + clsid2catid, catid2name = get_categories( + self.cfg.metric, anno_file=anno_file) + + # Run Infer + self.status['mode'] = 'test' + self.model.eval() + if self.cfg.get('print_flops', False): + self._flops(loader) + results = [] + for step_id, data in enumerate(loader): + self.status['step_id'] = step_id + # forward + outs = self.model(data) + + for key in ['im_shape', 'scale_factor', 'im_id']: + outs[key] = data[key] + for key, value in outs.items(): + if hasattr(value, 'numpy'): + outs[key] = value.numpy() + results.append(outs) + # sniper + if type(self.dataset) == SniperCOCODataSet: + results = self.dataset.anno_cropper.aggregate_chips_detections( + results) + + for outs in results: + batch_res = get_infer_results(outs, clsid2catid) + bbox_num = outs['bbox_num'] + + start = 0 + for i, im_id in enumerate(outs['im_id']): + image_path = imid2path[int(im_id)] + image = Image.open(image_path).convert('RGB') + image = ImageOps.exif_transpose(image) + self.status['original_image'] = np.array(image.copy()) + + end = start + bbox_num[i] + bbox_res = batch_res['bbox'][start:end] \ + if 'bbox' in batch_res else None + mask_res = batch_res['mask'][start:end] \ + if 'mask' in batch_res else None + segm_res = batch_res['segm'][start:end] \ + if 'segm' in batch_res else None + keypoint_res = batch_res['keypoint'][start:end] \ + if 'keypoint' in batch_res else None + image = visualize_results( + image, bbox_res, mask_res, segm_res, keypoint_res, + int(im_id), catid2name, draw_threshold) + self.status['result_image'] = np.array(image.copy()) + if self._compose_callback: + self._compose_callback.on_step_end(self.status) + # save image with detection + save_name = self._get_save_image_name(output_dir, image_path) + logger.info("Detection bbox results save in {}".format( + save_name)) + image.save(save_name, quality=95) + if save_txt: + save_path = os.path.splitext(save_name)[0] + '.txt' + results = {} + results["im_id"] = im_id + if bbox_res: + results["bbox_res"] = bbox_res + if keypoint_res: + results["keypoint_res"] = keypoint_res + save_result(save_path, results, catid2name, draw_threshold) + start = end + + def _get_save_image_name(self, output_dir, image_path): + """ + Get save image name from source image path. + """ + if not os.path.exists(output_dir): + os.makedirs(output_dir) + image_name = os.path.split(image_path)[-1] + name, ext = os.path.splitext(image_name) + return os.path.join(output_dir, "{}".format(name)) + ext + + def _get_infer_cfg_and_input_spec(self, save_dir, prune_input=True): + image_shape = None + im_shape = [None, 2] + scale_factor = [None, 2] + if self.cfg.architecture in MOT_ARCH: + test_reader_name = 'TestMOTReader' + else: + test_reader_name = 'TestReader' + if 'inputs_def' in self.cfg[test_reader_name]: + inputs_def = self.cfg[test_reader_name]['inputs_def'] + image_shape = inputs_def.get('image_shape', None) + # set image_shape=[None, 3, -1, -1] as default + if image_shape is None: + image_shape = [None, 3, -1, -1] + + if len(image_shape) == 3: + image_shape = [None] + image_shape + else: + im_shape = [image_shape[0], 2] + scale_factor = [image_shape[0], 2] + + if hasattr(self.model, 'deploy'): + self.model.deploy = True + if hasattr(self.model, 'fuse_norm'): + self.model.fuse_norm = self.cfg['TestReader'].get('fuse_normalize', + False) + + # Save infer cfg + _dump_infer_config(self.cfg, + os.path.join(save_dir, 'infer_cfg.yml'), image_shape, + self.model) + + input_spec = [{ + "image": InputSpec( + shape=image_shape, name='image'), + "im_shape": InputSpec( + shape=im_shape, name='im_shape'), + "scale_factor": InputSpec( + shape=scale_factor, name='scale_factor') + }] + if self.cfg.architecture == 'DeepSORT': + input_spec[0].update({ + "crops": InputSpec( + shape=[None, 3, 192, 64], name='crops') + }) + if prune_input: + static_model = paddle.jit.to_static( + self.model, input_spec=input_spec) + # NOTE: dy2st do not pruned program, but jit.save will prune program + # input spec, prune input spec here and save with pruned input spec + pruned_input_spec = _prune_input_spec( + input_spec, static_model.forward.main_program, + static_model.forward.outputs) + else: + static_model = None + pruned_input_spec = input_spec + + # TODO: Hard code, delete it when support prune input_spec. + if self.cfg.architecture == 'PicoDet': + pruned_input_spec = [{ + "image": InputSpec( + shape=image_shape, name='image') + }] + + return static_model, pruned_input_spec + + def export(self, output_dir='output_inference'): + self.model.eval() + model_name = os.path.splitext(os.path.split(self.cfg.filename)[-1])[0] + save_dir = os.path.join(output_dir, model_name) + if not os.path.exists(save_dir): + os.makedirs(save_dir) + + static_model, pruned_input_spec = self._get_infer_cfg_and_input_spec( + save_dir) + + # dy2st and save model + if 'slim' not in self.cfg or self.cfg['slim_type'] != 'QAT': + paddle.jit.save( + static_model, + os.path.join(save_dir, 'model'), + input_spec=pruned_input_spec) + else: + self.cfg.slim.save_quantized_model( + self.model, + os.path.join(save_dir, 'model'), + input_spec=pruned_input_spec) + logger.info("Export model and saved in {}".format(save_dir)) + + def post_quant(self, output_dir='output_inference'): + model_name = os.path.splitext(os.path.split(self.cfg.filename)[-1])[0] + save_dir = os.path.join(output_dir, model_name) + if not os.path.exists(save_dir): + os.makedirs(save_dir) + + for idx, data in enumerate(self.loader): + self.model(data) + if idx == int(self.cfg.get('quant_batch_num', 10)): + break + + # TODO: support prune input_spec + _, pruned_input_spec = self._get_infer_cfg_and_input_spec( + save_dir, prune_input=False) + + self.cfg.slim.save_quantized_model( + self.model, + os.path.join(save_dir, 'model'), + input_spec=pruned_input_spec) + logger.info("Export Post-Quant model and saved in {}".format(save_dir)) + + def _flops(self, loader): + self.model.eval() + try: + import paddleslim + except Exception as e: + logger.warning( + 'Unable to calculate flops, please install paddleslim, for example: `pip install paddleslim`' + ) + return + + from paddleslim.analysis import dygraph_flops as flops + input_data = None + for data in loader: + input_data = data + break + + input_spec = [{ + "image": input_data['image'][0].unsqueeze(0), + "im_shape": input_data['im_shape'][0].unsqueeze(0), + "scale_factor": input_data['scale_factor'][0].unsqueeze(0) + }] + flops = flops(self.model, input_spec) / (1000**3) + logger.info(" Model FLOPs : {:.6f}G. (image shape is {})".format( + flops, input_data['image'][0].unsqueeze(0).shape)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/README.md b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/README.md new file mode 100644 index 000000000..7ada0acf7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/README.md @@ -0,0 +1,38 @@ +# 鑷畾涔塐P缂栬瘧 +鏃嬭浆妗咺OU璁＄畻OP鏄弬鑰僛鑷畾涔夊閮ㄧ畻瀛怾(https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/07_new_op/new_custom_op.html) 銆 + +## 1. 鐜渚濊禆 +- Paddle >= 2.0.1 +- gcc 8.2 + +## 2. 瀹夎 +``` +python3.7 setup.py install +``` + +鎸夌収濡備笅鏂瑰紡浣跨敤 +``` +# 寮曞叆鑷畾涔塷p +from rbox_iou_ops import rbox_iou + +paddle.set_device('gpu:0') +paddle.disable_static() + +rbox1 = np.random.rand(13000, 5) +rbox2 = np.random.rand(7, 5) + +pd_rbox1 = paddle.to_tensor(rbox1) +pd_rbox2 = paddle.to_tensor(rbox2) + +iou = rbox_iou(pd_rbox1, pd_rbox2) +print('iou', iou) +``` + +## 3. 鍗曞厓娴嬭瘯 +鍗曞厓娴嬭瘯`test.py`鏂囦欢涓紝閫氳繃瀵规瘮python瀹炵幇鐨勭粨鏋滃拰娴嬭瘯鑷畾涔塷p缁撴灉銆 + +鐢变簬python璁＄畻缁嗚妭涓巆pp璁＄畻缁嗚妭鐣ユ湁鍖哄埆锛岃宸尯闂磋缃负0.02銆 +``` +python3.7 test.py +``` +鎻愮ず`rbox_iou OP compute right!`璇存槑OP娴嬭瘯閫氳繃銆 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cc new file mode 100644 index 000000000..6031953d2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cc @@ -0,0 +1,97 @@ +// Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// The code is based on https://github.com/csuhan/s2anet/blob/master/mmdet/ops/box_iou_rotated + +#include "rbox_iou_op.h" +#include "paddle/extension.h" + + +template +void rbox_iou_cpu_kernel( + const int rbox1_num, + const int rbox2_num, + const T* rbox1_data_ptr, + const T* rbox2_data_ptr, + T* output_data_ptr) { + + int i, j; + for (i = 0; i < rbox1_num; i++) { + for (j = 0; j < rbox2_num; j++) { + int offset = i * rbox2_num + j; + output_data_ptr[offset] = rbox_iou_single(rbox1_data_ptr + i * 5, rbox2_data_ptr + j * 5); + } + } +} + + +#define CHECK_INPUT_CPU(x) PD_CHECK(x.place() == paddle::PlaceType::kCPU, #x " must be a CPU Tensor.") + +std::vector RboxIouCPUForward(const paddle::Tensor& rbox1, const paddle::Tensor& rbox2) { + CHECK_INPUT_CPU(rbox1); + CHECK_INPUT_CPU(rbox2); + + auto rbox1_num = rbox1.shape()[0]; + auto rbox2_num = rbox2.shape()[0]; + + auto output = paddle::Tensor(paddle::PlaceType::kCPU, {rbox1_num, rbox2_num}); + + PD_DISPATCH_FLOATING_TYPES( + rbox1.type(), + "rbox_iou_cpu_kernel", + ([&] { + rbox_iou_cpu_kernel( + rbox1_num, + rbox2_num, + rbox1.data(), + rbox2.data(), + output.mutable_data()); + })); + + return {output}; +} + + +#ifdef PADDLE_WITH_CUDA +std::vector RboxIouCUDAForward(const paddle::Tensor& rbox1, const paddle::Tensor& rbox2); +#endif + + +#define CHECK_INPUT_SAME(x1, x2) PD_CHECK(x1.place() == x2.place(), "input must be smae pacle.") + +std::vector RboxIouForward(const paddle::Tensor& rbox1, const paddle::Tensor& rbox2) { + CHECK_INPUT_SAME(rbox1, rbox2); + if (rbox1.place() == paddle::PlaceType::kCPU) { + return RboxIouCPUForward(rbox1, rbox2); +#ifdef PADDLE_WITH_CUDA + } else if (rbox1.place() == paddle::PlaceType::kGPU) { + return RboxIouCUDAForward(rbox1, rbox2); +#endif + } +} + +std::vector> InferShape(std::vector rbox1_shape, std::vector rbox2_shape) { + return {{rbox1_shape[0], rbox2_shape[0]}}; +} + +std::vector InferDtype(paddle::DataType t1, paddle::DataType t2) { + return {t1}; +} + +PD_BUILD_OP(rbox_iou) + .Inputs({"RBOX1", "RBOX2"}) + .Outputs({"Output"}) + .SetKernelFn(PD_KERNEL(RboxIouForward)) + .SetInferShapeFn(PD_INFER_SHAPE(InferShape)) + .SetInferDtypeFn(PD_INFER_DTYPE(InferDtype)); diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cu b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cu new file mode 100644 index 000000000..8ec43e54b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.cu @@ -0,0 +1,120 @@ +// Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// The code is based on https://github.com/csuhan/s2anet/blob/master/mmdet/ops/box_iou_rotated + +#include "rbox_iou_op.h" +#include "paddle/extension.h" + +// 2D block with 32 * 16 = 512 threads per block +const int BLOCK_DIM_X = 32; +const int BLOCK_DIM_Y = 16; + +/** + Computes ceil(a / b) +*/ + +static inline int CeilDiv(const int a, const int b) { + return (a + b - 1) / b; +} + +template +__global__ void rbox_iou_cuda_kernel( + const int rbox1_num, + const int rbox2_num, + const T* rbox1_data_ptr, + const T* rbox2_data_ptr, + T* output_data_ptr) { + + // get row_start and col_start + const int rbox1_block_idx = blockIdx.x * blockDim.x; + const int rbox2_block_idx = blockIdx.y * blockDim.y; + + const int rbox1_thread_num = min(rbox1_num - rbox1_block_idx, blockDim.x); + const int rbox2_thread_num = min(rbox2_num - rbox2_block_idx, blockDim.y); + + __shared__ T block_boxes1[BLOCK_DIM_X * 5]; + __shared__ T block_boxes2[BLOCK_DIM_Y * 5]; + + + // It's safe to copy using threadIdx.x since BLOCK_DIM_X >= BLOCK_DIM_Y + if (threadIdx.x < rbox1_thread_num && threadIdx.y == 0) { + block_boxes1[threadIdx.x * 5 + 0] = + rbox1_data_ptr[(rbox1_block_idx + threadIdx.x) * 5 + 0]; + block_boxes1[threadIdx.x * 5 + 1] = + rbox1_data_ptr[(rbox1_block_idx + threadIdx.x) * 5 + 1]; + block_boxes1[threadIdx.x * 5 + 2] = + rbox1_data_ptr[(rbox1_block_idx + threadIdx.x) * 5 + 2]; + block_boxes1[threadIdx.x * 5 + 3] = + rbox1_data_ptr[(rbox1_block_idx + threadIdx.x) * 5 + 3]; + block_boxes1[threadIdx.x * 5 + 4] = + rbox1_data_ptr[(rbox1_block_idx + threadIdx.x) * 5 + 4]; + } + + // threadIdx.x < BLOCK_DIM_Y=rbox2_thread_num, just use same condition as above: threadIdx.y == 0 + if (threadIdx.x < rbox2_thread_num && threadIdx.y == 0) { + block_boxes2[threadIdx.x * 5 + 0] = + rbox2_data_ptr[(rbox2_block_idx + threadIdx.x) * 5 + 0]; + block_boxes2[threadIdx.x * 5 + 1] = + rbox2_data_ptr[(rbox2_block_idx + threadIdx.x) * 5 + 1]; + block_boxes2[threadIdx.x * 5 + 2] = + rbox2_data_ptr[(rbox2_block_idx + threadIdx.x) * 5 + 2]; + block_boxes2[threadIdx.x * 5 + 3] = + rbox2_data_ptr[(rbox2_block_idx + threadIdx.x) * 5 + 3]; + block_boxes2[threadIdx.x * 5 + 4] = + rbox2_data_ptr[(rbox2_block_idx + threadIdx.x) * 5 + 4]; + } + + // sync + __syncthreads(); + + if (threadIdx.x < rbox1_thread_num && threadIdx.y < rbox2_thread_num) { + int offset = (rbox1_block_idx + threadIdx.x) * rbox2_num + rbox2_block_idx + threadIdx.y; + output_data_ptr[offset] = rbox_iou_single(block_boxes1 + threadIdx.x * 5, block_boxes2 + threadIdx.y * 5); + } +} + +#define CHECK_INPUT_GPU(x) PD_CHECK(x.place() == paddle::PlaceType::kGPU, #x " must be a GPU Tensor.") + +std::vector RboxIouCUDAForward(const paddle::Tensor& rbox1, const paddle::Tensor& rbox2) { + CHECK_INPUT_GPU(rbox1); + CHECK_INPUT_GPU(rbox2); + + auto rbox1_num = rbox1.shape()[0]; + auto rbox2_num = rbox2.shape()[0]; + + auto output = paddle::Tensor(paddle::PlaceType::kGPU, {rbox1_num, rbox2_num}); + + const int blocks_x = CeilDiv(rbox1_num, BLOCK_DIM_X); + const int blocks_y = CeilDiv(rbox2_num, BLOCK_DIM_Y); + + dim3 blocks(blocks_x, blocks_y); + dim3 threads(BLOCK_DIM_X, BLOCK_DIM_Y); + + PD_DISPATCH_FLOATING_TYPES( + rbox1.type(), + "rbox_iou_cuda_kernel", + ([&] { + rbox_iou_cuda_kernel<<>>( + rbox1_num, + rbox2_num, + rbox1.data(), + rbox2.data(), + output.mutable_data()); + })); + + return {output}; +} + + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.h b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.h new file mode 100644 index 000000000..77fb62e39 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/rbox_iou_op.h @@ -0,0 +1,356 @@ +// Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +// +// Licensed under the Apache License, Version 2.0 (the "License"); +// you may not use this file except in compliance with the License. +// You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, software +// distributed under the License is distributed on an "AS IS" BASIS, +// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +// See the License for the specific language governing permissions and +// limitations under the License. +// +// The code is based on https://github.com/csuhan/s2anet/blob/master/mmdet/ops/box_iou_rotated + +#pragma once + +#include +#include +#include + +#ifdef __CUDACC__ +// Designates functions callable from the host (CPU) and the device (GPU) +#define HOST_DEVICE __host__ __device__ +#define HOST_DEVICE_INLINE HOST_DEVICE __forceinline__ +#else +#include +#define HOST_DEVICE +#define HOST_DEVICE_INLINE HOST_DEVICE inline +#endif + +namespace { + +template +struct RotatedBox { + T x_ctr, y_ctr, w, h, a; +}; + +template +struct Point { + T x, y; + HOST_DEVICE_INLINE Point(const T& px = 0, const T& py = 0) : x(px), y(py) {} + HOST_DEVICE_INLINE Point operator+(const Point& p) const { + return Point(x + p.x, y + p.y); + } + HOST_DEVICE_INLINE Point& operator+=(const Point& p) { + x += p.x; + y += p.y; + return *this; + } + HOST_DEVICE_INLINE Point operator-(const Point& p) const { + return Point(x - p.x, y - p.y); + } + HOST_DEVICE_INLINE Point operator*(const T coeff) const { + return Point(x * coeff, y * coeff); + } +}; + +template +HOST_DEVICE_INLINE T dot_2d(const Point& A, const Point& B) { + return A.x * B.x + A.y * B.y; +} + +template +HOST_DEVICE_INLINE T cross_2d(const Point& A, const Point& B) { + return A.x * B.y - B.x * A.y; +} + +template +HOST_DEVICE_INLINE void get_rotated_vertices( + const RotatedBox& box, + Point (&pts)[4]) { + // M_PI / 180. == 0.01745329251 + //double theta = box.a * 0.01745329251; + //MODIFIED + double theta = box.a; + T cosTheta2 = (T)cos(theta) * 0.5f; + T sinTheta2 = (T)sin(theta) * 0.5f; + + // y: top --> down; x: left --> right + pts[0].x = box.x_ctr - sinTheta2 * box.h - cosTheta2 * box.w; + pts[0].y = box.y_ctr + cosTheta2 * box.h - sinTheta2 * box.w; + pts[1].x = box.x_ctr + sinTheta2 * box.h - cosTheta2 * box.w; + pts[1].y = box.y_ctr - cosTheta2 * box.h - sinTheta2 * box.w; + pts[2].x = 2 * box.x_ctr - pts[0].x; + pts[2].y = 2 * box.y_ctr - pts[0].y; + pts[3].x = 2 * box.x_ctr - pts[1].x; + pts[3].y = 2 * box.y_ctr - pts[1].y; +} + +template +HOST_DEVICE_INLINE int get_intersection_points( + const Point (&pts1)[4], + const Point (&pts2)[4], + Point (&intersections)[24]) { + // Line vector + // A line from p1 to p2 is: p1 + (p2-p1)*t, t=[0,1] + Point vec1[4], vec2[4]; + for (int i = 0; i < 4; i++) { + vec1[i] = pts1[(i + 1) % 4] - pts1[i]; + vec2[i] = pts2[(i + 1) % 4] - pts2[i]; + } + + // Line test - test all line combos for intersection + int num = 0; // number of intersections + for (int i = 0; i < 4; i++) { + for (int j = 0; j < 4; j++) { + // Solve for 2x2 Ax=b + T det = cross_2d(vec2[j], vec1[i]); + + // This takes care of parallel lines + if (fabs(det) <= 1e-14) { + continue; + } + + auto vec12 = pts2[j] - pts1[i]; + + T t1 = cross_2d(vec2[j], vec12) / det; + T t2 = cross_2d(vec1[i], vec12) / det; + + if (t1 >= 0.0f && t1 <= 1.0f && t2 >= 0.0f && t2 <= 1.0f) { + intersections[num++] = pts1[i] + vec1[i] * t1; + } + } + } + + // Check for vertices of rect1 inside rect2 + { + const auto& AB = vec2[0]; + const auto& DA = vec2[3]; + auto ABdotAB = dot_2d(AB, AB); + auto ADdotAD = dot_2d(DA, DA); + for (int i = 0; i < 4; i++) { + // assume ABCD is the rectangle, and P is the point to be judged + // P is inside ABCD iff. P's projection on AB lies within AB + // and P's projection on AD lies within AD + + auto AP = pts1[i] - pts2[0]; + + auto APdotAB = dot_2d(AP, AB); + auto APdotAD = -dot_2d(AP, DA); + + if ((APdotAB >= 0) && (APdotAD >= 0) && (APdotAB <= ABdotAB) && + (APdotAD <= ADdotAD)) { + intersections[num++] = pts1[i]; + } + } + } + + // Reverse the check - check for vertices of rect2 inside rect1 + { + const auto& AB = vec1[0]; + const auto& DA = vec1[3]; + auto ABdotAB = dot_2d(AB, AB); + auto ADdotAD = dot_2d(DA, DA); + for (int i = 0; i < 4; i++) { + auto AP = pts2[i] - pts1[0]; + + auto APdotAB = dot_2d(AP, AB); + auto APdotAD = -dot_2d(AP, DA); + + if ((APdotAB >= 0) && (APdotAD >= 0) && (APdotAB <= ABdotAB) && + (APdotAD <= ADdotAD)) { + intersections[num++] = pts2[i]; + } + } + } + + return num; +} + +template +HOST_DEVICE_INLINE int convex_hull_graham( + const Point (&p)[24], + const int& num_in, + Point (&q)[24], + bool shift_to_zero = false) { + assert(num_in >= 2); + + // Step 1: + // Find point with minimum y + // if more than 1 points have the same minimum y, + // pick the one with the minimum x. + int t = 0; + for (int i = 1; i < num_in; i++) { + if (p[i].y < p[t].y || (p[i].y == p[t].y && p[i].x < p[t].x)) { + t = i; + } + } + auto& start = p[t]; // starting point + + // Step 2: + // Subtract starting point from every points (for sorting in the next step) + for (int i = 0; i < num_in; i++) { + q[i] = p[i] - start; + } + + // Swap the starting point to position 0 + auto tmp = q[0]; + q[0] = q[t]; + q[t] = tmp; + + // Step 3: + // Sort point 1 ~ num_in according to their relative cross-product values + // (essentially sorting according to angles) + // If the angles are the same, sort according to their distance to origin + T dist[24]; + for (int i = 0; i < num_in; i++) { + dist[i] = dot_2d(q[i], q[i]); + } + +#ifdef __CUDACC__ + // CUDA version + // In the future, we can potentially use thrust + // for sorting here to improve speed (though not guaranteed) + for (int i = 1; i < num_in - 1; i++) { + for (int j = i + 1; j < num_in; j++) { + T crossProduct = cross_2d(q[i], q[j]); + if ((crossProduct < -1e-6) || + (fabs(crossProduct) < 1e-6 && dist[i] > dist[j])) { + auto q_tmp = q[i]; + q[i] = q[j]; + q[j] = q_tmp; + auto dist_tmp = dist[i]; + dist[i] = dist[j]; + dist[j] = dist_tmp; + } + } + } +#else + // CPU version + std::sort( + q + 1, q + num_in, [](const Point& A, const Point& B) -> bool { + T temp = cross_2d(A, B); + if (fabs(temp) < 1e-6) { + return dot_2d(A, A) < dot_2d(B, B); + } else { + return temp > 0; + } + }); +#endif + + // Step 4: + // Make sure there are at least 2 points (that don't overlap with each other) + // in the stack + int k; // index of the non-overlapped second point + for (k = 1; k < num_in; k++) { + if (dist[k] > 1e-8) { + break; + } + } + if (k == num_in) { + // We reach the end, which means the convex hull is just one point + q[0] = p[t]; + return 1; + } + q[1] = q[k]; + int m = 2; // 2 points in the stack + // Step 5: + // Finally we can start the scanning process. + // When a non-convex relationship between the 3 points is found + // (either concave shape or duplicated points), + // we pop the previous point from the stack + // until the 3-point relationship is convex again, or + // until the stack only contains two points + for (int i = k + 1; i < num_in; i++) { + while (m > 1 && cross_2d(q[i] - q[m - 2], q[m - 1] - q[m - 2]) >= 0) { + m--; + } + q[m++] = q[i]; + } + + // Step 6 (Optional): + // In general sense we need the original coordinates, so we + // need to shift the points back (reverting Step 2) + // But if we're only interested in getting the area/perimeter of the shape + // We can simply return. + if (!shift_to_zero) { + for (int i = 0; i < m; i++) { + q[i] += start; + } + } + + return m; +} + +template +HOST_DEVICE_INLINE T polygon_area(const Point (&q)[24], const int& m) { + if (m <= 2) { + return 0; + } + + T area = 0; + for (int i = 1; i < m - 1; i++) { + area += fabs(cross_2d(q[i] - q[0], q[i + 1] - q[0])); + } + + return area / 2.0; +} + +template +HOST_DEVICE_INLINE T rboxes_intersection( + const RotatedBox& box1, + const RotatedBox& box2) { + // There are up to 4 x 4 + 4 + 4 = 24 intersections (including dups) returned + // from rotated_rect_intersection_pts + Point intersectPts[24], orderedPts[24]; + + Point pts1[4]; + Point pts2[4]; + get_rotated_vertices(box1, pts1); + get_rotated_vertices(box2, pts2); + + int num = get_intersection_points(pts1, pts2, intersectPts); + + if (num <= 2) { + return 0.0; + } + + // Convex Hull to order the intersection points in clockwise order and find + // the contour area. + int num_convex = convex_hull_graham(intersectPts, num, orderedPts, true); + return polygon_area(orderedPts, num_convex); +} + +} // namespace + +template +HOST_DEVICE_INLINE T +rbox_iou_single(T const* const box1_raw, T const* const box2_raw) { + // shift center to the middle point to achieve higher precision in result + RotatedBox box1, box2; + auto center_shift_x = (box1_raw[0] + box2_raw[0]) / 2.0; + auto center_shift_y = (box1_raw[1] + box2_raw[1]) / 2.0; + box1.x_ctr = box1_raw[0] - center_shift_x; + box1.y_ctr = box1_raw[1] - center_shift_y; + box1.w = box1_raw[2]; + box1.h = box1_raw[3]; + box1.a = box1_raw[4]; + box2.x_ctr = box2_raw[0] - center_shift_x; + box2.y_ctr = box2_raw[1] - center_shift_y; + box2.w = box2_raw[2]; + box2.h = box2_raw[3]; + box2.a = box2_raw[4]; + + const T area1 = box1.w * box1.h; + const T area2 = box2.w * box2.h; + if (area1 < 1e-14 || area2 < 1e-14) { + return 0.f; + } + + const T intersection = rboxes_intersection(box1, box2); + const T iou = intersection / (area1 + area2 - intersection); + return iou; +} diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/setup.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/setup.py new file mode 100644 index 000000000..d364db7ed --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/setup.py @@ -0,0 +1,14 @@ +import paddle +from paddle.utils.cpp_extension import CppExtension, CUDAExtension, setup + +if __name__ == "__main__": + if paddle.device.is_compiled_with_cuda(): + setup( + name='rbox_iou_ops', + ext_modules=CUDAExtension( + sources=['rbox_iou_op.cc', 'rbox_iou_op.cu'], + extra_compile_args={'cxx': ['-DPADDLE_WITH_CUDA']})) + else: + setup( + name='rbox_iou_ops', + ext_modules=CppExtension(sources=['rbox_iou_op.cc'])) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/test.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/test.py new file mode 100644 index 000000000..85872e484 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/ext_op/test.py @@ -0,0 +1,156 @@ +import numpy as np +import sys +import time +from shapely.geometry import Polygon +import paddle +import unittest + +try: + from rbox_iou_ops import rbox_iou +except Exception as e: + print('import rbox_iou_ops error', e) + sys.exit(-1) + + +def rbox2poly_single(rrect, get_best_begin_point=False): + """ + rrect:[x_ctr,y_ctr,w,h,angle] + to + poly:[x0,y0,x1,y1,x2,y2,x3,y3] + """ + x_ctr, y_ctr, width, height, angle = rrect[:5] + tl_x, tl_y, br_x, br_y = -width / 2, -height / 2, width / 2, height / 2 + # rect 2x4 + rect = np.array([[tl_x, br_x, br_x, tl_x], [tl_y, tl_y, br_y, br_y]]) + R = np.array([[np.cos(angle), -np.sin(angle)], + [np.sin(angle), np.cos(angle)]]) + # poly + poly = R.dot(rect) + x0, x1, x2, x3 = poly[0, :4] + x_ctr + y0, y1, y2, y3 = poly[1, :4] + y_ctr + poly = np.array([x0, y0, x1, y1, x2, y2, x3, y3], dtype=np.float64) + return poly + + +def intersection(g, p): + """ + Intersection. + """ + + g = g[:8].reshape((4, 2)) + p = p[:8].reshape((4, 2)) + + a = g + b = p + + use_filter = True + if use_filter: + # step1: + inter_x1 = np.maximum(np.min(a[:, 0]), np.min(b[:, 0])) + inter_x2 = np.minimum(np.max(a[:, 0]), np.max(b[:, 0])) + inter_y1 = np.maximum(np.min(a[:, 1]), np.min(b[:, 1])) + inter_y2 = np.minimum(np.max(a[:, 1]), np.max(b[:, 1])) + if inter_x1 >= inter_x2 or inter_y1 >= inter_y2: + return 0. + x1 = np.minimum(np.min(a[:, 0]), np.min(b[:, 0])) + x2 = np.maximum(np.max(a[:, 0]), np.max(b[:, 0])) + y1 = np.minimum(np.min(a[:, 1]), np.min(b[:, 1])) + y2 = np.maximum(np.max(a[:, 1]), np.max(b[:, 1])) + if x1 >= x2 or y1 >= y2 or (x2 - x1) < 2 or (y2 - y1) < 2: + return 0. + + g = Polygon(g) + p = Polygon(p) + if not g.is_valid or not p.is_valid: + return 0 + + inter = Polygon(g).intersection(Polygon(p)).area + union = g.area + p.area - inter + if union == 0: + return 0 + else: + return inter / union + + +def rbox_overlaps(anchors, gt_bboxes, use_cv2=False): + """ + + Args: + anchors: [NA, 5] x1,y1,x2,y2,angle + gt_bboxes: [M, 5] x1,y1,x2,y2,angle + + Returns: + + """ + assert anchors.shape[1] == 5 + assert gt_bboxes.shape[1] == 5 + + gt_bboxes_ploy = [rbox2poly_single(e) for e in gt_bboxes] + anchors_ploy = [rbox2poly_single(e) for e in anchors] + + num_gt, num_anchors = len(gt_bboxes_ploy), len(anchors_ploy) + iou = np.zeros((num_gt, num_anchors), dtype=np.float64) + + start_time = time.time() + for i in range(num_gt): + for j in range(num_anchors): + try: + iou[i, j] = intersection(gt_bboxes_ploy[i], anchors_ploy[j]) + except Exception as e: + print('cur gt_bboxes_ploy[i]', gt_bboxes_ploy[i], + 'anchors_ploy[j]', anchors_ploy[j], e) + iou = iou.T + return iou + + +def gen_sample(n): + rbox = np.random.rand(n, 5) + rbox[:, 0:4] = rbox[:, 0:4] * 0.45 + 0.001 + rbox[:, 4] = rbox[:, 4] - 0.5 + return rbox + + +class RBoxIoUTest(unittest.TestCase): + def setUp(self): + self.initTestCase() + self.rbox1 = gen_sample(self.n) + self.rbox2 = gen_sample(self.m) + + def initTestCase(self): + self.n = 13000 + self.m = 7 + + def assertAllClose(self, x, y, msg, atol=5e-1, rtol=1e-2): + self.assertTrue(np.allclose(x, y, atol=atol, rtol=rtol), msg=msg) + + def get_places(self): + places = [paddle.CPUPlace()] + if paddle.device.is_compiled_with_cuda(): + places.append(paddle.CUDAPlace(0)) + + return places + + def check_output(self, place): + paddle.disable_static() + pd_rbox1 = paddle.to_tensor(self.rbox1, place=place) + pd_rbox2 = paddle.to_tensor(self.rbox2, place=place) + actual_t = rbox_iou(pd_rbox1, pd_rbox2).numpy() + poly_rbox1 = self.rbox1 + poly_rbox2 = self.rbox2 + poly_rbox1[:, 0:4] = self.rbox1[:, 0:4] * 1024 + poly_rbox2[:, 0:4] = self.rbox2[:, 0:4] * 1024 + expect_t = rbox_overlaps(poly_rbox1, poly_rbox2, use_cv2=False) + self.assertAllClose( + actual_t, + expect_t, + msg="rbox_iou has diff at {} \nExpect {}\nBut got {}".format( + str(place), str(expect_t), str(actual_t))) + + def test_output(self): + places = self.get_places() + for place in places: + self.check_output(place) + + +if __name__ == "__main__": + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__init__.py new file mode 100644 index 000000000..d69e8af0f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__init__.py @@ -0,0 +1,29 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import metrics +from . import keypoint_metrics + +from .metrics import * +from .keypoint_metrics import * + +__all__ = metrics.__all__ + keypoint_metrics.__all__ + +from . import mot_metrics +from .mot_metrics import * +__all__ = metrics.__all__ + mot_metrics.__all__ + +from . import mcmot_metrics +from .mcmot_metrics import * +__all__ = metrics.__all__ + mcmot_metrics.__all__ \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..08ea6f44d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/coco_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/coco_utils.cpython-37.pyc new file mode 100644 index 000000000..c55b3d121 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/coco_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/json_results.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/json_results.cpython-37.pyc new file mode 100644 index 000000000..d99601c30 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/json_results.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/keypoint_metrics.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/keypoint_metrics.cpython-37.pyc new file mode 100644 index 000000000..3ea557dda Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/keypoint_metrics.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/map_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/map_utils.cpython-37.pyc new file mode 100644 index 000000000..c2a9ed2f0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/map_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mcmot_metrics.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mcmot_metrics.cpython-37.pyc new file mode 100644 index 000000000..1390aa87b Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mcmot_metrics.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/metrics.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/metrics.cpython-37.pyc new file mode 100644 index 000000000..0a4895277 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/metrics.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mot_metrics.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mot_metrics.cpython-37.pyc new file mode 100644 index 000000000..683014969 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/mot_metrics.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/munkres.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/munkres.cpython-37.pyc new file mode 100644 index 000000000..c00df5d8f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/munkres.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/widerface_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/widerface_utils.cpython-37.pyc new file mode 100644 index 000000000..bd9959299 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/__pycache__/widerface_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/coco_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/coco_utils.py new file mode 100644 index 000000000..47b92bc62 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/coco_utils.py @@ -0,0 +1,184 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys +import numpy as np +import itertools + +from ppdet.metrics.json_results import get_det_res, get_det_poly_res, get_seg_res, get_solov2_segm_res, get_keypoint_res +from ppdet.metrics.map_utils import draw_pr_curve + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +def get_infer_results(outs, catid, bias=0): + """ + Get result at the stage of inference. + The output format is dictionary containing bbox or mask result. + + For example, bbox result is a list and each element contains + image_id, category_id, bbox and score. + """ + if outs is None or len(outs) == 0: + raise ValueError( + 'The number of valid detection result if zero. Please use reasonable model and check input data.' + ) + + im_id = outs['im_id'] + + infer_res = {} + if 'bbox' in outs: + if len(outs['bbox']) > 0 and len(outs['bbox'][0]) > 6: + infer_res['bbox'] = get_det_poly_res( + outs['bbox'], outs['bbox_num'], im_id, catid, bias=bias) + else: + infer_res['bbox'] = get_det_res( + outs['bbox'], outs['bbox_num'], im_id, catid, bias=bias) + + if 'mask' in outs: + # mask post process + infer_res['mask'] = get_seg_res(outs['mask'], outs['bbox'], + outs['bbox_num'], im_id, catid) + + if 'segm' in outs: + infer_res['segm'] = get_solov2_segm_res(outs, im_id, catid) + + if 'keypoint' in outs: + infer_res['keypoint'] = get_keypoint_res(outs, im_id) + outs['bbox_num'] = [len(infer_res['keypoint'])] + + return infer_res + + +def cocoapi_eval(jsonfile, + style, + coco_gt=None, + anno_file=None, + max_dets=(100, 300, 1000), + classwise=False, + sigmas=None, + use_area=True): + """ + Args: + jsonfile (str): Evaluation json file, eg: bbox.json, mask.json. + style (str): COCOeval style, can be `bbox` , `segm` , `proposal`, `keypoints` and `keypoints_crowd`. + coco_gt (str): Whether to load COCOAPI through anno_file, + eg: coco_gt = COCO(anno_file) + anno_file (str): COCO annotations file. + max_dets (tuple): COCO evaluation maxDets. + classwise (bool): Whether per-category AP and draw P-R Curve or not. + sigmas (nparray): keypoint labelling sigmas. + use_area (bool): If gt annotations (eg. CrowdPose, AIC) + do not have 'area', please set use_area=False. + """ + assert coco_gt != None or anno_file != None + if style == 'keypoints_crowd': + #please install xtcocotools==1.6 + from xtcocotools.coco import COCO + from xtcocotools.cocoeval import COCOeval + else: + from pycocotools.coco import COCO + from pycocotools.cocoeval import COCOeval + + if coco_gt == None: + coco_gt = COCO(anno_file) + logger.info("Start evaluate...") + coco_dt = coco_gt.loadRes(jsonfile) + if style == 'proposal': + coco_eval = COCOeval(coco_gt, coco_dt, 'bbox') + coco_eval.params.useCats = 0 + coco_eval.params.maxDets = list(max_dets) + elif style == 'keypoints_crowd': + coco_eval = COCOeval(coco_gt, coco_dt, style, sigmas, use_area) + else: + coco_eval = COCOeval(coco_gt, coco_dt, style) + coco_eval.evaluate() + coco_eval.accumulate() + coco_eval.summarize() + if classwise: + # Compute per-category AP and PR curve + try: + from terminaltables import AsciiTable + except Exception as e: + logger.error( + 'terminaltables not found, plaese install terminaltables. ' + 'for example: `pip install terminaltables`.') + raise e + precisions = coco_eval.eval['precision'] + cat_ids = coco_gt.getCatIds() + # precision: (iou, recall, cls, area range, max dets) + assert len(cat_ids) == precisions.shape[2] + results_per_category = [] + for idx, catId in enumerate(cat_ids): + # area range index 0: all area ranges + # max dets index -1: typically 100 per image + nm = coco_gt.loadCats(catId)[0] + precision = precisions[:, :, idx, 0, -1] + precision = precision[precision > -1] + if precision.size: + ap = np.mean(precision) + else: + ap = float('nan') + results_per_category.append( + (str(nm["name"]), '{:0.3f}'.format(float(ap)))) + pr_array = precisions[0, :, idx, 0, 2] + recall_array = np.arange(0.0, 1.01, 0.01) + draw_pr_curve( + pr_array, + recall_array, + out_dir=style + '_pr_curve', + file_name='{}_precision_recall_curve.jpg'.format(nm["name"])) + + num_columns = min(6, len(results_per_category) * 2) + results_flatten = list(itertools.chain(*results_per_category)) + headers = ['category', 'AP'] * (num_columns // 2) + results_2d = itertools.zip_longest( + *[results_flatten[i::num_columns] for i in range(num_columns)]) + table_data = [headers] + table_data += [result for result in results_2d] + table = AsciiTable(table_data) + logger.info('Per-category of {} AP: \n{}'.format(style, table.table)) + logger.info("per-category PR curve has output to {} folder.".format( + style + '_pr_curve')) + # flush coco evaluation result + sys.stdout.flush() + return coco_eval.stats + + +def json_eval_results(metric, json_directory, dataset): + """ + cocoapi eval with already exists proposal.json, bbox.json or mask.json + """ + assert metric == 'COCO' + anno_file = dataset.get_anno() + json_file_list = ['proposal.json', 'bbox.json', 'mask.json'] + if json_directory: + assert os.path.exists( + json_directory), "The json directory:{} does not exist".format( + json_directory) + for k, v in enumerate(json_file_list): + json_file_list[k] = os.path.join(str(json_directory), v) + + coco_eval_style = ['proposal', 'bbox', 'segm'] + for i, v_json in enumerate(json_file_list): + if os.path.exists(v_json): + cocoapi_eval(v_json, coco_eval_style[i], anno_file=anno_file) + else: + logger.info("{} not exists!".format(v_json)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/json_results.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/json_results.py new file mode 100644 index 000000000..c703de63b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/json_results.py @@ -0,0 +1,149 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import six +import numpy as np + + +def get_det_res(bboxes, bbox_nums, image_id, label_to_cat_id_map, bias=0): + det_res = [] + k = 0 + for i in range(len(bbox_nums)): + cur_image_id = int(image_id[i][0]) + det_nums = bbox_nums[i] + for j in range(det_nums): + dt = bboxes[k] + k = k + 1 + num_id, score, xmin, ymin, xmax, ymax = dt.tolist() + if int(num_id) < 0: + continue + category_id = label_to_cat_id_map[int(num_id)] + w = xmax - xmin + bias + h = ymax - ymin + bias + bbox = [xmin, ymin, w, h] + dt_res = { + 'image_id': cur_image_id, + 'category_id': category_id, + 'bbox': bbox, + 'score': score + } + det_res.append(dt_res) + return det_res + + +def get_det_poly_res(bboxes, bbox_nums, image_id, label_to_cat_id_map, bias=0): + det_res = [] + k = 0 + for i in range(len(bbox_nums)): + cur_image_id = int(image_id[i][0]) + det_nums = bbox_nums[i] + for j in range(det_nums): + dt = bboxes[k] + k = k + 1 + num_id, score, x1, y1, x2, y2, x3, y3, x4, y4 = dt.tolist() + if int(num_id) < 0: + continue + category_id = label_to_cat_id_map[int(num_id)] + rbox = [x1, y1, x2, y2, x3, y3, x4, y4] + dt_res = { + 'image_id': cur_image_id, + 'category_id': category_id, + 'bbox': rbox, + 'score': score + } + det_res.append(dt_res) + return det_res + + +def get_seg_res(masks, bboxes, mask_nums, image_id, label_to_cat_id_map): + import pycocotools.mask as mask_util + seg_res = [] + k = 0 + for i in range(len(mask_nums)): + cur_image_id = int(image_id[i][0]) + det_nums = mask_nums[i] + for j in range(det_nums): + mask = masks[k].astype(np.uint8) + score = float(bboxes[k][1]) + label = int(bboxes[k][0]) + k = k + 1 + if label == -1: + continue + cat_id = label_to_cat_id_map[label] + rle = mask_util.encode( + np.array( + mask[:, :, None], order="F", dtype="uint8"))[0] + if six.PY3: + if 'counts' in rle: + rle['counts'] = rle['counts'].decode("utf8") + sg_res = { + 'image_id': cur_image_id, + 'category_id': cat_id, + 'segmentation': rle, + 'score': score + } + seg_res.append(sg_res) + return seg_res + + +def get_solov2_segm_res(results, image_id, num_id_to_cat_id_map): + import pycocotools.mask as mask_util + segm_res = [] + # for each batch + segms = results['segm'].astype(np.uint8) + clsid_labels = results['cate_label'] + clsid_scores = results['cate_score'] + lengths = segms.shape[0] + im_id = int(image_id[0][0]) + if lengths == 0 or segms is None: + return None + # for each sample + for i in range(lengths - 1): + clsid = int(clsid_labels[i]) + catid = num_id_to_cat_id_map[clsid] + score = float(clsid_scores[i]) + mask = segms[i] + segm = mask_util.encode(np.array(mask[:, :, np.newaxis], order='F'))[0] + segm['counts'] = segm['counts'].decode('utf8') + coco_res = { + 'image_id': im_id, + 'category_id': catid, + 'segmentation': segm, + 'score': score + } + segm_res.append(coco_res) + return segm_res + + +def get_keypoint_res(results, im_id): + anns = [] + preds = results['keypoint'] + for idx in range(im_id.shape[0]): + image_id = im_id[idx].item() + kpts, scores = preds[idx] + for kpt, score in zip(kpts, scores): + kpt = kpt.flatten() + ann = { + 'image_id': image_id, + 'category_id': 1, # XXX hard code + 'keypoints': kpt.tolist(), + 'score': float(score) + } + x = kpt[0::3] + y = kpt[1::3] + x0, x1, y0, y1 = np.min(x).item(), np.max(x).item(), np.min(y).item( + ), np.max(y).item() + ann['area'] = (x1 - x0) * (y1 - y0) + ann['bbox'] = [x0, y0, x1 - x0, y1 - y0] + anns.append(ann) + return anns diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/keypoint_metrics.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/keypoint_metrics.py new file mode 100644 index 000000000..d8bc0e782 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/keypoint_metrics.py @@ -0,0 +1,402 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import json +from collections import defaultdict, OrderedDict +import numpy as np +from pycocotools.coco import COCO +from pycocotools.cocoeval import COCOeval +from ..modeling.keypoint_utils import oks_nms +from scipy.io import loadmat, savemat +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['KeyPointTopDownCOCOEval', 'KeyPointTopDownMPIIEval'] + + +class KeyPointTopDownCOCOEval(object): + ''' + Adapted from + https://github.com/leoxiaobin/deep-high-resolution-net.pytorch + Copyright (c) Microsoft, under the MIT License. + ''' + + def __init__(self, + anno_file, + num_samples, + num_joints, + output_eval, + iou_type='keypoints', + in_vis_thre=0.2, + oks_thre=0.9, + save_prediction_only=False): + super(KeyPointTopDownCOCOEval, self).__init__() + self.coco = COCO(anno_file) + self.num_samples = num_samples + self.num_joints = num_joints + self.iou_type = iou_type + self.in_vis_thre = in_vis_thre + self.oks_thre = oks_thre + self.output_eval = output_eval + self.res_file = os.path.join(output_eval, "keypoints_results.json") + self.save_prediction_only = save_prediction_only + self.reset() + + def reset(self): + self.results = { + 'all_preds': np.zeros( + (self.num_samples, self.num_joints, 3), dtype=np.float32), + 'all_boxes': np.zeros((self.num_samples, 6)), + 'image_path': [] + } + self.eval_results = {} + self.idx = 0 + + def update(self, inputs, outputs): + kpts, _ = outputs['keypoint'][0] + + num_images = inputs['image'].shape[0] + self.results['all_preds'][self.idx:self.idx + num_images, :, 0: + 3] = kpts[:, :, 0:3] + self.results['all_boxes'][self.idx:self.idx + num_images, 0:2] = inputs[ + 'center'].numpy()[:, 0:2] + self.results['all_boxes'][self.idx:self.idx + num_images, 2:4] = inputs[ + 'scale'].numpy()[:, 0:2] + self.results['all_boxes'][self.idx:self.idx + num_images, 4] = np.prod( + inputs['scale'].numpy() * 200, 1) + self.results['all_boxes'][self.idx:self.idx + num_images, + 5] = np.squeeze(inputs['score'].numpy()) + self.results['image_path'].extend(inputs['im_id'].numpy()) + + self.idx += num_images + + def _write_coco_keypoint_results(self, keypoints): + data_pack = [{ + 'cat_id': 1, + 'cls': 'person', + 'ann_type': 'keypoints', + 'keypoints': keypoints + }] + results = self._coco_keypoint_results_one_category_kernel(data_pack[0]) + if not os.path.exists(self.output_eval): + os.makedirs(self.output_eval) + with open(self.res_file, 'w') as f: + json.dump(results, f, sort_keys=True, indent=4) + logger.info(f'The keypoint result is saved to {self.res_file}.') + try: + json.load(open(self.res_file)) + except Exception: + content = [] + with open(self.res_file, 'r') as f: + for line in f: + content.append(line) + content[-1] = ']' + with open(self.res_file, 'w') as f: + for c in content: + f.write(c) + + def _coco_keypoint_results_one_category_kernel(self, data_pack): + cat_id = data_pack['cat_id'] + keypoints = data_pack['keypoints'] + cat_results = [] + + for img_kpts in keypoints: + if len(img_kpts) == 0: + continue + + _key_points = np.array( + [img_kpts[k]['keypoints'] for k in range(len(img_kpts))]) + _key_points = _key_points.reshape(_key_points.shape[0], -1) + + result = [{ + 'image_id': img_kpts[k]['image'], + 'category_id': cat_id, + 'keypoints': _key_points[k].tolist(), + 'score': img_kpts[k]['score'], + 'center': list(img_kpts[k]['center']), + 'scale': list(img_kpts[k]['scale']) + } for k in range(len(img_kpts))] + cat_results.extend(result) + + return cat_results + + def get_final_results(self, preds, all_boxes, img_path): + _kpts = [] + for idx, kpt in enumerate(preds): + _kpts.append({ + 'keypoints': kpt, + 'center': all_boxes[idx][0:2], + 'scale': all_boxes[idx][2:4], + 'area': all_boxes[idx][4], + 'score': all_boxes[idx][5], + 'image': int(img_path[idx]) + }) + # image x person x (keypoints) + kpts = defaultdict(list) + for kpt in _kpts: + kpts[kpt['image']].append(kpt) + + # rescoring and oks nms + num_joints = preds.shape[1] + in_vis_thre = self.in_vis_thre + oks_thre = self.oks_thre + oks_nmsed_kpts = [] + for img in kpts.keys(): + img_kpts = kpts[img] + for n_p in img_kpts: + box_score = n_p['score'] + kpt_score = 0 + valid_num = 0 + for n_jt in range(0, num_joints): + t_s = n_p['keypoints'][n_jt][2] + if t_s > in_vis_thre: + kpt_score = kpt_score + t_s + valid_num = valid_num + 1 + if valid_num != 0: + kpt_score = kpt_score / valid_num + # rescoring + n_p['score'] = kpt_score * box_score + + keep = oks_nms([img_kpts[i] for i in range(len(img_kpts))], + oks_thre) + + if len(keep) == 0: + oks_nmsed_kpts.append(img_kpts) + else: + oks_nmsed_kpts.append([img_kpts[_keep] for _keep in keep]) + + self._write_coco_keypoint_results(oks_nmsed_kpts) + + def accumulate(self): + self.get_final_results(self.results['all_preds'], + self.results['all_boxes'], + self.results['image_path']) + if self.save_prediction_only: + logger.info(f'The keypoint result is saved to {self.res_file} ' + 'and do not evaluate the mAP.') + return + coco_dt = self.coco.loadRes(self.res_file) + coco_eval = COCOeval(self.coco, coco_dt, 'keypoints') + coco_eval.params.useSegm = None + coco_eval.evaluate() + coco_eval.accumulate() + coco_eval.summarize() + + keypoint_stats = [] + for ind in range(len(coco_eval.stats)): + keypoint_stats.append((coco_eval.stats[ind])) + self.eval_results['keypoint'] = keypoint_stats + + def log(self): + if self.save_prediction_only: + return + stats_names = [ + 'AP', 'Ap .5', 'AP .75', 'AP (M)', 'AP (L)', 'AR', 'AR .5', + 'AR .75', 'AR (M)', 'AR (L)' + ] + num_values = len(stats_names) + print(' '.join(['| {}'.format(name) for name in stats_names]) + ' |') + print('|---' * (num_values + 1) + '|') + + print(' '.join([ + '| {:.3f}'.format(value) for value in self.eval_results['keypoint'] + ]) + ' |') + + def get_results(self): + return self.eval_results + + +class KeyPointTopDownMPIIEval(object): + def __init__(self, + anno_file, + num_samples, + num_joints, + output_eval, + oks_thre=0.9, + save_prediction_only=False): + super(KeyPointTopDownMPIIEval, self).__init__() + self.ann_file = anno_file + self.res_file = os.path.join(output_eval, "keypoints_results.json") + self.save_prediction_only = save_prediction_only + self.reset() + + def reset(self): + self.results = [] + self.eval_results = {} + self.idx = 0 + + def update(self, inputs, outputs): + kpts, _ = outputs['keypoint'][0] + + num_images = inputs['image'].shape[0] + results = {} + results['preds'] = kpts[:, :, 0:3] + results['boxes'] = np.zeros((num_images, 6)) + results['boxes'][:, 0:2] = inputs['center'].numpy()[:, 0:2] + results['boxes'][:, 2:4] = inputs['scale'].numpy()[:, 0:2] + results['boxes'][:, 4] = np.prod(inputs['scale'].numpy() * 200, 1) + results['boxes'][:, 5] = np.squeeze(inputs['score'].numpy()) + results['image_path'] = inputs['image_file'] + + self.results.append(results) + + def accumulate(self): + self._mpii_keypoint_results_save() + if self.save_prediction_only: + logger.info(f'The keypoint result is saved to {self.res_file} ' + 'and do not evaluate the mAP.') + return + + self.eval_results = self.evaluate(self.results) + + def _mpii_keypoint_results_save(self): + results = [] + for res in self.results: + if len(res) == 0: + continue + result = [{ + 'preds': res['preds'][k].tolist(), + 'boxes': res['boxes'][k].tolist(), + 'image_path': res['image_path'][k], + } for k in range(len(res))] + results.extend(result) + with open(self.res_file, 'w') as f: + json.dump(results, f, sort_keys=True, indent=4) + logger.info(f'The keypoint result is saved to {self.res_file}.') + + def log(self): + if self.save_prediction_only: + return + for item, value in self.eval_results.items(): + print("{} : {}".format(item, value)) + + def get_results(self): + return self.eval_results + + def evaluate(self, outputs, savepath=None): + """Evaluate PCKh for MPII dataset. Adapted from + https://github.com/leoxiaobin/deep-high-resolution-net.pytorch + Copyright (c) Microsoft, under the MIT License. + + Args: + outputs(list(preds, boxes)): + + * preds (np.ndarray[N,K,3]): The first two dimensions are + coordinates, score is the third dimension of the array. + * boxes (np.ndarray[N,6]): [center[0], center[1], scale[0] + , scale[1],area, score] + + Returns: + dict: PCKh for each joint + """ + + kpts = [] + for output in outputs: + preds = output['preds'] + batch_size = preds.shape[0] + for i in range(batch_size): + kpts.append({'keypoints': preds[i]}) + + preds = np.stack([kpt['keypoints'] for kpt in kpts]) + + # convert 0-based index to 1-based index, + # and get the first two dimensions. + preds = preds[..., :2] + 1.0 + + if savepath is not None: + pred_file = os.path.join(savepath, 'pred.mat') + savemat(pred_file, mdict={'preds': preds}) + + SC_BIAS = 0.6 + threshold = 0.5 + + gt_file = os.path.join( + os.path.dirname(self.ann_file), 'mpii_gt_val.mat') + gt_dict = loadmat(gt_file) + dataset_joints = gt_dict['dataset_joints'] + jnt_missing = gt_dict['jnt_missing'] + pos_gt_src = gt_dict['pos_gt_src'] + headboxes_src = gt_dict['headboxes_src'] + + pos_pred_src = np.transpose(preds, [1, 2, 0]) + + head = np.where(dataset_joints == 'head')[1][0] + lsho = np.where(dataset_joints == 'lsho')[1][0] + lelb = np.where(dataset_joints == 'lelb')[1][0] + lwri = np.where(dataset_joints == 'lwri')[1][0] + lhip = np.where(dataset_joints == 'lhip')[1][0] + lkne = np.where(dataset_joints == 'lkne')[1][0] + lank = np.where(dataset_joints == 'lank')[1][0] + + rsho = np.where(dataset_joints == 'rsho')[1][0] + relb = np.where(dataset_joints == 'relb')[1][0] + rwri = np.where(dataset_joints == 'rwri')[1][0] + rkne = np.where(dataset_joints == 'rkne')[1][0] + rank = np.where(dataset_joints == 'rank')[1][0] + rhip = np.where(dataset_joints == 'rhip')[1][0] + + jnt_visible = 1 - jnt_missing + uv_error = pos_pred_src - pos_gt_src + uv_err = np.linalg.norm(uv_error, axis=1) + headsizes = headboxes_src[1, :, :] - headboxes_src[0, :, :] + headsizes = np.linalg.norm(headsizes, axis=0) + headsizes *= SC_BIAS + scale = headsizes * np.ones((len(uv_err), 1), dtype=np.float32) + scaled_uv_err = uv_err / scale + scaled_uv_err = scaled_uv_err * jnt_visible + jnt_count = np.sum(jnt_visible, axis=1) + less_than_threshold = (scaled_uv_err <= threshold) * jnt_visible + PCKh = 100. * np.sum(less_than_threshold, axis=1) / jnt_count + + # save + rng = np.arange(0, 0.5 + 0.01, 0.01) + pckAll = np.zeros((len(rng), 16), dtype=np.float32) + + for r, threshold in enumerate(rng): + less_than_threshold = (scaled_uv_err <= threshold) * jnt_visible + pckAll[r, :] = 100. * np.sum(less_than_threshold, + axis=1) / jnt_count + + PCKh = np.ma.array(PCKh, mask=False) + PCKh.mask[6:8] = True + + jnt_count = np.ma.array(jnt_count, mask=False) + jnt_count.mask[6:8] = True + jnt_ratio = jnt_count / np.sum(jnt_count).astype(np.float64) + + name_value = [ #noqa + ('Head', PCKh[head]), + ('Shoulder', 0.5 * (PCKh[lsho] + PCKh[rsho])), + ('Elbow', 0.5 * (PCKh[lelb] + PCKh[relb])), + ('Wrist', 0.5 * (PCKh[lwri] + PCKh[rwri])), + ('Hip', 0.5 * (PCKh[lhip] + PCKh[rhip])), + ('Knee', 0.5 * (PCKh[lkne] + PCKh[rkne])), + ('Ankle', 0.5 * (PCKh[lank] + PCKh[rank])), + ('PCKh', np.sum(PCKh * jnt_ratio)), + ('PCKh@0.1', np.sum(pckAll[11, :] * jnt_ratio)) + ] + name_value = OrderedDict(name_value) + + return name_value + + def _sort_and_unique_bboxes(self, kpts, key='bbox_id'): + """sort kpts and remove the repeated ones.""" + kpts = sorted(kpts, key=lambda x: x[key]) + num = len(kpts) + for i in range(num - 1, 0, -1): + if kpts[i][key] == kpts[i - 1][key]: + del kpts[i] + + return kpts diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/map_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/map_utils.py new file mode 100644 index 000000000..9c96b9235 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/map_utils.py @@ -0,0 +1,443 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +import os +import sys +import numpy as np +import itertools +import paddle +from ppdet.modeling.bbox_utils import poly2rbox, rbox2poly_np + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = [ + 'draw_pr_curve', + 'bbox_area', + 'jaccard_overlap', + 'prune_zero_padding', + 'DetectionMAP', + 'ap_per_class', + 'compute_ap', +] + + +def draw_pr_curve(precision, + recall, + iou=0.5, + out_dir='pr_curve', + file_name='precision_recall_curve.jpg'): + if not os.path.exists(out_dir): + os.makedirs(out_dir) + output_path = os.path.join(out_dir, file_name) + try: + import matplotlib.pyplot as plt + except Exception as e: + logger.error('Matplotlib not found, plaese install matplotlib.' + 'for example: `pip install matplotlib`.') + raise e + plt.cla() + plt.figure('P-R Curve') + plt.title('Precision/Recall Curve(IoU={})'.format(iou)) + plt.xlabel('Recall') + plt.ylabel('Precision') + plt.grid(True) + plt.plot(recall, precision) + plt.savefig(output_path) + + +def bbox_area(bbox, is_bbox_normalized): + """ + Calculate area of a bounding box + """ + norm = 1. - float(is_bbox_normalized) + width = bbox[2] - bbox[0] + norm + height = bbox[3] - bbox[1] + norm + return width * height + + +def jaccard_overlap(pred, gt, is_bbox_normalized=False): + """ + Calculate jaccard overlap ratio between two bounding box + """ + if pred[0] >= gt[2] or pred[2] <= gt[0] or \ + pred[1] >= gt[3] or pred[3] <= gt[1]: + return 0. + inter_xmin = max(pred[0], gt[0]) + inter_ymin = max(pred[1], gt[1]) + inter_xmax = min(pred[2], gt[2]) + inter_ymax = min(pred[3], gt[3]) + inter_size = bbox_area([inter_xmin, inter_ymin, inter_xmax, inter_ymax], + is_bbox_normalized) + pred_size = bbox_area(pred, is_bbox_normalized) + gt_size = bbox_area(gt, is_bbox_normalized) + overlap = float(inter_size) / (pred_size + gt_size - inter_size) + return overlap + + +def calc_rbox_iou(pred, gt_rbox): + """ + calc iou between rotated bbox + """ + # calc iou of bounding box for speedup + pred = np.array(pred, np.float32).reshape(-1, 8) + pred = pred.reshape(-1, 2) + gt_poly = rbox2poly_np(np.array(gt_rbox).reshape(-1, 5))[0] + gt_poly = gt_poly.reshape(-1, 2) + pred_rect = [ + np.min(pred[:, 0]), np.min(pred[:, 1]), np.max(pred[:, 0]), + np.max(pred[:, 1]) + ] + gt_rect = [ + np.min(gt_poly[:, 0]), np.min(gt_poly[:, 1]), np.max(gt_poly[:, 0]), + np.max(gt_poly[:, 1]) + ] + iou = jaccard_overlap(pred_rect, gt_rect, False) + + if iou <= 0: + return iou + + # calc rbox iou + pred = pred.reshape(-1, 8) + + pred = np.array(pred, np.float32).reshape(-1, 8) + pred_rbox = poly2rbox(pred) + pred_rbox = pred_rbox.reshape(-1, 5) + pred_rbox = pred_rbox.reshape(-1, 5) + try: + from rbox_iou_ops import rbox_iou + except Exception as e: + print("import custom_ops error, try install rbox_iou_ops " \ + "following ppdet/ext_op/README.md", e) + sys.stdout.flush() + sys.exit(-1) + gt_rbox = np.array(gt_rbox, np.float32).reshape(-1, 5) + pd_gt_rbox = paddle.to_tensor(gt_rbox, dtype='float32') + pd_pred_rbox = paddle.to_tensor(pred_rbox, dtype='float32') + iou = rbox_iou(pd_gt_rbox, pd_pred_rbox) + iou = iou.numpy() + return iou[0][0] + + +def prune_zero_padding(gt_box, gt_label, difficult=None): + valid_cnt = 0 + for i in range(len(gt_box)): + if gt_box[i, 0] == 0 and gt_box[i, 1] == 0 and \ + gt_box[i, 2] == 0 and gt_box[i, 3] == 0: + break + valid_cnt += 1 + return (gt_box[:valid_cnt], gt_label[:valid_cnt], difficult[:valid_cnt] + if difficult is not None else None) + + +class DetectionMAP(object): + """ + Calculate detection mean average precision. + Currently support two types: 11point and integral + + Args: + class_num (int): The class number. + overlap_thresh (float): The threshold of overlap + ratio between prediction bounding box and + ground truth bounding box for deciding + true/false positive. Default 0.5. + map_type (str): Calculation method of mean average + precision, currently support '11point' and + 'integral'. Default '11point'. + is_bbox_normalized (bool): Whether bounding boxes + is normalized to range[0, 1]. Default False. + evaluate_difficult (bool): Whether to evaluate + difficult bounding boxes. Default False. + catid2name (dict): Mapping between category id and category name. + classwise (bool): Whether per-category AP and draw + P-R Curve or not. + """ + + def __init__(self, + class_num, + overlap_thresh=0.5, + map_type='11point', + is_bbox_normalized=False, + evaluate_difficult=False, + catid2name=None, + classwise=False): + self.class_num = class_num + self.overlap_thresh = overlap_thresh + assert map_type in ['11point', 'integral'], \ + "map_type currently only support '11point' "\ + "and 'integral'" + self.map_type = map_type + self.is_bbox_normalized = is_bbox_normalized + self.evaluate_difficult = evaluate_difficult + self.classwise = classwise + self.classes = [] + for cname in catid2name.values(): + self.classes.append(cname) + self.reset() + + def update(self, bbox, score, label, gt_box, gt_label, difficult=None): + """ + Update metric statics from given prediction and ground + truth infomations. + """ + if difficult is None: + difficult = np.zeros_like(gt_label) + + # record class gt count + for gtl, diff in zip(gt_label, difficult): + if self.evaluate_difficult or int(diff) == 0: + self.class_gt_counts[int(np.array(gtl))] += 1 + + # record class score positive + visited = [False] * len(gt_label) + for b, s, l in zip(bbox, score, label): + pred = b.tolist() if isinstance(b, np.ndarray) else b + max_idx = -1 + max_overlap = -1.0 + for i, gl in enumerate(gt_label): + if int(gl) == int(l): + if len(gt_box[i]) == 5: + overlap = calc_rbox_iou(pred, gt_box[i]) + else: + overlap = jaccard_overlap(pred, gt_box[i], + self.is_bbox_normalized) + if overlap > max_overlap: + max_overlap = overlap + max_idx = i + + if max_overlap > self.overlap_thresh: + if self.evaluate_difficult or \ + int(np.array(difficult[max_idx])) == 0: + if not visited[max_idx]: + self.class_score_poss[int(l)].append([s, 1.0]) + visited[max_idx] = True + else: + self.class_score_poss[int(l)].append([s, 0.0]) + else: + self.class_score_poss[int(l)].append([s, 0.0]) + + def reset(self): + """ + Reset metric statics + """ + self.class_score_poss = [[] for _ in range(self.class_num)] + self.class_gt_counts = [0] * self.class_num + self.mAP = 0.0 + + def accumulate(self): + """ + Accumulate metric results and calculate mAP + """ + mAP = 0. + valid_cnt = 0 + eval_results = [] + for score_pos, count in zip(self.class_score_poss, + self.class_gt_counts): + if count == 0: continue + if len(score_pos) == 0: + valid_cnt += 1 + continue + + accum_tp_list, accum_fp_list = \ + self._get_tp_fp_accum(score_pos) + precision = [] + recall = [] + for ac_tp, ac_fp in zip(accum_tp_list, accum_fp_list): + precision.append(float(ac_tp) / (ac_tp + ac_fp)) + recall.append(float(ac_tp) / count) + + one_class_ap = 0.0 + if self.map_type == '11point': + max_precisions = [0.] * 11 + start_idx = len(precision) - 1 + for j in range(10, -1, -1): + for i in range(start_idx, -1, -1): + if recall[i] < float(j) / 10.: + start_idx = i + if j > 0: + max_precisions[j - 1] = max_precisions[j] + break + else: + if max_precisions[j] < precision[i]: + max_precisions[j] = precision[i] + one_class_ap = sum(max_precisions) / 11. + mAP += one_class_ap + valid_cnt += 1 + elif self.map_type == 'integral': + import math + prev_recall = 0. + for i in range(len(precision)): + recall_gap = math.fabs(recall[i] - prev_recall) + if recall_gap > 1e-6: + one_class_ap += precision[i] * recall_gap + prev_recall = recall[i] + mAP += one_class_ap + valid_cnt += 1 + else: + logger.error("Unspported mAP type {}".format(self.map_type)) + sys.exit(1) + eval_results.append({ + 'class': self.classes[valid_cnt - 1], + 'ap': one_class_ap, + 'precision': precision, + 'recall': recall, + }) + self.eval_results = eval_results + self.mAP = mAP / float(valid_cnt) if valid_cnt > 0 else mAP + + def get_map(self): + """ + Get mAP result + """ + if self.mAP is None: + logger.error("mAP is not calculated.") + if self.classwise: + # Compute per-category AP and PR curve + try: + from terminaltables import AsciiTable + except Exception as e: + logger.error( + 'terminaltables not found, plaese install terminaltables. ' + 'for example: `pip install terminaltables`.') + raise e + results_per_category = [] + for eval_result in self.eval_results: + results_per_category.append( + (str(eval_result['class']), + '{:0.3f}'.format(float(eval_result['ap'])))) + draw_pr_curve( + eval_result['precision'], + eval_result['recall'], + out_dir='voc_pr_curve', + file_name='{}_precision_recall_curve.jpg'.format( + eval_result['class'])) + + num_columns = min(6, len(results_per_category) * 2) + results_flatten = list(itertools.chain(*results_per_category)) + headers = ['category', 'AP'] * (num_columns // 2) + results_2d = itertools.zip_longest( + *[results_flatten[i::num_columns] for i in range(num_columns)]) + table_data = [headers] + table_data += [result for result in results_2d] + table = AsciiTable(table_data) + logger.info('Per-category of VOC AP: \n{}'.format(table.table)) + logger.info( + "per-category PR curve has output to voc_pr_curve folder.") + return self.mAP + + def _get_tp_fp_accum(self, score_pos_list): + """ + Calculate accumulating true/false positive results from + [score, pos] records + """ + sorted_list = sorted(score_pos_list, key=lambda s: s[0], reverse=True) + accum_tp = 0 + accum_fp = 0 + accum_tp_list = [] + accum_fp_list = [] + for (score, pos) in sorted_list: + accum_tp += int(pos) + accum_tp_list.append(accum_tp) + accum_fp += 1 - int(pos) + accum_fp_list.append(accum_fp) + return accum_tp_list, accum_fp_list + + +def ap_per_class(tp, conf, pred_cls, target_cls): + """ + Computes the average precision, given the recall and precision curves. + Method originally from https://github.com/rafaelpadilla/Object-Detection-Metrics. + + Args: + tp (list): True positives. + conf (list): Objectness value from 0-1. + pred_cls (list): Predicted object classes. + target_cls (list): Target object classes. + """ + tp, conf, pred_cls, target_cls = np.array(tp), np.array(conf), np.array( + pred_cls), np.array(target_cls) + + # Sort by objectness + i = np.argsort(-conf) + tp, conf, pred_cls = tp[i], conf[i], pred_cls[i] + + # Find unique classes + unique_classes = np.unique(np.concatenate((pred_cls, target_cls), 0)) + + # Create Precision-Recall curve and compute AP for each class + ap, p, r = [], [], [] + for c in unique_classes: + i = pred_cls == c + n_gt = sum(target_cls == c) # Number of ground truth objects + n_p = sum(i) # Number of predicted objects + + if (n_p == 0) and (n_gt == 0): + continue + elif (n_p == 0) or (n_gt == 0): + ap.append(0) + r.append(0) + p.append(0) + else: + # Accumulate FPs and TPs + fpc = np.cumsum(1 - tp[i]) + tpc = np.cumsum(tp[i]) + + # Recall + recall_curve = tpc / (n_gt + 1e-16) + r.append(tpc[-1] / (n_gt + 1e-16)) + + # Precision + precision_curve = tpc / (tpc + fpc) + p.append(tpc[-1] / (tpc[-1] + fpc[-1])) + + # AP from recall-precision curve + ap.append(compute_ap(recall_curve, precision_curve)) + + return np.array(ap), unique_classes.astype('int32'), np.array(r), np.array( + p) + + +def compute_ap(recall, precision): + """ + Computes the average precision, given the recall and precision curves. + Code originally from https://github.com/rbgirshick/py-faster-rcnn. + + Args: + recall (list): The recall curve. + precision (list): The precision curve. + + Returns: + The average precision as computed in py-faster-rcnn. + """ + # correct AP calculation + # first append sentinel values at the end + mrec = np.concatenate(([0.], recall, [1.])) + mpre = np.concatenate(([0.], precision, [0.])) + + # compute the precision envelope + for i in range(mpre.size - 1, 0, -1): + mpre[i - 1] = np.maximum(mpre[i - 1], mpre[i]) + + # to calculate area under PR curve, look for points + # where X axis (recall) changes value + i = np.where(mrec[1:] != mrec[:-1])[0] + + # and sum (\Delta recall) * prec + ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1]) + return ap diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mcmot_metrics.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mcmot_metrics.py new file mode 100644 index 000000000..9f329c8e0 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mcmot_metrics.py @@ -0,0 +1,467 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import copy +import sys +import math +from collections import defaultdict +from motmetrics.math_util import quiet_divide + +import numpy as np +import pandas as pd + +import paddle +import paddle.nn.functional as F +from .metrics import Metric +import motmetrics as mm +import openpyxl +metrics = mm.metrics.motchallenge_metrics +mh = mm.metrics.create() +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['MCMOTEvaluator', 'MCMOTMetric'] + +METRICS_LIST = [ + 'num_frames', 'num_matches', 'num_switches', 'num_transfer', 'num_ascend', + 'num_migrate', 'num_false_positives', 'num_misses', 'num_detections', + 'num_objects', 'num_predictions', 'num_unique_objects', 'mostly_tracked', + 'partially_tracked', 'mostly_lost', 'num_fragmentations', 'motp', 'mota', + 'precision', 'recall', 'idfp', 'idfn', 'idtp', 'idp', 'idr', 'idf1' +] + +NAME_MAP = { + 'num_frames': 'num_frames', + 'num_matches': 'num_matches', + 'num_switches': 'IDs', + 'num_transfer': 'IDt', + 'num_ascend': 'IDa', + 'num_migrate': 'IDm', + 'num_false_positives': 'FP', + 'num_misses': 'FN', + 'num_detections': 'num_detections', + 'num_objects': 'num_objects', + 'num_predictions': 'num_predictions', + 'num_unique_objects': 'GT', + 'mostly_tracked': 'MT', + 'partially_tracked': 'partially_tracked', + 'mostly_lost': 'ML', + 'num_fragmentations': 'FM', + 'motp': 'MOTP', + 'mota': 'MOTA', + 'precision': 'Prcn', + 'recall': 'Rcll', + 'idfp': 'idfp', + 'idfn': 'idfn', + 'idtp': 'idtp', + 'idp': 'IDP', + 'idr': 'IDR', + 'idf1': 'IDF1' +} + + +def parse_accs_metrics(seq_acc, index_name, verbose=False): + """ + Parse the evaluation indicators of multiple MOTAccumulator + """ + mh = mm.metrics.create() + summary = MCMOTEvaluator.get_summary(seq_acc, index_name, METRICS_LIST) + summary.loc['OVERALL', 'motp'] = (summary['motp'] * summary['num_detections']).sum() / \ + summary.loc['OVERALL', 'num_detections'] + if verbose: + strsummary = mm.io.render_summary( + summary, formatters=mh.formatters, namemap=NAME_MAP) + print(strsummary) + + return summary + + +def seqs_overall_metrics(summary_df, verbose=False): + """ + Calculate overall metrics for multiple sequences + """ + add_col = [ + 'num_frames', 'num_matches', 'num_switches', 'num_transfer', + 'num_ascend', 'num_migrate', 'num_false_positives', 'num_misses', + 'num_detections', 'num_objects', 'num_predictions', + 'num_unique_objects', 'mostly_tracked', 'partially_tracked', + 'mostly_lost', 'num_fragmentations', 'idfp', 'idfn', 'idtp' + ] + calc_col = ['motp', 'mota', 'precision', 'recall', 'idp', 'idr', 'idf1'] + calc_df = summary_df.copy() + + overall_dic = {} + for col in add_col: + overall_dic[col] = calc_df[col].sum() + + for col in calc_col: + overall_dic[col] = getattr(MCMOTMetricOverall, col + '_overall')( + calc_df, overall_dic) + + overall_df = pd.DataFrame(overall_dic, index=['overall_calc']) + calc_df = pd.concat([calc_df, overall_df]) + + if verbose: + mh = mm.metrics.create() + str_calc_df = mm.io.render_summary( + calc_df, formatters=mh.formatters, namemap=NAME_MAP) + print(str_calc_df) + + return calc_df + + +class MCMOTMetricOverall(object): + def motp_overall(summary_df, overall_dic): + motp = quiet_divide((summary_df['motp'] * + summary_df['num_detections']).sum(), + overall_dic['num_detections']) + return motp + + def mota_overall(summary_df, overall_dic): + del summary_df + mota = 1. - quiet_divide( + (overall_dic['num_misses'] + overall_dic['num_switches'] + + overall_dic['num_false_positives']), overall_dic['num_objects']) + return mota + + def precision_overall(summary_df, overall_dic): + del summary_df + precision = quiet_divide(overall_dic['num_detections'], ( + overall_dic['num_false_positives'] + overall_dic['num_detections'])) + return precision + + def recall_overall(summary_df, overall_dic): + del summary_df + recall = quiet_divide(overall_dic['num_detections'], + overall_dic['num_objects']) + return recall + + def idp_overall(summary_df, overall_dic): + del summary_df + idp = quiet_divide(overall_dic['idtp'], + (overall_dic['idtp'] + overall_dic['idfp'])) + return idp + + def idr_overall(summary_df, overall_dic): + del summary_df + idr = quiet_divide(overall_dic['idtp'], + (overall_dic['idtp'] + overall_dic['idfn'])) + return idr + + def idf1_overall(summary_df, overall_dic): + del summary_df + idf1 = quiet_divide(2. * overall_dic['idtp'], ( + overall_dic['num_objects'] + overall_dic['num_predictions'])) + return idf1 + + +def read_mcmot_results_union(filename, is_gt, is_ignore): + results_dict = dict() + if os.path.isfile(filename): + all_result = np.loadtxt(filename, delimiter=',') + if all_result.shape[0] == 0 or all_result.shape[1] < 7: + return results_dict + if is_ignore: + return results_dict + if is_gt: + # only for test use + all_result = all_result[all_result[:, 7] != 0] + all_result[:, 7] = all_result[:, 7] - 1 + + if all_result.shape[0] == 0: + return results_dict + + class_unique = np.unique(all_result[:, 7]) + + last_max_id = 0 + result_cls_list = [] + for cls in class_unique: + result_cls_split = all_result[all_result[:, 7] == cls] + result_cls_split[:, 1] = result_cls_split[:, 1] + last_max_id + # make sure track id different between every category + last_max_id = max(np.unique(result_cls_split[:, 1])) + 1 + result_cls_list.append(result_cls_split) + + results_con = np.concatenate(result_cls_list) + + for line in range(len(results_con)): + linelist = results_con[line] + fid = int(linelist[0]) + if fid < 1: + continue + results_dict.setdefault(fid, list()) + + if is_gt: + score = 1 + else: + score = float(linelist[6]) + + tlwh = tuple(map(float, linelist[2:6])) + target_id = int(linelist[1]) + cls = int(linelist[7]) + + results_dict[fid].append((tlwh, target_id, cls, score)) + + return results_dict + + +def read_mcmot_results(filename, is_gt, is_ignore): + results_dict = dict() + if os.path.isfile(filename): + with open(filename, 'r') as f: + for line in f.readlines(): + linelist = line.strip().split(',') + if len(linelist) < 7: + continue + fid = int(linelist[0]) + if fid < 1: + continue + cid = int(linelist[7]) + if is_gt: + score = 1 + # only for test use + cid -= 1 + else: + score = float(linelist[6]) + + cls_result_dict = results_dict.setdefault(cid, dict()) + cls_result_dict.setdefault(fid, list()) + + tlwh = tuple(map(float, linelist[2:6])) + target_id = int(linelist[1]) + cls_result_dict[fid].append((tlwh, target_id, score)) + return results_dict + + +def read_results(filename, + data_type, + is_gt=False, + is_ignore=False, + multi_class=False, + union=False): + if data_type in ['mcmot', 'lab']: + if multi_class: + if union: + # The results are evaluated by union all the categories. + # Track IDs between different categories cannot be duplicate. + read_fun = read_mcmot_results_union + else: + # The results are evaluated separately by category. + read_fun = read_mcmot_results + else: + raise ValueError('multi_class: {}, MCMOT should have cls_id.'. + format(multi_class)) + else: + raise ValueError('Unknown data type: {}'.format(data_type)) + + return read_fun(filename, is_gt, is_ignore) + + +def unzip_objs(objs): + if len(objs) > 0: + tlwhs, ids, scores = zip(*objs) + else: + tlwhs, ids, scores = [], [], [] + tlwhs = np.asarray(tlwhs, dtype=float).reshape(-1, 4) + return tlwhs, ids, scores + + +def unzip_objs_cls(objs): + if len(objs) > 0: + tlwhs, ids, cls, scores = zip(*objs) + else: + tlwhs, ids, cls, scores = [], [], [], [] + tlwhs = np.asarray(tlwhs, dtype=float).reshape(-1, 4) + ids = np.array(ids) + cls = np.array(cls) + scores = np.array(scores) + return tlwhs, ids, cls, scores + + +class MCMOTEvaluator(object): + def __init__(self, data_root, seq_name, data_type, num_classes): + self.data_root = data_root + self.seq_name = seq_name + self.data_type = data_type + self.num_classes = num_classes + + self.load_annotations() + self.reset_accumulator() + + self.class_accs = [] + + def load_annotations(self): + assert self.data_type == 'mcmot' + self.gt_filename = os.path.join(self.data_root, '../', '../', + 'sequences', + '{}.txt'.format(self.seq_name)) + + def reset_accumulator(self): + import motmetrics as mm + mm.lap.default_solver = 'lap' + self.acc = mm.MOTAccumulator(auto_id=True) + + def eval_frame_dict(self, trk_objs, gt_objs, rtn_events=False, union=False): + import motmetrics as mm + mm.lap.default_solver = 'lap' + if union: + trk_tlwhs, trk_ids, trk_cls = unzip_objs_cls(trk_objs)[:3] + gt_tlwhs, gt_ids, gt_cls = unzip_objs_cls(gt_objs)[:3] + + # get distance matrix + iou_distance = mm.distances.iou_matrix( + gt_tlwhs, trk_tlwhs, max_iou=0.5) + + # Set the distance between objects of different categories to nan + gt_cls_len = len(gt_cls) + trk_cls_len = len(trk_cls) + # When the number of GT or Trk is 0, iou_distance dimension is (0,0) + if gt_cls_len != 0 and trk_cls_len != 0: + gt_cls = gt_cls.reshape(gt_cls_len, 1) + gt_cls = np.repeat(gt_cls, trk_cls_len, axis=1) + trk_cls = trk_cls.reshape(1, trk_cls_len) + trk_cls = np.repeat(trk_cls, gt_cls_len, axis=0) + iou_distance = np.where(gt_cls == trk_cls, iou_distance, np.nan) + + else: + trk_tlwhs, trk_ids = unzip_objs(trk_objs)[:2] + gt_tlwhs, gt_ids = unzip_objs(gt_objs)[:2] + + # get distance matrix + iou_distance = mm.distances.iou_matrix( + gt_tlwhs, trk_tlwhs, max_iou=0.5) + + self.acc.update(gt_ids, trk_ids, iou_distance) + + if rtn_events and iou_distance.size > 0 and hasattr(self.acc, + 'mot_events'): + events = self.acc.mot_events # only supported by https://github.com/longcw/py-motmetrics + else: + events = None + return events + + def eval_file(self, result_filename): + # evaluation of each category + gt_frame_dict = read_results( + self.gt_filename, + self.data_type, + is_gt=True, + multi_class=True, + union=False) + result_frame_dict = read_results( + result_filename, + self.data_type, + is_gt=False, + multi_class=True, + union=False) + + for cid in range(self.num_classes): + self.reset_accumulator() + cls_result_frame_dict = result_frame_dict.setdefault(cid, dict()) + cls_gt_frame_dict = gt_frame_dict.setdefault(cid, dict()) + + # only labeled frames will be evaluated + frames = sorted(list(set(cls_gt_frame_dict.keys()))) + + for frame_id in frames: + trk_objs = cls_result_frame_dict.get(frame_id, []) + gt_objs = cls_gt_frame_dict.get(frame_id, []) + self.eval_frame_dict(trk_objs, gt_objs, rtn_events=False) + + self.class_accs.append(self.acc) + + return self.class_accs + + @staticmethod + def get_summary(accs, + names, + metrics=('mota', 'num_switches', 'idp', 'idr', 'idf1', + 'precision', 'recall')): + import motmetrics as mm + mm.lap.default_solver = 'lap' + + names = copy.deepcopy(names) + if metrics is None: + metrics = mm.metrics.motchallenge_metrics + metrics = copy.deepcopy(metrics) + + mh = mm.metrics.create() + summary = mh.compute_many( + accs, metrics=metrics, names=names, generate_overall=True) + + return summary + + @staticmethod + def save_summary(summary, filename): + import pandas as pd + writer = pd.ExcelWriter(filename) + summary.to_excel(writer) + writer.save() + + +class MCMOTMetric(Metric): + def __init__(self, num_classes, save_summary=False): + self.num_classes = num_classes + self.save_summary = save_summary + self.MCMOTEvaluator = MCMOTEvaluator + self.result_root = None + self.reset() + + self.seqs_overall = defaultdict(list) + + def reset(self): + self.accs = [] + self.seqs = [] + + def update(self, data_root, seq, data_type, result_root, result_filename): + evaluator = self.MCMOTEvaluator(data_root, seq, data_type, + self.num_classes) + seq_acc = evaluator.eval_file(result_filename) + self.accs.append(seq_acc) + self.seqs.append(seq) + self.result_root = result_root + + cls_index_name = [ + '{}_{}'.format(seq, i) for i in range(self.num_classes) + ] + summary = parse_accs_metrics(seq_acc, cls_index_name) + summary.rename( + index={'OVERALL': '{}_OVERALL'.format(seq)}, inplace=True) + for row in range(len(summary)): + self.seqs_overall[row].append(summary.iloc[row:row + 1]) + + def accumulate(self): + self.cls_summary_list = [] + for row in range(self.num_classes): + seqs_cls_df = pd.concat(self.seqs_overall[row]) + seqs_cls_summary = seqs_overall_metrics(seqs_cls_df) + cls_summary_overall = seqs_cls_summary.iloc[-1:].copy() + cls_summary_overall.rename( + index={'overall_calc': 'overall_calc_{}'.format(row)}, + inplace=True) + self.cls_summary_list.append(cls_summary_overall) + + def log(self): + seqs_summary = seqs_overall_metrics( + pd.concat(self.seqs_overall[self.num_classes]), verbose=True) + class_summary = seqs_overall_metrics( + pd.concat(self.cls_summary_list), verbose=True) + + def get_results(self): + return 1 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/metrics.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/metrics.py new file mode 100644 index 000000000..f9913b7fb --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/metrics.py @@ -0,0 +1,432 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys +import json +import paddle +import numpy as np + +from .map_utils import prune_zero_padding, DetectionMAP +from .coco_utils import get_infer_results, cocoapi_eval +from .widerface_utils import face_eval_run +from ppdet.data.source.category import get_categories + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = [ + 'Metric', + 'COCOMetric', + 'VOCMetric', + 'WiderFaceMetric', + 'get_infer_results', + 'RBoxMetric', + 'SNIPERCOCOMetric' +] + +COCO_SIGMAS = np.array([ + .26, .25, .25, .35, .35, .79, .79, .72, .72, .62, .62, 1.07, 1.07, .87, .87, + .89, .89 +]) / 10.0 +CROWD_SIGMAS = np.array( + [.79, .79, .72, .72, .62, .62, 1.07, 1.07, .87, .87, .89, .89, .79, + .79]) / 10.0 + + +class Metric(paddle.metric.Metric): + def name(self): + return self.__class__.__name__ + + def reset(self): + pass + + def accumulate(self): + pass + + # paddle.metric.Metric defined :metch:`update`, :meth:`accumulate` + # :metch:`reset`, in ppdet, we also need following 2 methods: + + # abstract method for logging metric results + def log(self): + pass + + # abstract method for getting metric results + def get_results(self): + pass + + +class COCOMetric(Metric): + def __init__(self, anno_file, **kwargs): + assert os.path.isfile(anno_file), \ + "anno_file {} not a file".format(anno_file) + self.anno_file = anno_file + self.clsid2catid = kwargs.get('clsid2catid', None) + if self.clsid2catid is None: + self.clsid2catid, _ = get_categories('COCO', anno_file) + self.classwise = kwargs.get('classwise', False) + self.output_eval = kwargs.get('output_eval', None) + # TODO: bias should be unified + self.bias = kwargs.get('bias', 0) + self.save_prediction_only = kwargs.get('save_prediction_only', False) + self.iou_type = kwargs.get('IouType', 'bbox') + self.reset() + + def reset(self): + # only bbox and mask evaluation support currently + self.results = {'bbox': [], 'mask': [], 'segm': [], 'keypoint': []} + self.eval_results = {} + + def update(self, inputs, outputs): + outs = {} + # outputs Tensor -> numpy.ndarray + for k, v in outputs.items(): + outs[k] = v.numpy() if isinstance(v, paddle.Tensor) else v + + im_id = inputs['im_id'] + outs['im_id'] = im_id.numpy() if isinstance(im_id, + paddle.Tensor) else im_id + + infer_results = get_infer_results( + outs, self.clsid2catid, bias=self.bias) + self.results['bbox'] += infer_results[ + 'bbox'] if 'bbox' in infer_results else [] + self.results['mask'] += infer_results[ + 'mask'] if 'mask' in infer_results else [] + self.results['segm'] += infer_results[ + 'segm'] if 'segm' in infer_results else [] + self.results['keypoint'] += infer_results[ + 'keypoint'] if 'keypoint' in infer_results else [] + + def accumulate(self): + if len(self.results['bbox']) > 0: + output = "bbox.json" + if self.output_eval: + output = os.path.join(self.output_eval, output) + with open(output, 'w') as f: + json.dump(self.results['bbox'], f) + logger.info('The bbox result is saved to bbox.json.') + + if self.save_prediction_only: + logger.info('The bbox result is saved to {} and do not ' + 'evaluate the mAP.'.format(output)) + else: + bbox_stats = cocoapi_eval( + output, + 'bbox', + anno_file=self.anno_file, + classwise=self.classwise) + self.eval_results['bbox'] = bbox_stats + sys.stdout.flush() + + if len(self.results['mask']) > 0: + output = "mask.json" + if self.output_eval: + output = os.path.join(self.output_eval, output) + with open(output, 'w') as f: + json.dump(self.results['mask'], f) + logger.info('The mask result is saved to mask.json.') + + if self.save_prediction_only: + logger.info('The mask result is saved to {} and do not ' + 'evaluate the mAP.'.format(output)) + else: + seg_stats = cocoapi_eval( + output, + 'segm', + anno_file=self.anno_file, + classwise=self.classwise) + self.eval_results['mask'] = seg_stats + sys.stdout.flush() + + if len(self.results['segm']) > 0: + output = "segm.json" + if self.output_eval: + output = os.path.join(self.output_eval, output) + with open(output, 'w') as f: + json.dump(self.results['segm'], f) + logger.info('The segm result is saved to segm.json.') + + if self.save_prediction_only: + logger.info('The segm result is saved to {} and do not ' + 'evaluate the mAP.'.format(output)) + else: + seg_stats = cocoapi_eval( + output, + 'segm', + anno_file=self.anno_file, + classwise=self.classwise) + self.eval_results['mask'] = seg_stats + sys.stdout.flush() + + if len(self.results['keypoint']) > 0: + output = "keypoint.json" + if self.output_eval: + output = os.path.join(self.output_eval, output) + with open(output, 'w') as f: + json.dump(self.results['keypoint'], f) + logger.info('The keypoint result is saved to keypoint.json.') + + if self.save_prediction_only: + logger.info('The keypoint result is saved to {} and do not ' + 'evaluate the mAP.'.format(output)) + else: + style = 'keypoints' + use_area = True + sigmas = COCO_SIGMAS + if self.iou_type == 'keypoints_crowd': + style = 'keypoints_crowd' + use_area = False + sigmas = CROWD_SIGMAS + keypoint_stats = cocoapi_eval( + output, + style, + anno_file=self.anno_file, + classwise=self.classwise, + sigmas=sigmas, + use_area=use_area) + self.eval_results['keypoint'] = keypoint_stats + sys.stdout.flush() + + def log(self): + pass + + def get_results(self): + return self.eval_results + + +class VOCMetric(Metric): + def __init__(self, + label_list, + class_num=20, + overlap_thresh=0.5, + map_type='11point', + is_bbox_normalized=False, + evaluate_difficult=False, + classwise=False): + assert os.path.isfile(label_list), \ + "label_list {} not a file".format(label_list) + self.clsid2catid, self.catid2name = get_categories('VOC', label_list) + + self.overlap_thresh = overlap_thresh + self.map_type = map_type + self.evaluate_difficult = evaluate_difficult + self.detection_map = DetectionMAP( + class_num=class_num, + overlap_thresh=overlap_thresh, + map_type=map_type, + is_bbox_normalized=is_bbox_normalized, + evaluate_difficult=evaluate_difficult, + catid2name=self.catid2name, + classwise=classwise) + + self.reset() + + def reset(self): + self.detection_map.reset() + + def update(self, inputs, outputs): + bbox_np = outputs['bbox'].numpy() + bboxes = bbox_np[:, 2:] + scores = bbox_np[:, 1] + labels = bbox_np[:, 0] + bbox_lengths = outputs['bbox_num'].numpy() + + if bboxes.shape == (1, 1) or bboxes is None: + return + gt_boxes = inputs['gt_bbox'] + gt_labels = inputs['gt_class'] + difficults = inputs['difficult'] if not self.evaluate_difficult \ + else None + + scale_factor = inputs['scale_factor'].numpy( + ) if 'scale_factor' in inputs else np.ones( + (gt_boxes.shape[0], 2)).astype('float32') + + bbox_idx = 0 + for i in range(len(gt_boxes)): + gt_box = gt_boxes[i].numpy() + h, w = scale_factor[i] + gt_box = gt_box / np.array([w, h, w, h]) + gt_label = gt_labels[i].numpy() + difficult = None if difficults is None \ + else difficults[i].numpy() + bbox_num = bbox_lengths[i] + bbox = bboxes[bbox_idx:bbox_idx + bbox_num] + score = scores[bbox_idx:bbox_idx + bbox_num] + label = labels[bbox_idx:bbox_idx + bbox_num] + gt_box, gt_label, difficult = prune_zero_padding(gt_box, gt_label, + difficult) + self.detection_map.update(bbox, score, label, gt_box, gt_label, + difficult) + bbox_idx += bbox_num + + def accumulate(self): + logger.info("Accumulating evaluatation results...") + self.detection_map.accumulate() + + def log(self): + map_stat = 100. * self.detection_map.get_map() + logger.info("mAP({:.2f}, {}) = {:.2f}%".format(self.overlap_thresh, + self.map_type, map_stat)) + + def get_results(self): + return {'bbox': [self.detection_map.get_map()]} + + +class WiderFaceMetric(Metric): + def __init__(self, image_dir, anno_file, multi_scale=True): + self.image_dir = image_dir + self.anno_file = anno_file + self.multi_scale = multi_scale + self.clsid2catid, self.catid2name = get_categories('widerface') + + def update(self, model): + + face_eval_run( + model, + self.image_dir, + self.anno_file, + pred_dir='output/pred', + eval_mode='widerface', + multi_scale=self.multi_scale) + + +class RBoxMetric(Metric): + def __init__(self, anno_file, **kwargs): + assert os.path.isfile(anno_file), \ + "anno_file {} not a file".format(anno_file) + assert os.path.exists(anno_file), "anno_file {} not exists".format( + anno_file) + self.anno_file = anno_file + self.gt_anno = json.load(open(self.anno_file)) + cats = self.gt_anno['categories'] + self.clsid2catid = {i: cat['id'] for i, cat in enumerate(cats)} + self.catid2clsid = {cat['id']: i for i, cat in enumerate(cats)} + self.catid2name = {cat['id']: cat['name'] for cat in cats} + self.classwise = kwargs.get('classwise', False) + self.output_eval = kwargs.get('output_eval', None) + # TODO: bias should be unified + self.bias = kwargs.get('bias', 0) + self.save_prediction_only = kwargs.get('save_prediction_only', False) + self.iou_type = kwargs.get('IouType', 'bbox') + self.overlap_thresh = kwargs.get('overlap_thresh', 0.5) + self.map_type = kwargs.get('map_type', '11point') + self.evaluate_difficult = kwargs.get('evaluate_difficult', False) + class_num = len(self.catid2name) + self.detection_map = DetectionMAP( + class_num=class_num, + overlap_thresh=self.overlap_thresh, + map_type=self.map_type, + is_bbox_normalized=False, + evaluate_difficult=self.evaluate_difficult, + catid2name=self.catid2name, + classwise=self.classwise) + + self.reset() + + def reset(self): + self.result_bbox = [] + self.detection_map.reset() + + def update(self, inputs, outputs): + outs = {} + # outputs Tensor -> numpy.ndarray + for k, v in outputs.items(): + outs[k] = v.numpy() if isinstance(v, paddle.Tensor) else v + + im_id = inputs['im_id'] + outs['im_id'] = im_id.numpy() if isinstance(im_id, + paddle.Tensor) else im_id + + infer_results = get_infer_results( + outs, self.clsid2catid, bias=self.bias) + self.result_bbox += infer_results[ + 'bbox'] if 'bbox' in infer_results else [] + bbox = [b['bbox'] for b in self.result_bbox] + score = [b['score'] for b in self.result_bbox] + label = [b['category_id'] for b in self.result_bbox] + label = [self.catid2clsid[e] for e in label] + gt_box = [ + e['bbox'] for e in self.gt_anno['annotations'] + if e['image_id'] == outs['im_id'] + ] + gt_label = [ + e['category_id'] for e in self.gt_anno['annotations'] + if e['image_id'] == outs['im_id'] + ] + gt_label = [self.catid2clsid[e] for e in gt_label] + self.detection_map.update(bbox, score, label, gt_box, gt_label) + + def accumulate(self): + if len(self.result_bbox) > 0: + output = "bbox.json" + if self.output_eval: + output = os.path.join(self.output_eval, output) + with open(output, 'w') as f: + json.dump(self.result_bbox, f) + logger.info('The bbox result is saved to bbox.json.') + + if self.save_prediction_only: + logger.info('The bbox result is saved to {} and do not ' + 'evaluate the mAP.'.format(output)) + else: + logger.info("Accumulating evaluatation results...") + self.detection_map.accumulate() + + def log(self): + map_stat = 100. * self.detection_map.get_map() + logger.info("mAP({:.2f}, {}) = {:.2f}%".format(self.overlap_thresh, + self.map_type, map_stat)) + + def get_results(self): + return {'bbox': [self.detection_map.get_map()]} + + +class SNIPERCOCOMetric(COCOMetric): + def __init__(self, anno_file, **kwargs): + super(SNIPERCOCOMetric, self).__init__(anno_file, **kwargs) + self.dataset = kwargs["dataset"] + self.chip_results = [] + + def reset(self): + # only bbox and mask evaluation support currently + self.results = {'bbox': [], 'mask': [], 'segm': [], 'keypoint': []} + self.eval_results = {} + self.chip_results = [] + + def update(self, inputs, outputs): + outs = {} + # outputs Tensor -> numpy.ndarray + for k, v in outputs.items(): + outs[k] = v.numpy() if isinstance(v, paddle.Tensor) else v + + im_id = inputs['im_id'] + outs['im_id'] = im_id.numpy() if isinstance(im_id, + paddle.Tensor) else im_id + + self.chip_results.append(outs) + + + def accumulate(self): + results = self.dataset.anno_cropper.aggregate_chips_detections(self.chip_results) + for outs in results: + infer_results = get_infer_results(outs, self.clsid2catid, bias=self.bias) + self.results['bbox'] += infer_results['bbox'] if 'bbox' in infer_results else [] + + super(SNIPERCOCOMetric, self).accumulate() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mot_metrics.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mot_metrics.py new file mode 100644 index 000000000..85cba3630 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/mot_metrics.py @@ -0,0 +1,1232 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import copy +import sys +import math +from collections import defaultdict +import numpy as np +import paddle +import paddle.nn.functional as F +from ppdet.modeling.bbox_utils import bbox_iou_np_expand +from .map_utils import ap_per_class +from .metrics import Metric +from .munkres import Munkres + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['MOTEvaluator', 'MOTMetric', 'JDEDetMetric', 'KITTIMOTMetric'] + + +def read_mot_results(filename, is_gt=False, is_ignore=False): + valid_labels = {1} + ignore_labels = {2, 7, 8, 12} # only in motchallenge datasets like 'MOT16' + results_dict = dict() + if os.path.isfile(filename): + with open(filename, 'r') as f: + for line in f.readlines(): + linelist = line.split(',') + if len(linelist) < 7: + continue + fid = int(linelist[0]) + if fid < 1: + continue + results_dict.setdefault(fid, list()) + + box_size = float(linelist[4]) * float(linelist[5]) + + if is_gt: + label = int(float(linelist[7])) + mark = int(float(linelist[6])) + if mark == 0 or label not in valid_labels: + continue + score = 1 + elif is_ignore: + if 'MOT16-' in filename or 'MOT17-' in filename or 'MOT15-' in filename or 'MOT20-' in filename: + label = int(float(linelist[7])) + vis_ratio = float(linelist[8]) + if label not in ignore_labels and vis_ratio >= 0: + continue + else: + continue + score = 1 + else: + score = float(linelist[6]) + + tlwh = tuple(map(float, linelist[2:6])) + target_id = int(linelist[1]) + + results_dict[fid].append((tlwh, target_id, score)) + return results_dict + + +""" +MOT dataset label list, see in https://motchallenge.net +labels={'ped', ... % 1 + 'person_on_vhcl', ... % 2 + 'car', ... % 3 + 'bicycle', ... % 4 + 'mbike', ... % 5 + 'non_mot_vhcl', ... % 6 + 'static_person', ... % 7 + 'distractor', ... % 8 + 'occluder', ... % 9 + 'occluder_on_grnd', ... % 10 + 'occluder_full', ... % 11 + 'reflection', ... % 12 + 'crowd' ... % 13 +}; +""" + + +def unzip_objs(objs): + if len(objs) > 0: + tlwhs, ids, scores = zip(*objs) + else: + tlwhs, ids, scores = [], [], [] + tlwhs = np.asarray(tlwhs, dtype=float).reshape(-1, 4) + return tlwhs, ids, scores + + +class MOTEvaluator(object): + def __init__(self, data_root, seq_name, data_type): + self.data_root = data_root + self.seq_name = seq_name + self.data_type = data_type + + self.load_annotations() + self.reset_accumulator() + + def load_annotations(self): + assert self.data_type == 'mot' + gt_filename = os.path.join(self.data_root, self.seq_name, 'gt', + 'gt.txt') + self.gt_frame_dict = read_mot_results(gt_filename, is_gt=True) + self.gt_ignore_frame_dict = read_mot_results( + gt_filename, is_ignore=True) + + def reset_accumulator(self): + import motmetrics as mm + mm.lap.default_solver = 'lap' + self.acc = mm.MOTAccumulator(auto_id=True) + + def eval_frame(self, frame_id, trk_tlwhs, trk_ids, rtn_events=False): + import motmetrics as mm + mm.lap.default_solver = 'lap' + # results + trk_tlwhs = np.copy(trk_tlwhs) + trk_ids = np.copy(trk_ids) + + # gts + gt_objs = self.gt_frame_dict.get(frame_id, []) + gt_tlwhs, gt_ids = unzip_objs(gt_objs)[:2] + + # ignore boxes + ignore_objs = self.gt_ignore_frame_dict.get(frame_id, []) + ignore_tlwhs = unzip_objs(ignore_objs)[0] + + # remove ignored results + keep = np.ones(len(trk_tlwhs), dtype=bool) + iou_distance = mm.distances.iou_matrix( + ignore_tlwhs, trk_tlwhs, max_iou=0.5) + if len(iou_distance) > 0: + match_is, match_js = mm.lap.linear_sum_assignment(iou_distance) + match_is, match_js = map(lambda a: np.asarray(a, dtype=int), [match_is, match_js]) + match_ious = iou_distance[match_is, match_js] + + match_js = np.asarray(match_js, dtype=int) + match_js = match_js[np.logical_not(np.isnan(match_ious))] + keep[match_js] = False + trk_tlwhs = trk_tlwhs[keep] + trk_ids = trk_ids[keep] + + # get distance matrix + iou_distance = mm.distances.iou_matrix(gt_tlwhs, trk_tlwhs, max_iou=0.5) + + # acc + self.acc.update(gt_ids, trk_ids, iou_distance) + + if rtn_events and iou_distance.size > 0 and hasattr(self.acc, + 'last_mot_events'): + events = self.acc.last_mot_events # only supported by https://github.com/longcw/py-motmetrics + else: + events = None + return events + + def eval_file(self, filename): + self.reset_accumulator() + + result_frame_dict = read_mot_results(filename, is_gt=False) + frames = sorted(list(set(result_frame_dict.keys()))) + for frame_id in frames: + trk_objs = result_frame_dict.get(frame_id, []) + trk_tlwhs, trk_ids = unzip_objs(trk_objs)[:2] + self.eval_frame(frame_id, trk_tlwhs, trk_ids, rtn_events=False) + + return self.acc + + @staticmethod + def get_summary(accs, + names, + metrics=('mota', 'num_switches', 'idp', 'idr', 'idf1', + 'precision', 'recall')): + import motmetrics as mm + mm.lap.default_solver = 'lap' + names = copy.deepcopy(names) + if metrics is None: + metrics = mm.metrics.motchallenge_metrics + metrics = copy.deepcopy(metrics) + + mh = mm.metrics.create() + summary = mh.compute_many( + accs, metrics=metrics, names=names, generate_overall=True) + return summary + + @staticmethod + def save_summary(summary, filename): + import pandas as pd + writer = pd.ExcelWriter(filename) + summary.to_excel(writer) + writer.save() + + +class MOTMetric(Metric): + def __init__(self, save_summary=False): + self.save_summary = save_summary + self.MOTEvaluator = MOTEvaluator + self.result_root = None + self.reset() + + def reset(self): + self.accs = [] + self.seqs = [] + + def update(self, data_root, seq, data_type, result_root, result_filename): + evaluator = self.MOTEvaluator(data_root, seq, data_type) + self.accs.append(evaluator.eval_file(result_filename)) + self.seqs.append(seq) + self.result_root = result_root + + def accumulate(self): + import motmetrics as mm + import openpyxl + metrics = mm.metrics.motchallenge_metrics + mh = mm.metrics.create() + summary = self.MOTEvaluator.get_summary(self.accs, self.seqs, metrics) + self.strsummary = mm.io.render_summary( + summary, + formatters=mh.formatters, + namemap=mm.io.motchallenge_metric_names) + if self.save_summary: + self.MOTEvaluator.save_summary( + summary, os.path.join(self.result_root, 'summary.xlsx')) + + def log(self): + print(self.strsummary) + + def get_results(self): + return self.strsummary + + +class JDEDetMetric(Metric): + # Note this detection AP metric is different from COCOMetric or VOCMetric, + # and the bboxes coordinates are not scaled to the original image + def __init__(self, overlap_thresh=0.5): + self.overlap_thresh = overlap_thresh + self.reset() + + def reset(self): + self.AP_accum = np.zeros(1) + self.AP_accum_count = np.zeros(1) + + def update(self, inputs, outputs): + bboxes = outputs['bbox'][:, 2:].numpy() + scores = outputs['bbox'][:, 1].numpy() + labels = outputs['bbox'][:, 0].numpy() + bbox_lengths = outputs['bbox_num'].numpy() + if bboxes.shape[0] == 1 and bboxes.sum() == 0.0: + return + + gt_boxes = inputs['gt_bbox'].numpy()[0] + gt_labels = inputs['gt_class'].numpy()[0] + if gt_labels.shape[0] == 0: + return + + correct = [] + detected = [] + for i in range(bboxes.shape[0]): + obj_pred = 0 + pred_bbox = bboxes[i].reshape(1, 4) + # Compute iou with target boxes + iou = bbox_iou_np_expand(pred_bbox, gt_boxes, x1y1x2y2=True)[0] + # Extract index of largest overlap + best_i = np.argmax(iou) + # If overlap exceeds threshold and classification is correct mark as correct + if iou[best_i] > self.overlap_thresh and obj_pred == gt_labels[ + best_i] and best_i not in detected: + correct.append(1) + detected.append(best_i) + else: + correct.append(0) + + # Compute Average Precision (AP) per class + target_cls = list(gt_labels.T[0]) + AP, AP_class, R, P = ap_per_class( + tp=correct, + conf=scores, + pred_cls=np.zeros_like(scores), + target_cls=target_cls) + self.AP_accum_count += np.bincount(AP_class, minlength=1) + self.AP_accum += np.bincount(AP_class, minlength=1, weights=AP) + + def accumulate(self): + logger.info("Accumulating evaluatation results...") + self.map_stat = self.AP_accum[0] / (self.AP_accum_count[0] + 1E-16) + + def log(self): + map_stat = 100. * self.map_stat + logger.info("mAP({:.2f}) = {:.2f}%".format(self.overlap_thresh, + map_stat)) + + def get_results(self): + return self.map_stat + + +""" +Following code is borrow from https://github.com/xingyizhou/CenterTrack/blob/master/src/tools/eval_kitti_track/evaluate_tracking.py +""" + + +class tData: + """ + Utility class to load data. + """ + def __init__(self,frame=-1,obj_type="unset",truncation=-1,occlusion=-1,\ + obs_angle=-10,x1=-1,y1=-1,x2=-1,y2=-1,w=-1,h=-1,l=-1,\ + X=-1000,Y=-1000,Z=-1000,yaw=-10,score=-1000,track_id=-1): + """ + Constructor, initializes the object given the parameters. + """ + self.frame = frame + self.track_id = track_id + self.obj_type = obj_type + self.truncation = truncation + self.occlusion = occlusion + self.obs_angle = obs_angle + self.x1 = x1 + self.y1 = y1 + self.x2 = x2 + self.y2 = y2 + self.w = w + self.h = h + self.l = l + self.X = X + self.Y = Y + self.Z = Z + self.yaw = yaw + self.score = score + self.ignored = False + self.valid = False + self.tracker = -1 + + def __str__(self): + attrs = vars(self) + return '\n'.join("%s: %s" % item for item in attrs.items()) + + +class KITTIEvaluation(object): + """ KITTI tracking statistics (CLEAR MOT, id-switches, fragments, ML/PT/MT, precision/recall) + MOTA - Multi-object tracking accuracy in [0,100] + MOTP - Multi-object tracking precision in [0,100] (3D) / [td,100] (2D) + MOTAL - Multi-object tracking accuracy in [0,100] with log10(id-switches) + + id-switches - number of id switches + fragments - number of fragmentations + + MT, PT, ML - number of mostly tracked, partially tracked and mostly lost trajectories + + recall - recall = percentage of detected targets + precision - precision = percentage of correctly detected targets + FAR - number of false alarms per frame + falsepositives - number of false positives (FP) + missed - number of missed targets (FN) + """ + def __init__(self, result_path, gt_path, min_overlap=0.5, max_truncation = 0,\ + min_height = 25, max_occlusion = 2, cls="car",\ + n_frames=[], seqs=[], n_sequences=0): + # get number of sequences and + # get number of frames per sequence from test mapping + # (created while extracting the benchmark) + self.gt_path = os.path.join(gt_path, "../labels") + self.n_frames = n_frames + self.sequence_name = seqs + self.n_sequences = n_sequences + + self.cls = cls # class to evaluate, i.e. pedestrian or car + + self.result_path = result_path + + # statistics and numbers for evaluation + self.n_gt = 0 # number of ground truth detections minus ignored false negatives and true positives + self.n_igt = 0 # number of ignored ground truth detections + self.n_gts = [ + ] # number of ground truth detections minus ignored false negatives and true positives PER SEQUENCE + self.n_igts = [ + ] # number of ground ignored truth detections PER SEQUENCE + self.n_gt_trajectories = 0 + self.n_gt_seq = [] + self.n_tr = 0 # number of tracker detections minus ignored tracker detections + self.n_trs = [ + ] # number of tracker detections minus ignored tracker detections PER SEQUENCE + self.n_itr = 0 # number of ignored tracker detections + self.n_itrs = [] # number of ignored tracker detections PER SEQUENCE + self.n_igttr = 0 # number of ignored ground truth detections where the corresponding associated tracker detection is also ignored + self.n_tr_trajectories = 0 + self.n_tr_seq = [] + self.MOTA = 0 + self.MOTP = 0 + self.MOTAL = 0 + self.MODA = 0 + self.MODP = 0 + self.MODP_t = [] + self.recall = 0 + self.precision = 0 + self.F1 = 0 + self.FAR = 0 + self.total_cost = 0 + self.itp = 0 # number of ignored true positives + self.itps = [] # number of ignored true positives PER SEQUENCE + self.tp = 0 # number of true positives including ignored true positives! + self.tps = [ + ] # number of true positives including ignored true positives PER SEQUENCE + self.fn = 0 # number of false negatives WITHOUT ignored false negatives + self.fns = [ + ] # number of false negatives WITHOUT ignored false negatives PER SEQUENCE + self.ifn = 0 # number of ignored false negatives + self.ifns = [] # number of ignored false negatives PER SEQUENCE + self.fp = 0 # number of false positives + # a bit tricky, the number of ignored false negatives and ignored true positives + # is subtracted, but if both tracker detection and ground truth detection + # are ignored this number is added again to avoid double counting + self.fps = [] # above PER SEQUENCE + self.mme = 0 + self.fragments = 0 + self.id_switches = 0 + self.MT = 0 + self.PT = 0 + self.ML = 0 + + self.min_overlap = min_overlap # minimum bounding box overlap for 3rd party metrics + self.max_truncation = max_truncation # maximum truncation of an object for evaluation + self.max_occlusion = max_occlusion # maximum occlusion of an object for evaluation + self.min_height = min_height # minimum height of an object for evaluation + self.n_sample_points = 500 + + # this should be enough to hold all groundtruth trajectories + # is expanded if necessary and reduced in any case + self.gt_trajectories = [[] for x in range(self.n_sequences)] + self.ign_trajectories = [[] for x in range(self.n_sequences)] + + def loadGroundtruth(self): + try: + self._loadData(self.gt_path, cls=self.cls, loading_groundtruth=True) + except IOError: + return False + return True + + def loadTracker(self): + try: + if not self._loadData( + self.result_path, cls=self.cls, loading_groundtruth=False): + return False + except IOError: + return False + return True + + def _loadData(self, + root_dir, + cls, + min_score=-1000, + loading_groundtruth=False): + """ + Generic loader for ground truth and tracking data. + Use loadGroundtruth() or loadTracker() to load this data. + Loads detections in KITTI format from textfiles. + """ + # construct objectDetections object to hold detection data + t_data = tData() + data = [] + eval_2d = True + eval_3d = True + + seq_data = [] + n_trajectories = 0 + n_trajectories_seq = [] + for seq, s_name in enumerate(self.sequence_name): + i = 0 + filename = os.path.join(root_dir, "%s.txt" % s_name) + f = open(filename, "r") + + f_data = [ + [] for x in range(self.n_frames[seq]) + ] # current set has only 1059 entries, sufficient length is checked anyway + ids = [] + n_in_seq = 0 + id_frame_cache = [] + for line in f: + # KITTI tracking benchmark data format: + # (frame,tracklet_id,objectType,truncation,occlusion,alpha,x1,y1,x2,y2,h,w,l,X,Y,Z,ry) + line = line.strip() + fields = line.split(" ") + # classes that should be loaded (ignored neighboring classes) + if "car" in cls.lower(): + classes = ["car", "van"] + elif "pedestrian" in cls.lower(): + classes = ["pedestrian", "person_sitting"] + else: + classes = [cls.lower()] + classes += ["dontcare"] + if not any([s for s in classes if s in fields[2].lower()]): + continue + # get fields from table + t_data.frame = int(float(fields[0])) # frame + t_data.track_id = int(float(fields[1])) # id + t_data.obj_type = fields[ + 2].lower() # object type [car, pedestrian, cyclist, ...] + t_data.truncation = int( + float(fields[3])) # truncation [-1,0,1,2] + t_data.occlusion = int( + float(fields[4])) # occlusion [-1,0,1,2] + t_data.obs_angle = float(fields[5]) # observation angle [rad] + t_data.x1 = float(fields[6]) # left [px] + t_data.y1 = float(fields[7]) # top [px] + t_data.x2 = float(fields[8]) # right [px] + t_data.y2 = float(fields[9]) # bottom [px] + t_data.h = float(fields[10]) # height [m] + t_data.w = float(fields[11]) # width [m] + t_data.l = float(fields[12]) # length [m] + t_data.X = float(fields[13]) # X [m] + t_data.Y = float(fields[14]) # Y [m] + t_data.Z = float(fields[15]) # Z [m] + t_data.yaw = float(fields[16]) # yaw angle [rad] + if not loading_groundtruth: + if len(fields) == 17: + t_data.score = -1 + elif len(fields) == 18: + t_data.score = float(fields[17]) # detection score + else: + logger.info("file is not in KITTI format") + return + + # do not consider objects marked as invalid + if t_data.track_id is -1 and t_data.obj_type != "dontcare": + continue + + idx = t_data.frame + # check if length for frame data is sufficient + if idx >= len(f_data): + print("extend f_data", idx, len(f_data)) + f_data += [[] for x in range(max(500, idx - len(f_data)))] + try: + id_frame = (t_data.frame, t_data.track_id) + if id_frame in id_frame_cache and not loading_groundtruth: + logger.info( + "track ids are not unique for sequence %d: frame %d" + % (seq, t_data.frame)) + logger.info( + "track id %d occured at least twice for this frame" + % t_data.track_id) + logger.info("Exiting...") + #continue # this allows to evaluate non-unique result files + return False + id_frame_cache.append(id_frame) + f_data[t_data.frame].append(copy.copy(t_data)) + except: + print(len(f_data), idx) + raise + + if t_data.track_id not in ids and t_data.obj_type != "dontcare": + ids.append(t_data.track_id) + n_trajectories += 1 + n_in_seq += 1 + + # check if uploaded data provides information for 2D and 3D evaluation + if not loading_groundtruth and eval_2d is True and ( + t_data.x1 == -1 or t_data.x2 == -1 or t_data.y1 == -1 or + t_data.y2 == -1): + eval_2d = False + if not loading_groundtruth and eval_3d is True and ( + t_data.X == -1000 or t_data.Y == -1000 or + t_data.Z == -1000): + eval_3d = False + + # only add existing frames + n_trajectories_seq.append(n_in_seq) + seq_data.append(f_data) + f.close() + + if not loading_groundtruth: + self.tracker = seq_data + self.n_tr_trajectories = n_trajectories + self.eval_2d = eval_2d + self.eval_3d = eval_3d + self.n_tr_seq = n_trajectories_seq + if self.n_tr_trajectories == 0: + return False + else: + # split ground truth and DontCare areas + self.dcareas = [] + self.groundtruth = [] + for seq_idx in range(len(seq_data)): + seq_gt = seq_data[seq_idx] + s_g, s_dc = [], [] + for f in range(len(seq_gt)): + all_gt = seq_gt[f] + g, dc = [], [] + for gg in all_gt: + if gg.obj_type == "dontcare": + dc.append(gg) + else: + g.append(gg) + s_g.append(g) + s_dc.append(dc) + self.dcareas.append(s_dc) + self.groundtruth.append(s_g) + self.n_gt_seq = n_trajectories_seq + self.n_gt_trajectories = n_trajectories + return True + + def boxoverlap(self, a, b, criterion="union"): + """ + boxoverlap computes intersection over union for bbox a and b in KITTI format. + If the criterion is 'union', overlap = (a inter b) / a union b). + If the criterion is 'a', overlap = (a inter b) / a, where b should be a dontcare area. + """ + x1 = max(a.x1, b.x1) + y1 = max(a.y1, b.y1) + x2 = min(a.x2, b.x2) + y2 = min(a.y2, b.y2) + + w = x2 - x1 + h = y2 - y1 + + if w <= 0. or h <= 0.: + return 0. + inter = w * h + aarea = (a.x2 - a.x1) * (a.y2 - a.y1) + barea = (b.x2 - b.x1) * (b.y2 - b.y1) + # intersection over union overlap + if criterion.lower() == "union": + o = inter / float(aarea + barea - inter) + elif criterion.lower() == "a": + o = float(inter) / float(aarea) + else: + raise TypeError("Unkown type for criterion") + return o + + def compute3rdPartyMetrics(self): + """ + Computes the metrics defined in + - Stiefelhagen 2008: Evaluating Multiple Object Tracking Performance: The CLEAR MOT Metrics + MOTA, MOTAL, MOTP + - Nevatia 2008: Global Data Association for Multi-Object Tracking Using Network Flows + MT/PT/ML + """ + # construct Munkres object for Hungarian Method association + hm = Munkres() + max_cost = 1e9 + + # go through all frames and associate ground truth and tracker results + # groundtruth and tracker contain lists for every single frame containing lists of KITTI format detections + fr, ids = 0, 0 + for seq_idx in range(len(self.groundtruth)): + seq_gt = self.groundtruth[seq_idx] + seq_dc = self.dcareas[seq_idx] # don't care areas + seq_tracker = self.tracker[seq_idx] + seq_trajectories = defaultdict(list) + seq_ignored = defaultdict(list) + + # statistics over the current sequence, check the corresponding + # variable comments in __init__ to get their meaning + seqtp = 0 + seqitp = 0 + seqfn = 0 + seqifn = 0 + seqfp = 0 + seqigt = 0 + seqitr = 0 + + last_ids = [[], []] + n_gts = 0 + n_trs = 0 + + for f in range(len(seq_gt)): + g = seq_gt[f] + dc = seq_dc[f] + + t = seq_tracker[f] + # counting total number of ground truth and tracker objects + self.n_gt += len(g) + self.n_tr += len(t) + + n_gts += len(g) + n_trs += len(t) + + # use hungarian method to associate, using boxoverlap 0..1 as cost + # build cost matrix + cost_matrix = [] + this_ids = [[], []] + for gg in g: + # save current ids + this_ids[0].append(gg.track_id) + this_ids[1].append(-1) + gg.tracker = -1 + gg.id_switch = 0 + gg.fragmentation = 0 + cost_row = [] + for tt in t: + # overlap == 1 is cost ==0 + c = 1 - self.boxoverlap(gg, tt) + # gating for boxoverlap + if c <= self.min_overlap: + cost_row.append(c) + else: + cost_row.append(max_cost) # = 1e9 + cost_matrix.append(cost_row) + # all ground truth trajectories are initially not associated + # extend groundtruth trajectories lists (merge lists) + seq_trajectories[gg.track_id].append(-1) + seq_ignored[gg.track_id].append(False) + + if len(g) is 0: + cost_matrix = [[]] + # associate + association_matrix = hm.compute(cost_matrix) + + # tmp variables for sanity checks and MODP computation + tmptp = 0 + tmpfp = 0 + tmpfn = 0 + tmpc = 0 # this will sum up the overlaps for all true positives + tmpcs = [0] * len( + g) # this will save the overlaps for all true positives + # the reason is that some true positives might be ignored + # later such that the corrsponding overlaps can + # be subtracted from tmpc for MODP computation + + # mapping for tracker ids and ground truth ids + for row, col in association_matrix: + # apply gating on boxoverlap + c = cost_matrix[row][col] + if c < max_cost: + g[row].tracker = t[col].track_id + this_ids[1][row] = t[col].track_id + t[col].valid = True + g[row].distance = c + self.total_cost += 1 - c + tmpc += 1 - c + tmpcs[row] = 1 - c + seq_trajectories[g[row].track_id][-1] = t[col].track_id + + # true positives are only valid associations + self.tp += 1 + tmptp += 1 + else: + g[row].tracker = -1 + self.fn += 1 + tmpfn += 1 + + # associate tracker and DontCare areas + # ignore tracker in neighboring classes + nignoredtracker = 0 # number of ignored tracker detections + ignoredtrackers = dict() # will associate the track_id with -1 + # if it is not ignored and 1 if it is + # ignored; + # this is used to avoid double counting ignored + # cases, see the next loop + + for tt in t: + ignoredtrackers[tt.track_id] = -1 + # ignore detection if it belongs to a neighboring class or is + # smaller or equal to the minimum height + + tt_height = abs(tt.y1 - tt.y2) + if ((self.cls == "car" and tt.obj_type == "van") or + (self.cls == "pedestrian" and + tt.obj_type == "person_sitting") or + tt_height <= self.min_height) and not tt.valid: + nignoredtracker += 1 + tt.ignored = True + ignoredtrackers[tt.track_id] = 1 + continue + for d in dc: + overlap = self.boxoverlap(tt, d, "a") + if overlap > 0.5 and not tt.valid: + tt.ignored = True + nignoredtracker += 1 + ignoredtrackers[tt.track_id] = 1 + break + + # check for ignored FN/TP (truncation or neighboring object class) + ignoredfn = 0 # the number of ignored false negatives + nignoredtp = 0 # the number of ignored true positives + nignoredpairs = 0 # the number of ignored pairs, i.e. a true positive + # which is ignored but where the associated tracker + # detection has already been ignored + + gi = 0 + for gg in g: + if gg.tracker < 0: + if gg.occlusion>self.max_occlusion or gg.truncation>self.max_truncation\ + or (self.cls=="car" and gg.obj_type=="van") or (self.cls=="pedestrian" and gg.obj_type=="person_sitting"): + seq_ignored[gg.track_id][-1] = True + gg.ignored = True + ignoredfn += 1 + + elif gg.tracker >= 0: + if gg.occlusion>self.max_occlusion or gg.truncation>self.max_truncation\ + or (self.cls=="car" and gg.obj_type=="van") or (self.cls=="pedestrian" and gg.obj_type=="person_sitting"): + + seq_ignored[gg.track_id][-1] = True + gg.ignored = True + nignoredtp += 1 + + # if the associated tracker detection is already ignored, + # we want to avoid double counting ignored detections + if ignoredtrackers[gg.tracker] > 0: + nignoredpairs += 1 + + # for computing MODP, the overlaps from ignored detections + # are subtracted + tmpc -= tmpcs[gi] + gi += 1 + + # the below might be confusion, check the comments in __init__ + # to see what the individual statistics represent + + # correct TP by number of ignored TP due to truncation + # ignored TP are shown as tracked in visualization + tmptp -= nignoredtp + + # count the number of ignored true positives + self.itp += nignoredtp + + # adjust the number of ground truth objects considered + self.n_gt -= (ignoredfn + nignoredtp) + + # count the number of ignored ground truth objects + self.n_igt += ignoredfn + nignoredtp + + # count the number of ignored tracker objects + self.n_itr += nignoredtracker + + # count the number of ignored pairs, i.e. associated tracker and + # ground truth objects that are both ignored + self.n_igttr += nignoredpairs + + # false negatives = associated gt bboxes exceding association threshold + non-associated gt bboxes + tmpfn += len(g) - len(association_matrix) - ignoredfn + self.fn += len(g) - len(association_matrix) - ignoredfn + self.ifn += ignoredfn + + # false positives = tracker bboxes - associated tracker bboxes + # mismatches (mme_t) + tmpfp += len( + t) - tmptp - nignoredtracker - nignoredtp + nignoredpairs + self.fp += len( + t) - tmptp - nignoredtracker - nignoredtp + nignoredpairs + + # update sequence data + seqtp += tmptp + seqitp += nignoredtp + seqfp += tmpfp + seqfn += tmpfn + seqifn += ignoredfn + seqigt += ignoredfn + nignoredtp + seqitr += nignoredtracker + + # sanity checks + # - the number of true positives minues ignored true positives + # should be greater or equal to 0 + # - the number of false negatives should be greater or equal to 0 + # - the number of false positives needs to be greater or equal to 0 + # otherwise ignored detections might be counted double + # - the number of counted true positives (plus ignored ones) + # and the number of counted false negatives (plus ignored ones) + # should match the total number of ground truth objects + # - the number of counted true positives (plus ignored ones) + # and the number of counted false positives + # plus the number of ignored tracker detections should + # match the total number of tracker detections; note that + # nignoredpairs is subtracted here to avoid double counting + # of ignored detection sin nignoredtp and nignoredtracker + if tmptp < 0: + print(tmptp, nignoredtp) + raise NameError("Something went wrong! TP is negative") + if tmpfn < 0: + print(tmpfn, + len(g), + len(association_matrix), ignoredfn, nignoredpairs) + raise NameError("Something went wrong! FN is negative") + if tmpfp < 0: + print(tmpfp, + len(t), tmptp, nignoredtracker, nignoredtp, + nignoredpairs) + raise NameError("Something went wrong! FP is negative") + if tmptp + tmpfn is not len(g) - ignoredfn - nignoredtp: + print("seqidx", seq_idx) + print("frame ", f) + print("TP ", tmptp) + print("FN ", tmpfn) + print("FP ", tmpfp) + print("nGT ", len(g)) + print("nAss ", len(association_matrix)) + print("ign GT", ignoredfn) + print("ign TP", nignoredtp) + raise NameError( + "Something went wrong! nGroundtruth is not TP+FN") + if tmptp + tmpfp + nignoredtp + nignoredtracker - nignoredpairs is not len( + t): + print(seq_idx, f, len(t), tmptp, tmpfp) + print(len(association_matrix), association_matrix) + raise NameError( + "Something went wrong! nTracker is not TP+FP") + + # check for id switches or fragmentations + for i, tt in enumerate(this_ids[0]): + if tt in last_ids[0]: + idx = last_ids[0].index(tt) + tid = this_ids[1][i] + lid = last_ids[1][idx] + if tid != lid and lid != -1 and tid != -1: + if g[i].truncation < self.max_truncation: + g[i].id_switch = 1 + ids += 1 + if tid != lid and lid != -1: + if g[i].truncation < self.max_truncation: + g[i].fragmentation = 1 + fr += 1 + + # save current index + last_ids = this_ids + # compute MOTP_t + MODP_t = 1 + if tmptp != 0: + MODP_t = tmpc / float(tmptp) + self.MODP_t.append(MODP_t) + + # remove empty lists for current gt trajectories + self.gt_trajectories[seq_idx] = seq_trajectories + self.ign_trajectories[seq_idx] = seq_ignored + + # gather statistics for "per sequence" statistics. + self.n_gts.append(n_gts) + self.n_trs.append(n_trs) + self.tps.append(seqtp) + self.itps.append(seqitp) + self.fps.append(seqfp) + self.fns.append(seqfn) + self.ifns.append(seqifn) + self.n_igts.append(seqigt) + self.n_itrs.append(seqitr) + + # compute MT/PT/ML, fragments, idswitches for all groundtruth trajectories + n_ignored_tr_total = 0 + for seq_idx, ( + seq_trajectories, seq_ignored + ) in enumerate(zip(self.gt_trajectories, self.ign_trajectories)): + if len(seq_trajectories) == 0: + continue + tmpMT, tmpML, tmpPT, tmpId_switches, tmpFragments = [0] * 5 + n_ignored_tr = 0 + for g, ign_g in zip(seq_trajectories.values(), + seq_ignored.values()): + # all frames of this gt trajectory are ignored + if all(ign_g): + n_ignored_tr += 1 + n_ignored_tr_total += 1 + continue + # all frames of this gt trajectory are not assigned to any detections + if all([this == -1 for this in g]): + tmpML += 1 + self.ML += 1 + continue + # compute tracked frames in trajectory + last_id = g[0] + # first detection (necessary to be in gt_trajectories) is always tracked + tracked = 1 if g[0] >= 0 else 0 + lgt = 0 if ign_g[0] else 1 + for f in range(1, len(g)): + if ign_g[f]: + last_id = -1 + continue + lgt += 1 + if last_id != g[f] and last_id != -1 and g[f] != -1 and g[ + f - 1] != -1: + tmpId_switches += 1 + self.id_switches += 1 + if f < len(g) - 1 and g[f - 1] != g[ + f] and last_id != -1 and g[f] != -1 and g[f + + 1] != -1: + tmpFragments += 1 + self.fragments += 1 + if g[f] != -1: + tracked += 1 + last_id = g[f] + # handle last frame; tracked state is handled in for loop (g[f]!=-1) + if len(g) > 1 and g[f - 1] != g[f] and last_id != -1 and g[ + f] != -1 and not ign_g[f]: + tmpFragments += 1 + self.fragments += 1 + + # compute MT/PT/ML + tracking_ratio = tracked / float(len(g) - sum(ign_g)) + if tracking_ratio > 0.8: + tmpMT += 1 + self.MT += 1 + elif tracking_ratio < 0.2: + tmpML += 1 + self.ML += 1 + else: # 0.2 <= tracking_ratio <= 0.8 + tmpPT += 1 + self.PT += 1 + + if (self.n_gt_trajectories - n_ignored_tr_total) == 0: + self.MT = 0. + self.PT = 0. + self.ML = 0. + else: + self.MT /= float(self.n_gt_trajectories - n_ignored_tr_total) + self.PT /= float(self.n_gt_trajectories - n_ignored_tr_total) + self.ML /= float(self.n_gt_trajectories - n_ignored_tr_total) + + # precision/recall etc. + if (self.fp + self.tp) == 0 or (self.tp + self.fn) == 0: + self.recall = 0. + self.precision = 0. + else: + self.recall = self.tp / float(self.tp + self.fn) + self.precision = self.tp / float(self.fp + self.tp) + if (self.recall + self.precision) == 0: + self.F1 = 0. + else: + self.F1 = 2. * (self.precision * self.recall) / ( + self.precision + self.recall) + if sum(self.n_frames) == 0: + self.FAR = "n/a" + else: + self.FAR = self.fp / float(sum(self.n_frames)) + + # compute CLEARMOT + if self.n_gt == 0: + self.MOTA = -float("inf") + self.MODA = -float("inf") + else: + self.MOTA = 1 - (self.fn + self.fp + self.id_switches + ) / float(self.n_gt) + self.MODA = 1 - (self.fn + self.fp) / float(self.n_gt) + if self.tp == 0: + self.MOTP = float("inf") + else: + self.MOTP = self.total_cost / float(self.tp) + if self.n_gt != 0: + if self.id_switches == 0: + self.MOTAL = 1 - (self.fn + self.fp + self.id_switches + ) / float(self.n_gt) + else: + self.MOTAL = 1 - (self.fn + self.fp + + math.log10(self.id_switches) + ) / float(self.n_gt) + else: + self.MOTAL = -float("inf") + if sum(self.n_frames) == 0: + self.MODP = "n/a" + else: + self.MODP = sum(self.MODP_t) / float(sum(self.n_frames)) + return True + + def createSummary(self): + summary = "" + summary += "tracking evaluation summary".center(80, "=") + "\n" + summary += self.printEntry("Multiple Object Tracking Accuracy (MOTA)", + self.MOTA) + "\n" + summary += self.printEntry("Multiple Object Tracking Precision (MOTP)", + self.MOTP) + "\n" + summary += self.printEntry("Multiple Object Tracking Accuracy (MOTAL)", + self.MOTAL) + "\n" + summary += self.printEntry("Multiple Object Detection Accuracy (MODA)", + self.MODA) + "\n" + summary += self.printEntry("Multiple Object Detection Precision (MODP)", + self.MODP) + "\n" + summary += "\n" + summary += self.printEntry("Recall", self.recall) + "\n" + summary += self.printEntry("Precision", self.precision) + "\n" + summary += self.printEntry("F1", self.F1) + "\n" + summary += self.printEntry("False Alarm Rate", self.FAR) + "\n" + summary += "\n" + summary += self.printEntry("Mostly Tracked", self.MT) + "\n" + summary += self.printEntry("Partly Tracked", self.PT) + "\n" + summary += self.printEntry("Mostly Lost", self.ML) + "\n" + summary += "\n" + summary += self.printEntry("True Positives", self.tp) + "\n" + #summary += self.printEntry("True Positives per Sequence", self.tps) + "\n" + summary += self.printEntry("Ignored True Positives", self.itp) + "\n" + #summary += self.printEntry("Ignored True Positives per Sequence", self.itps) + "\n" + + summary += self.printEntry("False Positives", self.fp) + "\n" + #summary += self.printEntry("False Positives per Sequence", self.fps) + "\n" + summary += self.printEntry("False Negatives", self.fn) + "\n" + #summary += self.printEntry("False Negatives per Sequence", self.fns) + "\n" + summary += self.printEntry("ID-switches", self.id_switches) + "\n" + self.fp = self.fp / self.n_gt + self.fn = self.fn / self.n_gt + self.id_switches = self.id_switches / self.n_gt + summary += self.printEntry("False Positives Ratio", self.fp) + "\n" + #summary += self.printEntry("False Positives per Sequence", self.fps) + "\n" + summary += self.printEntry("False Negatives Ratio", self.fn) + "\n" + #summary += self.printEntry("False Negatives per Sequence", self.fns) + "\n" + summary += self.printEntry("Ignored False Negatives Ratio", + self.ifn) + "\n" + + #summary += self.printEntry("Ignored False Negatives per Sequence", self.ifns) + "\n" + summary += self.printEntry("Missed Targets", self.fn) + "\n" + summary += self.printEntry("ID-switches", self.id_switches) + "\n" + summary += self.printEntry("Fragmentations", self.fragments) + "\n" + summary += "\n" + summary += self.printEntry("Ground Truth Objects (Total)", self.n_gt + + self.n_igt) + "\n" + #summary += self.printEntry("Ground Truth Objects (Total) per Sequence", self.n_gts) + "\n" + summary += self.printEntry("Ignored Ground Truth Objects", + self.n_igt) + "\n" + #summary += self.printEntry("Ignored Ground Truth Objects per Sequence", self.n_igts) + "\n" + summary += self.printEntry("Ground Truth Trajectories", + self.n_gt_trajectories) + "\n" + summary += "\n" + summary += self.printEntry("Tracker Objects (Total)", self.n_tr) + "\n" + #summary += self.printEntry("Tracker Objects (Total) per Sequence", self.n_trs) + "\n" + summary += self.printEntry("Ignored Tracker Objects", self.n_itr) + "\n" + #summary += self.printEntry("Ignored Tracker Objects per Sequence", self.n_itrs) + "\n" + summary += self.printEntry("Tracker Trajectories", + self.n_tr_trajectories) + "\n" + #summary += "\n" + #summary += self.printEntry("Ignored Tracker Objects with Associated Ignored Ground Truth Objects", self.n_igttr) + "\n" + summary += "=" * 80 + return summary + + def printEntry(self, key, val, width=(70, 10)): + """ + Pretty print an entry in a table fashion. + """ + s_out = key.ljust(width[0]) + if type(val) == int: + s = "%%%dd" % width[1] + s_out += s % val + elif type(val) == float: + s = "%%%df" % (width[1]) + s_out += s % val + else: + s_out += ("%s" % val).rjust(width[1]) + return s_out + + def saveToStats(self, save_summary): + """ + Save the statistics in a whitespace separate file. + """ + summary = self.createSummary() + if save_summary: + filename = os.path.join(self.result_path, + "summary_%s.txt" % self.cls) + dump = open(filename, "w+") + dump.write(summary) + dump.close() + return summary + + +class KITTIMOTMetric(Metric): + def __init__(self, save_summary=True): + self.save_summary = save_summary + self.MOTEvaluator = KITTIEvaluation + self.result_root = None + self.reset() + + def reset(self): + self.seqs = [] + self.n_sequences = 0 + self.n_frames = [] + self.strsummary = '' + + def update(self, data_root, seq, data_type, result_root, result_filename): + assert data_type == 'kitti', "data_type should 'kitti'" + self.result_root = result_root + self.gt_path = data_root + gt_path = '{}/../labels/{}.txt'.format(data_root, seq) + gt = open(gt_path, "r") + max_frame = 0 + for line in gt: + line = line.strip() + line_list = line.split(" ") + if int(line_list[0]) > max_frame: + max_frame = int(line_list[0]) + rs = open(result_filename, "r") + for line in rs: + line = line.strip() + line_list = line.split(" ") + if int(line_list[0]) > max_frame: + max_frame = int(line_list[0]) + gt.close() + rs.close() + self.n_frames.append(max_frame + 1) + self.seqs.append(seq) + self.n_sequences += 1 + + def accumulate(self): + logger.info("Processing Result for KITTI Tracking Benchmark") + e = self.MOTEvaluator(result_path=self.result_root, gt_path=self.gt_path,\ + n_frames=self.n_frames, seqs=self.seqs, n_sequences=self.n_sequences) + try: + if not e.loadTracker(): + return + logger.info("Loading Results - Success") + logger.info("Evaluate Object Class: %s" % c.upper()) + except: + logger.info("Caught exception while loading result data.") + if not e.loadGroundtruth(): + raise ValueError("Ground truth not found.") + logger.info("Loading Groundtruth - Success") + # sanity checks + if len(e.groundtruth) is not len(e.tracker): + logger.info( + "The uploaded data does not provide results for every sequence.") + return False + logger.info("Loaded %d Sequences." % len(e.groundtruth)) + logger.info("Start Evaluation...") + + if e.compute3rdPartyMetrics(): + self.strsummary = e.saveToStats(self.save_summary) + else: + logger.info( + "There seem to be no true positives or false positives at all in the submitted data." + ) + + def log(self): + print(self.strsummary) + + def get_results(self): + return self.strsummary diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/munkres.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/munkres.py new file mode 100644 index 000000000..fbd4a92d2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/munkres.py @@ -0,0 +1,428 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is borrow from https://github.com/xingyizhou/CenterTrack/blob/master/src/tools/eval_kitti_track/munkres.py +""" + +import sys + +__all__ = ['Munkres', 'make_cost_matrix'] + + +class Munkres: + """ + Calculate the Munkres solution to the classical assignment problem. + See the module documentation for usage. + """ + + def __init__(self): + """Create a new instance""" + self.C = None + self.row_covered = [] + self.col_covered = [] + self.n = 0 + self.Z0_r = 0 + self.Z0_c = 0 + self.marked = None + self.path = None + + def make_cost_matrix(profit_matrix, inversion_function): + """ + **DEPRECATED** + + Please use the module function ``make_cost_matrix()``. + """ + import munkres + return munkres.make_cost_matrix(profit_matrix, inversion_function) + + make_cost_matrix = staticmethod(make_cost_matrix) + + def pad_matrix(self, matrix, pad_value=0): + """ + Pad a possibly non-square matrix to make it square. + + :Parameters: + matrix : list of lists + matrix to pad + + pad_value : int + value to use to pad the matrix + + :rtype: list of lists + :return: a new, possibly padded, matrix + """ + max_columns = 0 + total_rows = len(matrix) + + for row in matrix: + max_columns = max(max_columns, len(row)) + + total_rows = max(max_columns, total_rows) + + new_matrix = [] + for row in matrix: + row_len = len(row) + new_row = row[:] + if total_rows > row_len: + # Row too short. Pad it. + new_row += [0] * (total_rows - row_len) + new_matrix += [new_row] + + while len(new_matrix) < total_rows: + new_matrix += [[0] * total_rows] + + return new_matrix + + def compute(self, cost_matrix): + """ + Compute the indexes for the lowest-cost pairings between rows and + columns in the database. Returns a list of (row, column) tuples + that can be used to traverse the matrix. + + :Parameters: + cost_matrix : list of lists + The cost matrix. If this cost matrix is not square, it + will be padded with zeros, via a call to ``pad_matrix()``. + (This method does *not* modify the caller's matrix. It + operates on a copy of the matrix.) + + **WARNING**: This code handles square and rectangular + matrices. It does *not* handle irregular matrices. + + :rtype: list + :return: A list of ``(row, column)`` tuples that describe the lowest + cost path through the matrix + + """ + self.C = self.pad_matrix(cost_matrix) + self.n = len(self.C) + self.original_length = len(cost_matrix) + self.original_width = len(cost_matrix[0]) + self.row_covered = [False for i in range(self.n)] + self.col_covered = [False for i in range(self.n)] + self.Z0_r = 0 + self.Z0_c = 0 + self.path = self.__make_matrix(self.n * 2, 0) + self.marked = self.__make_matrix(self.n, 0) + + done = False + step = 1 + + steps = { + 1: self.__step1, + 2: self.__step2, + 3: self.__step3, + 4: self.__step4, + 5: self.__step5, + 6: self.__step6 + } + + while not done: + try: + func = steps[step] + step = func() + except KeyError: + done = True + + # Look for the starred columns + results = [] + for i in range(self.original_length): + for j in range(self.original_width): + if self.marked[i][j] == 1: + results += [(i, j)] + + return results + + def __copy_matrix(self, matrix): + """Return an exact copy of the supplied matrix""" + return copy.deepcopy(matrix) + + def __make_matrix(self, n, val): + """Create an *n*x*n* matrix, populating it with the specific value.""" + matrix = [] + for i in range(n): + matrix += [[val for j in range(n)]] + return matrix + + def __step1(self): + """ + For each row of the matrix, find the smallest element and + subtract it from every element in its row. Go to Step 2. + """ + C = self.C + n = self.n + for i in range(n): + minval = min(self.C[i]) + # Find the minimum value for this row and subtract that minimum + # from every element in the row. + for j in range(n): + self.C[i][j] -= minval + + return 2 + + def __step2(self): + """ + Find a zero (Z) in the resulting matrix. If there is no starred + zero in its row or column, star Z. Repeat for each element in the + matrix. Go to Step 3. + """ + n = self.n + for i in range(n): + for j in range(n): + if (self.C[i][j] == 0) and \ + (not self.col_covered[j]) and \ + (not self.row_covered[i]): + self.marked[i][j] = 1 + self.col_covered[j] = True + self.row_covered[i] = True + + self.__clear_covers() + return 3 + + def __step3(self): + """ + Cover each column containing a starred zero. If K columns are + covered, the starred zeros describe a complete set of unique + assignments. In this case, Go to DONE, otherwise, Go to Step 4. + """ + n = self.n + count = 0 + for i in range(n): + for j in range(n): + if self.marked[i][j] == 1: + self.col_covered[j] = True + count += 1 + + if count >= n: + step = 7 # done + else: + step = 4 + + return step + + def __step4(self): + """ + Find a noncovered zero and prime it. If there is no starred zero + in the row containing this primed zero, Go to Step 5. Otherwise, + cover this row and uncover the column containing the starred + zero. Continue in this manner until there are no uncovered zeros + left. Save the smallest uncovered value and Go to Step 6. + """ + step = 0 + done = False + row = -1 + col = -1 + star_col = -1 + while not done: + (row, col) = self.__find_a_zero() + if row < 0: + done = True + step = 6 + else: + self.marked[row][col] = 2 + star_col = self.__find_star_in_row(row) + if star_col >= 0: + col = star_col + self.row_covered[row] = True + self.col_covered[col] = False + else: + done = True + self.Z0_r = row + self.Z0_c = col + step = 5 + + return step + + def __step5(self): + """ + Construct a series of alternating primed and starred zeros as + follows. Let Z0 represent the uncovered primed zero found in Step 4. + Let Z1 denote the starred zero in the column of Z0 (if any). + Let Z2 denote the primed zero in the row of Z1 (there will always + be one). Continue until the series terminates at a primed zero + that has no starred zero in its column. Unstar each starred zero + of the series, star each primed zero of the series, erase all + primes and uncover every line in the matrix. Return to Step 3 + """ + count = 0 + path = self.path + path[count][0] = self.Z0_r + path[count][1] = self.Z0_c + done = False + while not done: + row = self.__find_star_in_col(path[count][1]) + if row >= 0: + count += 1 + path[count][0] = row + path[count][1] = path[count - 1][1] + else: + done = True + + if not done: + col = self.__find_prime_in_row(path[count][0]) + count += 1 + path[count][0] = path[count - 1][0] + path[count][1] = col + + self.__convert_path(path, count) + self.__clear_covers() + self.__erase_primes() + return 3 + + def __step6(self): + """ + Add the value found in Step 4 to every element of each covered + row, and subtract it from every element of each uncovered column. + Return to Step 4 without altering any stars, primes, or covered + lines. + """ + minval = self.__find_smallest() + for i in range(self.n): + for j in range(self.n): + if self.row_covered[i]: + self.C[i][j] += minval + if not self.col_covered[j]: + self.C[i][j] -= minval + return 4 + + def __find_smallest(self): + """Find the smallest uncovered value in the matrix.""" + minval = 2e9 # sys.maxint + for i in range(self.n): + for j in range(self.n): + if (not self.row_covered[i]) and (not self.col_covered[j]): + if minval > self.C[i][j]: + minval = self.C[i][j] + return minval + + def __find_a_zero(self): + """Find the first uncovered element with value 0""" + row = -1 + col = -1 + i = 0 + n = self.n + done = False + + while not done: + j = 0 + while True: + if (self.C[i][j] == 0) and \ + (not self.row_covered[i]) and \ + (not self.col_covered[j]): + row = i + col = j + done = True + j += 1 + if j >= n: + break + i += 1 + if i >= n: + done = True + + return (row, col) + + def __find_star_in_row(self, row): + """ + Find the first starred element in the specified row. Returns + the column index, or -1 if no starred element was found. + """ + col = -1 + for j in range(self.n): + if self.marked[row][j] == 1: + col = j + break + + return col + + def __find_star_in_col(self, col): + """ + Find the first starred element in the specified row. Returns + the row index, or -1 if no starred element was found. + """ + row = -1 + for i in range(self.n): + if self.marked[i][col] == 1: + row = i + break + + return row + + def __find_prime_in_row(self, row): + """ + Find the first prime element in the specified row. Returns + the column index, or -1 if no starred element was found. + """ + col = -1 + for j in range(self.n): + if self.marked[row][j] == 2: + col = j + break + + return col + + def __convert_path(self, path, count): + for i in range(count + 1): + if self.marked[path[i][0]][path[i][1]] == 1: + self.marked[path[i][0]][path[i][1]] = 0 + else: + self.marked[path[i][0]][path[i][1]] = 1 + + def __clear_covers(self): + """Clear all covered matrix cells""" + for i in range(self.n): + self.row_covered[i] = False + self.col_covered[i] = False + + def __erase_primes(self): + """Erase all prime markings""" + for i in range(self.n): + for j in range(self.n): + if self.marked[i][j] == 2: + self.marked[i][j] = 0 + + +def make_cost_matrix(profit_matrix, inversion_function): + """ + Create a cost matrix from a profit matrix by calling + 'inversion_function' to invert each value. The inversion + function must take one numeric argument (of any type) and return + another numeric argument which is presumed to be the cost inverse + of the original profit. + + This is a static method. Call it like this: + + .. python:: + + cost_matrix = Munkres.make_cost_matrix(matrix, inversion_func) + + For example: + + .. python:: + + cost_matrix = Munkres.make_cost_matrix(matrix, lambda x : sys.maxint - x) + + :Parameters: + profit_matrix : list of lists + The matrix to convert from a profit to a cost matrix + + inversion_function : function + The function to use to invert each entry in the profit matrix + + :rtype: list of lists + :return: The converted matrix + """ + cost_matrix = [] + for row in profit_matrix: + cost_matrix.append([inversion_function(value) for value in row]) + return cost_matrix diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/widerface_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/widerface_utils.py new file mode 100644 index 000000000..2f64bf6d5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/metrics/widerface_utils.py @@ -0,0 +1,391 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import cv2 +import numpy as np +from collections import OrderedDict + +import paddle + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['face_eval_run', 'lmk2out'] + + +def face_eval_run(model, + image_dir, + gt_file, + pred_dir='output/pred', + eval_mode='widerface', + multi_scale=False): + # load ground truth files + with open(gt_file, 'r') as f: + gt_lines = f.readlines() + imid2path = [] + pos_gt = 0 + while pos_gt < len(gt_lines): + name_gt = gt_lines[pos_gt].strip('\n\t').split()[0] + imid2path.append(name_gt) + pos_gt += 1 + n_gt = int(gt_lines[pos_gt].strip('\n\t').split()[0]) + pos_gt += 1 + n_gt + logger.info('The ground truth file load {} images'.format(len(imid2path))) + + dets_dist = OrderedDict() + for iter_id, im_path in enumerate(imid2path): + image_path = os.path.join(image_dir, im_path) + if eval_mode == 'fddb': + image_path += '.jpg' + assert os.path.exists(image_path) + image = cv2.imread(image_path) + image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) + if multi_scale: + shrink, max_shrink = get_shrink(image.shape[0], image.shape[1]) + det0 = detect_face(model, image, shrink) + det1 = flip_test(model, image, shrink) + [det2, det3] = multi_scale_test(model, image, max_shrink) + det4 = multi_scale_test_pyramid(model, image, max_shrink) + det = np.row_stack((det0, det1, det2, det3, det4)) + dets = bbox_vote(det) + else: + dets = detect_face(model, image, 1) + if eval_mode == 'widerface': + save_widerface_bboxes(image_path, dets, pred_dir) + else: + dets_dist[im_path] = dets + if iter_id % 100 == 0: + logger.info('Test iter {}'.format(iter_id)) + if eval_mode == 'fddb': + save_fddb_bboxes(dets_dist, pred_dir) + logger.info("Finish evaluation.") + + +def detect_face(model, image, shrink): + image_shape = [image.shape[0], image.shape[1]] + if shrink != 1: + h, w = int(image_shape[0] * shrink), int(image_shape[1] * shrink) + image = cv2.resize(image, (w, h)) + image_shape = [h, w] + + img = face_img_process(image) + image_shape = np.asarray([image_shape]) + scale_factor = np.asarray([[shrink, shrink]]) + data = { + "image": paddle.to_tensor( + img, dtype='float32'), + "im_shape": paddle.to_tensor( + image_shape, dtype='float32'), + "scale_factor": paddle.to_tensor( + scale_factor, dtype='float32') + } + model.eval() + detection = model(data) + detection = detection['bbox'].numpy() + # layout: xmin, ymin, xmax. ymax, score + if np.prod(detection.shape) == 1: + logger.info("No face detected") + return np.array([[0, 0, 0, 0, 0]]) + det_conf = detection[:, 1] + det_xmin = detection[:, 2] + det_ymin = detection[:, 3] + det_xmax = detection[:, 4] + det_ymax = detection[:, 5] + + det = np.column_stack((det_xmin, det_ymin, det_xmax, det_ymax, det_conf)) + return det + + +def flip_test(model, image, shrink): + img = cv2.flip(image, 1) + det_f = detect_face(model, img, shrink) + det_t = np.zeros(det_f.shape) + img_width = image.shape[1] + det_t[:, 0] = img_width - det_f[:, 2] + det_t[:, 1] = det_f[:, 1] + det_t[:, 2] = img_width - det_f[:, 0] + det_t[:, 3] = det_f[:, 3] + det_t[:, 4] = det_f[:, 4] + return det_t + + +def multi_scale_test(model, image, max_shrink): + # Shrink detecting is only used to detect big faces + st = 0.5 if max_shrink >= 0.75 else 0.5 * max_shrink + det_s = detect_face(model, image, st) + index = np.where( + np.maximum(det_s[:, 2] - det_s[:, 0] + 1, det_s[:, 3] - det_s[:, 1] + 1) + > 30)[0] + det_s = det_s[index, :] + # Enlarge one times + bt = min(2, max_shrink) if max_shrink > 1 else (st + max_shrink) / 2 + det_b = detect_face(model, image, bt) + + # Enlarge small image x times for small faces + if max_shrink > 2: + bt *= 2 + while bt < max_shrink: + det_b = np.row_stack((det_b, detect_face(model, image, bt))) + bt *= 2 + det_b = np.row_stack((det_b, detect_face(model, image, max_shrink))) + + # Enlarged images are only used to detect small faces. + if bt > 1: + index = np.where( + np.minimum(det_b[:, 2] - det_b[:, 0] + 1, + det_b[:, 3] - det_b[:, 1] + 1) < 100)[0] + det_b = det_b[index, :] + # Shrinked images are only used to detect big faces. + else: + index = np.where( + np.maximum(det_b[:, 2] - det_b[:, 0] + 1, + det_b[:, 3] - det_b[:, 1] + 1) > 30)[0] + det_b = det_b[index, :] + return det_s, det_b + + +def multi_scale_test_pyramid(model, image, max_shrink): + # Use image pyramids to detect faces + det_b = detect_face(model, image, 0.25) + index = np.where( + np.maximum(det_b[:, 2] - det_b[:, 0] + 1, det_b[:, 3] - det_b[:, 1] + 1) + > 30)[0] + det_b = det_b[index, :] + + st = [0.75, 1.25, 1.5, 1.75] + for i in range(len(st)): + if st[i] <= max_shrink: + det_temp = detect_face(model, image, st[i]) + # Enlarged images are only used to detect small faces. + if st[i] > 1: + index = np.where( + np.minimum(det_temp[:, 2] - det_temp[:, 0] + 1, + det_temp[:, 3] - det_temp[:, 1] + 1) < 100)[0] + det_temp = det_temp[index, :] + # Shrinked images are only used to detect big faces. + else: + index = np.where( + np.maximum(det_temp[:, 2] - det_temp[:, 0] + 1, + det_temp[:, 3] - det_temp[:, 1] + 1) > 30)[0] + det_temp = det_temp[index, :] + det_b = np.row_stack((det_b, det_temp)) + return det_b + + +def to_chw(image): + """ + Transpose image from HWC to CHW. + Args: + image (np.array): an image with HWC layout. + """ + # HWC to CHW + if len(image.shape) == 3: + image = np.swapaxes(image, 1, 2) + image = np.swapaxes(image, 1, 0) + return image + + +def face_img_process(image, + mean=[104., 117., 123.], + std=[127.502231, 127.502231, 127.502231]): + img = np.array(image) + img = to_chw(img) + img = img.astype('float32') + img -= np.array(mean)[:, np.newaxis, np.newaxis].astype('float32') + img /= np.array(std)[:, np.newaxis, np.newaxis].astype('float32') + img = [img] + img = np.array(img) + return img + + +def get_shrink(height, width): + """ + Args: + height (int): image height. + width (int): image width. + """ + # avoid out of memory + max_shrink_v1 = (0x7fffffff / 577.0 / (height * width))**0.5 + max_shrink_v2 = ((678 * 1024 * 2.0 * 2.0) / (height * width))**0.5 + + def get_round(x, loc): + str_x = str(x) + if '.' in str_x: + str_before, str_after = str_x.split('.') + len_after = len(str_after) + if len_after >= 3: + str_final = str_before + '.' + str_after[0:loc] + return float(str_final) + else: + return x + + max_shrink = get_round(min(max_shrink_v1, max_shrink_v2), 2) - 0.3 + if max_shrink >= 1.5 and max_shrink < 2: + max_shrink = max_shrink - 0.1 + elif max_shrink >= 2 and max_shrink < 3: + max_shrink = max_shrink - 0.2 + elif max_shrink >= 3 and max_shrink < 4: + max_shrink = max_shrink - 0.3 + elif max_shrink >= 4 and max_shrink < 5: + max_shrink = max_shrink - 0.4 + elif max_shrink >= 5: + max_shrink = max_shrink - 0.5 + elif max_shrink <= 0.1: + max_shrink = 0.1 + + shrink = max_shrink if max_shrink < 1 else 1 + return shrink, max_shrink + + +def bbox_vote(det): + order = det[:, 4].ravel().argsort()[::-1] + det = det[order, :] + if det.shape[0] == 0: + dets = np.array([[10, 10, 20, 20, 0.002]]) + det = np.empty(shape=[0, 5]) + while det.shape[0] > 0: + # IOU + area = (det[:, 2] - det[:, 0] + 1) * (det[:, 3] - det[:, 1] + 1) + xx1 = np.maximum(det[0, 0], det[:, 0]) + yy1 = np.maximum(det[0, 1], det[:, 1]) + xx2 = np.minimum(det[0, 2], det[:, 2]) + yy2 = np.minimum(det[0, 3], det[:, 3]) + w = np.maximum(0.0, xx2 - xx1 + 1) + h = np.maximum(0.0, yy2 - yy1 + 1) + inter = w * h + o = inter / (area[0] + area[:] - inter) + + # nms + merge_index = np.where(o >= 0.3)[0] + det_accu = det[merge_index, :] + det = np.delete(det, merge_index, 0) + if merge_index.shape[0] <= 1: + if det.shape[0] == 0: + try: + dets = np.row_stack((dets, det_accu)) + except: + dets = det_accu + continue + det_accu[:, 0:4] = det_accu[:, 0:4] * np.tile(det_accu[:, -1:], (1, 4)) + max_score = np.max(det_accu[:, 4]) + det_accu_sum = np.zeros((1, 5)) + det_accu_sum[:, 0:4] = np.sum(det_accu[:, 0:4], + axis=0) / np.sum(det_accu[:, -1:]) + det_accu_sum[:, 4] = max_score + try: + dets = np.row_stack((dets, det_accu_sum)) + except: + dets = det_accu_sum + dets = dets[0:750, :] + keep_index = np.where(dets[:, 4] >= 0.01)[0] + dets = dets[keep_index, :] + return dets + + +def save_widerface_bboxes(image_path, bboxes_scores, output_dir): + image_name = image_path.split('/')[-1] + image_class = image_path.split('/')[-2] + odir = os.path.join(output_dir, image_class) + if not os.path.exists(odir): + os.makedirs(odir) + + ofname = os.path.join(odir, '%s.txt' % (image_name[:-4])) + f = open(ofname, 'w') + f.write('{:s}\n'.format(image_class + '/' + image_name)) + f.write('{:d}\n'.format(bboxes_scores.shape[0])) + for box_score in bboxes_scores: + xmin, ymin, xmax, ymax, score = box_score + f.write('{:.1f} {:.1f} {:.1f} {:.1f} {:.3f}\n'.format(xmin, ymin, ( + xmax - xmin + 1), (ymax - ymin + 1), score)) + f.close() + logger.info("The predicted result is saved as {}".format(ofname)) + + +def save_fddb_bboxes(bboxes_scores, + output_dir, + output_fname='pred_fddb_res.txt'): + if not os.path.exists(output_dir): + os.makedirs(output_dir) + predict_file = os.path.join(output_dir, output_fname) + f = open(predict_file, 'w') + for image_path, dets in bboxes_scores.iteritems(): + f.write('{:s}\n'.format(image_path)) + f.write('{:d}\n'.format(dets.shape[0])) + for box_score in dets: + xmin, ymin, xmax, ymax, score = box_score + width, height = xmax - xmin, ymax - ymin + f.write('{:.1f} {:.1f} {:.1f} {:.1f} {:.3f}\n' + .format(xmin, ymin, width, height, score)) + logger.info("The predicted result is saved as {}".format(predict_file)) + return predict_file + + +def lmk2out(results, is_bbox_normalized=False): + """ + Args: + results: request a dict, should include: `landmark`, `im_id`, + if is_bbox_normalized=True, also need `im_shape`. + is_bbox_normalized: whether or not landmark is normalized. + """ + xywh_res = [] + for t in results: + bboxes = t['bbox'][0] + lengths = t['bbox'][1][0] + im_ids = np.array(t['im_id'][0]).flatten() + if bboxes.shape == (1, 1) or bboxes is None: + continue + face_index = t['face_index'][0] + prior_box = t['prior_boxes'][0] + predict_lmk = t['landmark'][0] + prior = np.reshape(prior_box, (-1, 4)) + predictlmk = np.reshape(predict_lmk, (-1, 10)) + + k = 0 + for a in range(len(lengths)): + num = lengths[a] + im_id = int(im_ids[a]) + for i in range(num): + score = bboxes[k][1] + theindex = face_index[i][0] + me_prior = prior[theindex, :] + lmk_pred = predictlmk[theindex, :] + prior_w = me_prior[2] - me_prior[0] + prior_h = me_prior[3] - me_prior[1] + prior_w_center = (me_prior[2] + me_prior[0]) / 2 + prior_h_center = (me_prior[3] + me_prior[1]) / 2 + lmk_decode = np.zeros((10)) + for j in [0, 2, 4, 6, 8]: + lmk_decode[j] = lmk_pred[j] * 0.1 * prior_w + prior_w_center + for j in [1, 3, 5, 7, 9]: + lmk_decode[j] = lmk_pred[j] * 0.1 * prior_h + prior_h_center + im_shape = t['im_shape'][0][a].tolist() + image_h, image_w = int(im_shape[0]), int(im_shape[1]) + if is_bbox_normalized: + lmk_decode = lmk_decode * np.array([ + image_w, image_h, image_w, image_h, image_w, image_h, + image_w, image_h, image_w, image_h + ]) + lmk_res = { + 'image_id': im_id, + 'landmark': lmk_decode, + 'score': score, + } + xywh_res.append(lmk_res) + k += 1 + return xywh_res diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/.gitignore b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/.gitignore new file mode 100644 index 000000000..f296851d6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/.gitignore @@ -0,0 +1 @@ +MODEL_ZOO diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__init__.py new file mode 100644 index 000000000..6db6eb6c6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__init__.py @@ -0,0 +1,18 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import model_zoo +from .model_zoo import * + +__all__ = model_zoo.__all__ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..dccbb6902 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/model_zoo.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/model_zoo.cpython-37.pyc new file mode 100644 index 000000000..dd8c6c51a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/__pycache__/model_zoo.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/model_zoo.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/model_zoo.py new file mode 100644 index 000000000..27581ef79 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/model_zoo.py @@ -0,0 +1,84 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os.path as osp +import pkg_resources + +try: + from collections.abc import Sequence +except: + from collections import Sequence + +from ppdet.core.workspace import load_config, create +from ppdet.utils.checkpoint import load_weight +from ppdet.utils.download import get_config_path + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = [ + 'list_model', 'get_config_file', 'get_weights_url', 'get_model', + 'MODEL_ZOO_FILENAME' +] + +MODEL_ZOO_FILENAME = 'MODEL_ZOO' + + +def list_model(filters=[]): + model_zoo_file = pkg_resources.resource_filename('ppdet.model_zoo', + MODEL_ZOO_FILENAME) + with open(model_zoo_file) as f: + model_names = f.read().splitlines() + + # filter model_name + def filt(name): + for f in filters: + if name.find(f) < 0: + return False + return True + + if isinstance(filters, str) or not isinstance(filters, Sequence): + filters = [filters] + model_names = [name for name in model_names if filt(name)] + if len(model_names) == 0 and len(filters) > 0: + raise ValueError("no model found, please check filters seeting, " + "filters can be set as following kinds:\n" + "\tDataset: coco, voc ...\n" + "\tArchitecture: yolo, rcnn, ssd ...\n" + "\tBackbone: resnet, vgg, darknet ...\n") + + model_str = "Available Models:\n" + for model_name in model_names: + model_str += "\t{}\n".format(model_name) + logger.info(model_str) + + +# models and configs save on bcebos under dygraph directory +def get_config_file(model_name): + return get_config_path("ppdet://configs/{}.yml".format(model_name)) + + +def get_weights_url(model_name): + return "ppdet://models/{}.pdparams".format(osp.split(model_name)[-1]) + + +def get_model(model_name, pretrained=True): + cfg_file = get_config_file(model_name) + cfg = load_config(cfg_file) + model = create(cfg.architecture) + + if pretrained: + load_weight(model, get_weights_url(model_name)) + + return model diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/__init__.py new file mode 100644 index 000000000..6f0ea8534 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/__init__.py @@ -0,0 +1,13 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_get_model.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_get_model.py new file mode 100644 index 000000000..8887185e0 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_get_model.py @@ -0,0 +1,48 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import paddle +import ppdet +import unittest + +# NOTE: weights downloading costs time, we choose +# a small model for unittesting +MODEL_NAME = 'ppyolo/ppyolo_tiny_650e_coco' + + +class TestGetConfigFile(unittest.TestCase): + def test_main(self): + try: + cfg_file = ppdet.model_zoo.get_config_file(MODEL_NAME) + assert os.path.isfile(cfg_file) + except: + self.assertTrue(False) + + +class TestGetModel(unittest.TestCase): + def test_main(self): + try: + model = ppdet.model_zoo.get_model(MODEL_NAME) + assert isinstance(model, paddle.nn.Layer) + except: + self.assertTrue(False) + + +if __name__ == '__main__': + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_list_model.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_list_model.py new file mode 100644 index 000000000..8f91afe00 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/model_zoo/tests/test_list_model.py @@ -0,0 +1,68 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import unittest +import ppdet + + +class TestListModel(unittest.TestCase): + def setUp(self): + self._filter = [] + + def test_main(self): + try: + ppdet.model_zoo.list_model(self._filter) + self.assertTrue(True) + except: + self.assertTrue(False) + + +class TestListModelYOLO(TestListModel): + def setUp(self): + self._filter = ['yolo'] + + +class TestListModelRCNN(TestListModel): + def setUp(self): + self._filter = ['rcnn'] + + +class TestListModelSSD(TestListModel): + def setUp(self): + self._filter = ['ssd'] + + +class TestListModelMultiFilter(TestListModel): + def setUp(self): + self._filter = ['yolo', 'darknet'] + + +class TestListModelError(unittest.TestCase): + def setUp(self): + self._filter = ['xxx'] + + def test_main(self): + try: + ppdet.model_zoo.list_model(self._filter) + self.assertTrue(False) + except ValueError: + self.assertTrue(True) + + +if __name__ == '__main__': + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__init__.py new file mode 100644 index 000000000..cdcb5d1bf --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__init__.py @@ -0,0 +1,45 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import warnings +warnings.filterwarnings( + action='ignore', category=DeprecationWarning, module='ops') + +from . import ops +from . import backbones +from . import necks +from . import proposal_generator +from . import heads +from . import losses +from . import architectures +from . import post_process +from . import layers +from . import reid +from . import mot +from . import transformers +from . import assigners + +from .ops import * +from .backbones import * +from .necks import * +from .proposal_generator import * +from .heads import * +from .losses import * +from .architectures import * +from .post_process import * +from .layers import * +from .reid import * +from .mot import * +from .transformers import * +from .assigners import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..0d7ca5f63 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/bbox_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/bbox_utils.cpython-37.pyc new file mode 100644 index 000000000..ea3062ca1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/bbox_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/initializer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/initializer.cpython-37.pyc new file mode 100644 index 000000000..39ca157db Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/initializer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/keypoint_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/keypoint_utils.cpython-37.pyc new file mode 100644 index 000000000..482bd4353 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/keypoint_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/layers.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/layers.cpython-37.pyc new file mode 100644 index 000000000..7ccda4248 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/layers.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/ops.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/ops.cpython-37.pyc new file mode 100644 index 000000000..9a5aa8441 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/ops.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/post_process.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/post_process.cpython-37.pyc new file mode 100644 index 000000000..3ee8944b2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/post_process.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/shape_spec.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/shape_spec.cpython-37.pyc new file mode 100644 index 000000000..fe06b61b1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/__pycache__/shape_spec.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__init__.py new file mode 100644 index 000000000..b5feb06d8 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__init__.py @@ -0,0 +1,51 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +from . import meta_arch +from . import faster_rcnn +from . import mask_rcnn +from . import yolo +from . import cascade_rcnn +from . import ssd +from . import fcos +from . import solov2 +from . import ttfnet +from . import s2anet +from . import keypoint_hrhrnet +from . import keypoint_hrnet +from . import jde +from . import deepsort +from . import fairmot +from . import centernet +from . import gfl +from . import picodet +from . import detr +from . import sparse_rcnn +from . import tood + +from .meta_arch import * +from .faster_rcnn import * +from .mask_rcnn import * +from .yolo import * +from .cascade_rcnn import * +from .ssd import * +from .fcos import * +from .solov2 import * +from .ttfnet import * +from .s2anet import * +from .keypoint_hrhrnet import * +from .keypoint_hrnet import * +from .jde import * +from .deepsort import * +from .fairmot import * +from .centernet import * +from .blazeface import * +from .gfl import * +from .picodet import * +from .detr import * +from .sparse_rcnn import * +from .tood import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..8b5e1b231 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/blazeface.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/blazeface.cpython-37.pyc new file mode 100644 index 000000000..8481f88af Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/blazeface.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/cascade_rcnn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/cascade_rcnn.cpython-37.pyc new file mode 100644 index 000000000..3524b345f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/cascade_rcnn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/centernet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/centernet.cpython-37.pyc new file mode 100644 index 000000000..203ccb71f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/centernet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/deepsort.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/deepsort.cpython-37.pyc new file mode 100644 index 000000000..1530b32d4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/deepsort.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/detr.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/detr.cpython-37.pyc new file mode 100644 index 000000000..5b2a1625f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/detr.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fairmot.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fairmot.cpython-37.pyc new file mode 100644 index 000000000..7763b0b40 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fairmot.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/faster_rcnn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/faster_rcnn.cpython-37.pyc new file mode 100644 index 000000000..3727180f5 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/faster_rcnn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fcos.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fcos.cpython-37.pyc new file mode 100644 index 000000000..86ef96738 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/fcos.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/gfl.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/gfl.cpython-37.pyc new file mode 100644 index 000000000..a42fad809 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/gfl.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/jde.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/jde.cpython-37.pyc new file mode 100644 index 000000000..71c58f25d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/jde.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrhrnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrhrnet.cpython-37.pyc new file mode 100644 index 000000000..3c138dc87 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrhrnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrnet.cpython-37.pyc new file mode 100644 index 000000000..48a7b5e29 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/keypoint_hrnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/mask_rcnn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/mask_rcnn.cpython-37.pyc new file mode 100644 index 000000000..9f2423852 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/mask_rcnn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/meta_arch.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/meta_arch.cpython-37.pyc new file mode 100644 index 000000000..6ce3ee761 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/meta_arch.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/picodet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/picodet.cpython-37.pyc new file mode 100644 index 000000000..e27b87c8d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/picodet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/s2anet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/s2anet.cpython-37.pyc new file mode 100644 index 000000000..5328d43d4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/s2anet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/solov2.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/solov2.cpython-37.pyc new file mode 100644 index 000000000..38764e818 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/solov2.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/sparse_rcnn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/sparse_rcnn.cpython-37.pyc new file mode 100644 index 000000000..2c87c800e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/sparse_rcnn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ssd.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ssd.cpython-37.pyc new file mode 100644 index 000000000..e94cda5d4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ssd.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/tood.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/tood.cpython-37.pyc new file mode 100644 index 000000000..882d03fba Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/tood.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ttfnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ttfnet.cpython-37.pyc new file mode 100644 index 000000000..4033e47d7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/ttfnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/yolo.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/yolo.cpython-37.pyc new file mode 100644 index 000000000..27c693dec Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/__pycache__/yolo.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/blazeface.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/blazeface.py new file mode 100644 index 000000000..af6aa269d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/blazeface.py @@ -0,0 +1,91 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['BlazeFace'] + + +@register +class BlazeFace(BaseArch): + """ + BlazeFace: Sub-millisecond Neural Face Detection on Mobile GPUs, + see https://arxiv.org/abs/1907.05047 + + Args: + backbone (nn.Layer): backbone instance + neck (nn.Layer): neck instance + blaze_head (nn.Layer): `blazeHead` instance + post_process (object): `BBoxPostProcess` instance + """ + + __category__ = 'architecture' + __inject__ = ['post_process'] + + def __init__(self, backbone, blaze_head, neck, post_process): + super(BlazeFace, self).__init__() + self.backbone = backbone + self.neck = neck + self.blaze_head = blaze_head + self.post_process = post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + # fpn + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + # head + kwargs = {'input_shape': neck.out_shape} + blaze_head = create(cfg['blaze_head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + 'blaze_head': blaze_head, + } + + def _forward(self): + # Backbone + body_feats = self.backbone(self.inputs) + # neck + neck_feats = self.neck(body_feats) + # blaze Head + if self.training: + return self.blaze_head(neck_feats, self.inputs['image'], + self.inputs['gt_bbox'], + self.inputs['gt_class']) + else: + preds, anchors = self.blaze_head(neck_feats, self.inputs['image']) + bbox, bbox_num = self.post_process(preds, anchors, + self.inputs['im_shape'], + self.inputs['scale_factor']) + return bbox, bbox_num + + def get_loss(self, ): + return {"loss": self._forward()} + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = { + "bbox": bbox_pred, + "bbox_num": bbox_num, + } + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/cascade_rcnn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/cascade_rcnn.py new file mode 100644 index 000000000..ac29b775d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/cascade_rcnn.py @@ -0,0 +1,143 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['CascadeRCNN'] + + +@register +class CascadeRCNN(BaseArch): + """ + Cascade R-CNN network, see https://arxiv.org/abs/1712.00726 + + Args: + backbone (object): backbone instance + rpn_head (object): `RPNHead` instance + bbox_head (object): `BBoxHead` instance + bbox_post_process (object): `BBoxPostProcess` instance + neck (object): 'FPN' instance + mask_head (object): `MaskHead` instance + mask_post_process (object): `MaskPostProcess` instance + """ + __category__ = 'architecture' + __inject__ = [ + 'bbox_post_process', + 'mask_post_process', + ] + + def __init__(self, + backbone, + rpn_head, + bbox_head, + bbox_post_process, + neck=None, + mask_head=None, + mask_post_process=None): + super(CascadeRCNN, self).__init__() + self.backbone = backbone + self.rpn_head = rpn_head + self.bbox_head = bbox_head + self.bbox_post_process = bbox_post_process + self.neck = neck + self.mask_head = mask_head + self.mask_post_process = mask_post_process + self.with_mask = mask_head is not None + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + kwargs = {'input_shape': backbone.out_shape} + neck = cfg['neck'] and create(cfg['neck'], **kwargs) + + out_shape = neck and neck.out_shape or backbone.out_shape + kwargs = {'input_shape': out_shape} + rpn_head = create(cfg['rpn_head'], **kwargs) + bbox_head = create(cfg['bbox_head'], **kwargs) + + out_shape = neck and out_shape or bbox_head.get_head().out_shape + kwargs = {'input_shape': out_shape} + mask_head = cfg['mask_head'] and create(cfg['mask_head'], **kwargs) + return { + 'backbone': backbone, + 'neck': neck, + "rpn_head": rpn_head, + "bbox_head": bbox_head, + "mask_head": mask_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + if self.neck is not None: + body_feats = self.neck(body_feats) + + if self.training: + rois, rois_num, rpn_loss = self.rpn_head(body_feats, self.inputs) + bbox_loss, bbox_feat = self.bbox_head(body_feats, rois, rois_num, + self.inputs) + rois, rois_num = self.bbox_head.get_assigned_rois() + bbox_targets = self.bbox_head.get_assigned_targets() + if self.with_mask: + mask_loss = self.mask_head(body_feats, rois, rois_num, + self.inputs, bbox_targets, bbox_feat) + return rpn_loss, bbox_loss, mask_loss + else: + return rpn_loss, bbox_loss, {} + else: + rois, rois_num, _ = self.rpn_head(body_feats, self.inputs) + preds, _ = self.bbox_head(body_feats, rois, rois_num, self.inputs) + refined_rois = self.bbox_head.get_refined_rois() + + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + + bbox, bbox_num = self.bbox_post_process( + preds, (refined_rois, rois_num), im_shape, scale_factor) + # rescale the prediction back to origin image + bbox_pred = self.bbox_post_process.get_pred(bbox, bbox_num, + im_shape, scale_factor) + if not self.with_mask: + return bbox_pred, bbox_num, None + mask_out = self.mask_head(body_feats, bbox, bbox_num, self.inputs) + origin_shape = self.bbox_post_process.get_origin_shape() + mask_pred = self.mask_post_process(mask_out[:, 0, :, :], bbox_pred, + bbox_num, origin_shape) + return bbox_pred, bbox_num, mask_pred + + def get_loss(self, ): + rpn_loss, bbox_loss, mask_loss = self._forward() + loss = {} + loss.update(rpn_loss) + loss.update(bbox_loss) + if self.with_mask: + loss.update(mask_loss) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num, mask_pred = self._forward() + output = { + 'bbox': bbox_pred, + 'bbox_num': bbox_num, + } + if self.with_mask: + output.update({'mask': mask_pred}) + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/centernet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/centernet.py new file mode 100644 index 000000000..2287d743b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/centernet.py @@ -0,0 +1,108 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['CenterNet'] + + +@register +class CenterNet(BaseArch): + """ + CenterNet network, see http://arxiv.org/abs/1904.07850 + + Args: + backbone (object): backbone instance + neck (object): FPN instance, default use 'CenterNetDLAFPN' + head (object): 'CenterNetHead' instance + post_process (object): 'CenterNetPostProcess' instance + for_mot (bool): whether return other features used in tracking model + + """ + __category__ = 'architecture' + __inject__ = ['post_process'] + __shared__ = ['for_mot'] + + def __init__(self, + backbone, + neck='CenterNetDLAFPN', + head='CenterNetHead', + post_process='CenterNetPostProcess', + for_mot=False): + super(CenterNet, self).__init__() + self.backbone = backbone + self.neck = neck + self.head = head + self.post_process = post_process + self.for_mot = for_mot + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = cfg['neck'] and create(cfg['neck'], **kwargs) + + out_shape = neck and neck.out_shape or backbone.out_shape + kwargs = {'input_shape': out_shape} + head = create(cfg['head'], **kwargs) + + return {'backbone': backbone, 'neck': neck, "head": head} + + def _forward(self): + neck_feat = self.backbone(self.inputs) + if self.neck is not None: + neck_feat = self.neck(neck_feat) + head_out = self.head(neck_feat, self.inputs) + if self.for_mot: + head_out.update({'neck_feat': neck_feat}) + elif self.training: + head_out['loss'] = head_out.pop('det_loss') + return head_out + + def get_pred(self): + head_out = self._forward() + if self.for_mot: + bbox, bbox_inds, topk_clses = self.post_process( + head_out['heatmap'], + head_out['size'], + head_out['offset'], + im_shape=self.inputs['im_shape'], + scale_factor=self.inputs['scale_factor']) + output = { + "bbox": bbox, + "bbox_inds": bbox_inds, + "topk_clses": topk_clses, + "neck_feat": head_out['neck_feat'] + } + else: + bbox, bbox_num, _ = self.post_process( + head_out['heatmap'], + head_out['size'], + head_out['offset'], + im_shape=self.inputs['im_shape'], + scale_factor=self.inputs['scale_factor']) + output = { + "bbox": bbox, + "bbox_num": bbox_num, + } + return output + + def get_loss(self): + return self._forward() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/deepsort.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/deepsort.py new file mode 100644 index 000000000..066f7a4ce --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/deepsort.py @@ -0,0 +1,69 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch +from ppdet.modeling.mot.utils import Detection, get_crops, scale_coords, clip_box + +__all__ = ['DeepSORT'] + + +@register +class DeepSORT(BaseArch): + """ + DeepSORT network, see https://arxiv.org/abs/1703.07402 + + Args: + detector (object): detector model instance + reid (object): reid model instance + tracker (object): tracker instance + """ + __category__ = 'architecture' + + def __init__(self, + detector='YOLOv3', + reid='PCBPyramid', + tracker='DeepSORTTracker'): + super(DeepSORT, self).__init__() + self.detector = detector + self.reid = reid + self.tracker = tracker + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + if cfg['detector'] != 'None': + detector = create(cfg['detector']) + else: + detector = None + reid = create(cfg['reid']) + tracker = create(cfg['tracker']) + + return { + "detector": detector, + "reid": reid, + "tracker": tracker, + } + + def _forward(self): + crops = self.inputs['crops'] + features = self.reid(crops) + return features + + def get_pred(self): + return self._forward() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/detr.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/detr.py new file mode 100644 index 000000000..2c081bf6c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/detr.py @@ -0,0 +1,93 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from .meta_arch import BaseArch +from ppdet.core.workspace import register, create + +__all__ = ['DETR'] + + +@register +class DETR(BaseArch): + __category__ = 'architecture' + __inject__ = ['post_process'] + + def __init__(self, + backbone, + transformer, + detr_head, + post_process='DETRBBoxPostProcess'): + super(DETR, self).__init__() + self.backbone = backbone + self.transformer = transformer + self.detr_head = detr_head + self.post_process = post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + # transformer + kwargs = {'input_shape': backbone.out_shape} + transformer = create(cfg['transformer'], **kwargs) + # head + kwargs = { + 'hidden_dim': transformer.hidden_dim, + 'nhead': transformer.nhead, + 'input_shape': backbone.out_shape + } + detr_head = create(cfg['detr_head'], **kwargs) + + return { + 'backbone': backbone, + 'transformer': transformer, + "detr_head": detr_head, + } + + def _forward(self): + # Backbone + body_feats = self.backbone(self.inputs) + + # Transformer + out_transformer = self.transformer(body_feats, self.inputs['pad_mask']) + + # DETR Head + if self.training: + return self.detr_head(out_transformer, body_feats, self.inputs) + else: + preds = self.detr_head(out_transformer, body_feats) + bbox, bbox_num = self.post_process(preds, self.inputs['im_shape'], + self.inputs['scale_factor']) + return bbox, bbox_num + + def get_loss(self, ): + losses = self._forward() + losses.update({ + 'loss': + paddle.add_n([v for k, v in losses.items() if 'log' not in k]) + }) + return losses + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = { + "bbox": bbox_pred, + "bbox_num": bbox_num, + } + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fairmot.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fairmot.py new file mode 100644 index 000000000..271450839 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fairmot.py @@ -0,0 +1,100 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['FairMOT'] + + +@register +class FairMOT(BaseArch): + """ + FairMOT network, see http://arxiv.org/abs/2004.01888 + + Args: + detector (object): 'CenterNet' instance + reid (object): 'FairMOTEmbeddingHead' instance + tracker (object): 'JDETracker' instance + loss (object): 'FairMOTLoss' instance + + """ + + __category__ = 'architecture' + __inject__ = ['loss'] + + def __init__(self, + detector='CenterNet', + reid='FairMOTEmbeddingHead', + tracker='JDETracker', + loss='FairMOTLoss'): + super(FairMOT, self).__init__() + self.detector = detector + self.reid = reid + self.tracker = tracker + self.loss = loss + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + detector = create(cfg['detector']) + detector_out_shape = detector.neck and detector.neck.out_shape or detector.backbone.out_shape + + kwargs = {'input_shape': detector_out_shape} + reid = create(cfg['reid'], **kwargs) + loss = create(cfg['loss']) + tracker = create(cfg['tracker']) + + return { + 'detector': detector, + 'reid': reid, + 'loss': loss, + 'tracker': tracker + } + + def _forward(self): + loss = dict() + # det_outs keys: + # train: neck_feat, det_loss, heatmap_loss, size_loss, offset_loss (optional: iou_loss) + # eval/infer: neck_feat, bbox, bbox_inds + det_outs = self.detector(self.inputs) + neck_feat = det_outs['neck_feat'] + if self.training: + reid_loss = self.reid(neck_feat, self.inputs) + + det_loss = det_outs['det_loss'] + loss = self.loss(det_loss, reid_loss) + for k, v in det_outs.items(): + if 'loss' not in k: + continue + loss.update({k: v}) + loss.update({'reid_loss': reid_loss}) + return loss + else: + pred_dets, pred_embs = self.reid( + neck_feat, self.inputs, det_outs['bbox'], det_outs['bbox_inds'], + det_outs['topk_clses']) + return pred_dets, pred_embs + + def get_pred(self): + output = self._forward() + return output + + def get_loss(self): + loss = self._forward() + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/faster_rcnn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/faster_rcnn.py new file mode 100644 index 000000000..26a2672d6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/faster_rcnn.py @@ -0,0 +1,106 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['FasterRCNN'] + + +@register +class FasterRCNN(BaseArch): + """ + Faster R-CNN network, see https://arxiv.org/abs/1506.01497 + + Args: + backbone (object): backbone instance + rpn_head (object): `RPNHead` instance + bbox_head (object): `BBoxHead` instance + bbox_post_process (object): `BBoxPostProcess` instance + neck (object): 'FPN' instance + """ + __category__ = 'architecture' + __inject__ = ['bbox_post_process'] + + def __init__(self, + backbone, + rpn_head, + bbox_head, + bbox_post_process, + neck=None): + super(FasterRCNN, self).__init__() + self.backbone = backbone + self.neck = neck + self.rpn_head = rpn_head + self.bbox_head = bbox_head + self.bbox_post_process = bbox_post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + kwargs = {'input_shape': backbone.out_shape} + neck = cfg['neck'] and create(cfg['neck'], **kwargs) + + out_shape = neck and neck.out_shape or backbone.out_shape + kwargs = {'input_shape': out_shape} + rpn_head = create(cfg['rpn_head'], **kwargs) + bbox_head = create(cfg['bbox_head'], **kwargs) + return { + 'backbone': backbone, + 'neck': neck, + "rpn_head": rpn_head, + "bbox_head": bbox_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + if self.neck is not None: + body_feats = self.neck(body_feats) + if self.training: + rois, rois_num, rpn_loss = self.rpn_head(body_feats, self.inputs) + bbox_loss, _ = self.bbox_head(body_feats, rois, rois_num, + self.inputs) + return rpn_loss, bbox_loss + else: + rois, rois_num, _ = self.rpn_head(body_feats, self.inputs) + preds, _ = self.bbox_head(body_feats, rois, rois_num, None) + + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + bbox, bbox_num = self.bbox_post_process(preds, (rois, rois_num), + im_shape, scale_factor) + + # rescale the prediction back to origin image + bbox_pred = self.bbox_post_process.get_pred(bbox, bbox_num, + im_shape, scale_factor) + return bbox_pred, bbox_num + + def get_loss(self, ): + rpn_loss, bbox_loss = self._forward() + loss = {} + loss.update(rpn_loss) + loss.update(bbox_loss) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fcos.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fcos.py new file mode 100644 index 000000000..8fa5c569b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/fcos.py @@ -0,0 +1,105 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['FCOS'] + + +@register +class FCOS(BaseArch): + """ + FCOS network, see https://arxiv.org/abs/1904.01355 + + Args: + backbone (object): backbone instance + neck (object): 'FPN' instance + fcos_head (object): 'FCOSHead' instance + post_process (object): 'FCOSPostProcess' instance + """ + + __category__ = 'architecture' + __inject__ = ['fcos_post_process'] + + def __init__(self, + backbone, + neck, + fcos_head='FCOSHead', + fcos_post_process='FCOSPostProcess'): + super(FCOS, self).__init__() + self.backbone = backbone + self.neck = neck + self.fcos_head = fcos_head + self.fcos_post_process = fcos_post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + fcos_head = create(cfg['fcos_head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "fcos_head": fcos_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + fpn_feats = self.neck(body_feats) + fcos_head_outs = self.fcos_head(fpn_feats, self.training) + if not self.training: + scale_factor = self.inputs['scale_factor'] + bboxes = self.fcos_post_process(fcos_head_outs, scale_factor) + return bboxes + else: + return fcos_head_outs + + def get_loss(self, ): + loss = {} + tag_labels, tag_bboxes, tag_centerness = [], [], [] + for i in range(len(self.fcos_head.fpn_stride)): + # labels, reg_target, centerness + k_lbl = 'labels{}'.format(i) + if k_lbl in self.inputs: + tag_labels.append(self.inputs[k_lbl]) + k_box = 'reg_target{}'.format(i) + if k_box in self.inputs: + tag_bboxes.append(self.inputs[k_box]) + k_ctn = 'centerness{}'.format(i) + if k_ctn in self.inputs: + tag_centerness.append(self.inputs[k_ctn]) + + fcos_head_outs = self._forward() + loss_fcos = self.fcos_head.get_loss(fcos_head_outs, tag_labels, + tag_bboxes, tag_centerness) + loss.update(loss_fcos) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/gfl.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/gfl.py new file mode 100644 index 000000000..91c13077f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/gfl.py @@ -0,0 +1,87 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['GFL'] + + +@register +class GFL(BaseArch): + """ + Generalized Focal Loss network, see https://arxiv.org/abs/2006.04388 + + Args: + backbone (object): backbone instance + neck (object): 'FPN' instance + head (object): 'GFLHead' instance + """ + + __category__ = 'architecture' + + def __init__(self, backbone, neck, head='GFLHead'): + super(GFL, self).__init__() + self.backbone = backbone + self.neck = neck + self.head = head + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + head = create(cfg['head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "head": head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + fpn_feats = self.neck(body_feats) + head_outs = self.head(fpn_feats) + if not self.training: + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + bboxes, bbox_num = self.head.post_process(head_outs, im_shape, + scale_factor) + return bboxes, bbox_num + else: + return head_outs + + def get_loss(self, ): + loss = {} + + head_outs = self._forward() + loss_gfl = self.head.get_loss(head_outs, self.inputs) + loss.update(loss_gfl) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/jde.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/jde.py new file mode 100644 index 000000000..11b45c8c1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/jde.py @@ -0,0 +1,110 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['JDE'] + + +@register +class JDE(BaseArch): + __category__ = 'architecture' + __shared__ = ['metric'] + """ + JDE network, see https://arxiv.org/abs/1909.12605v1 + + Args: + detector (object): detector model instance + reid (object): reid model instance + tracker (object): tracker instance + metric (str): 'MOTDet' for training and detection evaluation, 'ReID' + for ReID embedding evaluation, or 'MOT' for multi object tracking + evaluation. + """ + + def __init__(self, + detector='YOLOv3', + reid='JDEEmbeddingHead', + tracker='JDETracker', + metric='MOT'): + super(JDE, self).__init__() + self.detector = detector + self.reid = reid + self.tracker = tracker + self.metric = metric + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + detector = create(cfg['detector']) + kwargs = {'input_shape': detector.neck.out_shape} + + reid = create(cfg['reid'], **kwargs) + + tracker = create(cfg['tracker']) + + return { + "detector": detector, + "reid": reid, + "tracker": tracker, + } + + def _forward(self): + det_outs = self.detector(self.inputs) + + if self.training: + emb_feats = det_outs['emb_feats'] + loss_confs = det_outs['det_losses']['loss_confs'] + loss_boxes = det_outs['det_losses']['loss_boxes'] + jde_losses = self.reid( + emb_feats, + self.inputs, + loss_confs=loss_confs, + loss_boxes=loss_boxes) + return jde_losses + else: + if self.metric == 'MOTDet': + det_results = { + 'bbox': det_outs['bbox'], + 'bbox_num': det_outs['bbox_num'], + } + return det_results + + elif self.metric == 'MOT': + emb_feats = det_outs['emb_feats'] + bboxes = det_outs['bbox'] + boxes_idx = det_outs['boxes_idx'] + nms_keep_idx = det_outs['nms_keep_idx'] + + pred_dets, pred_embs = self.reid( + emb_feats, + self.inputs, + bboxes=bboxes, + boxes_idx=boxes_idx, + nms_keep_idx=nms_keep_idx) + return pred_dets, pred_embs + + else: + raise ValueError("Unknown metric {} for multi object tracking.". + format(self.metric)) + + def get_loss(self): + return self._forward() + + def get_pred(self): + return self._forward() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrhrnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrhrnet.py new file mode 100644 index 000000000..6f62b4b21 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrhrnet.py @@ -0,0 +1,287 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from scipy.optimize import linear_sum_assignment +from collections import abc, defaultdict +import numpy as np +import paddle + +from ppdet.core.workspace import register, create, serializable +from .meta_arch import BaseArch +from .. import layers as L +from ..keypoint_utils import transpred + +__all__ = ['HigherHRNet'] + + +@register +class HigherHRNet(BaseArch): + __category__ = 'architecture' + + def __init__(self, + backbone='HRNet', + hrhrnet_head='HrHRNetHead', + post_process='HrHRNetPostProcess', + eval_flip=True, + flip_perm=None, + max_num_people=30): + """ + HigherHRNet network, see https://arxiv.org/abs/1908.10357锛 + HigherHRNet+swahr, see https://arxiv.org/abs/2012.15175 + + Args: + backbone (nn.Layer): backbone instance + hrhrnet_head (nn.Layer): keypoint_head instance + bbox_post_process (object): `BBoxPostProcess` instance + """ + super(HigherHRNet, self).__init__() + self.backbone = backbone + self.hrhrnet_head = hrhrnet_head + self.post_process = post_process + self.flip = eval_flip + self.flip_perm = paddle.to_tensor(flip_perm) + self.deploy = False + self.interpolate = L.Upsample(2, mode='bilinear') + self.pool = L.MaxPool(5, 1, 2) + self.max_num_people = max_num_people + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + # head + kwargs = {'input_shape': backbone.out_shape} + hrhrnet_head = create(cfg['hrhrnet_head'], **kwargs) + post_process = create(cfg['post_process']) + + return { + 'backbone': backbone, + "hrhrnet_head": hrhrnet_head, + "post_process": post_process, + } + + def _forward(self): + if self.flip and not self.training and not self.deploy: + self.inputs['image'] = paddle.concat( + (self.inputs['image'], paddle.flip(self.inputs['image'], [3]))) + body_feats = self.backbone(self.inputs) + + if self.training: + return self.hrhrnet_head(body_feats, self.inputs) + else: + outputs = self.hrhrnet_head(body_feats) + + if self.flip and not self.deploy: + outputs = [paddle.split(o, 2) for o in outputs] + output_rflip = [ + paddle.flip(paddle.gather(o[1], self.flip_perm, 1), [3]) + for o in outputs + ] + output1 = [o[0] for o in outputs] + heatmap = (output1[0] + output_rflip[0]) / 2. + tagmaps = [output1[1], output_rflip[1]] + outputs = [heatmap] + tagmaps + outputs = self.get_topk(outputs) + + if self.deploy: + return outputs + + res_lst = [] + h = self.inputs['im_shape'][0, 0].numpy().item() + w = self.inputs['im_shape'][0, 1].numpy().item() + kpts, scores = self.post_process(*outputs, h, w) + res_lst.append([kpts, scores]) + return res_lst + + def get_loss(self): + return self._forward() + + def get_pred(self): + outputs = {} + res_lst = self._forward() + outputs['keypoint'] = res_lst + return outputs + + def get_topk(self, outputs): + # resize to image size + outputs = [self.interpolate(x) for x in outputs] + if len(outputs) == 3: + tagmap = paddle.concat( + (outputs[1].unsqueeze(4), outputs[2].unsqueeze(4)), axis=4) + else: + tagmap = outputs[1].unsqueeze(4) + + heatmap = outputs[0] + N, J = 1, self.hrhrnet_head.num_joints + heatmap_maxpool = self.pool(heatmap) + # topk + maxmap = heatmap * (heatmap == heatmap_maxpool) + maxmap = maxmap.reshape([N, J, -1]) + heat_k, inds_k = maxmap.topk(self.max_num_people, axis=2) + + outputs = [heatmap, tagmap, heat_k, inds_k] + return outputs + + +@register +@serializable +class HrHRNetPostProcess(object): + ''' + HrHRNet postprocess contain: + 1) get topk keypoints in the output heatmap + 2) sample the tagmap's value corresponding to each of the topk coordinate + 3) match different joints to combine to some people with Hungary algorithm + 4) adjust the coordinate by +-0.25 to decrease error std + 5) salvage missing joints by check positivity of heatmap - tagdiff_norm + Args: + max_num_people (int): max number of people support in postprocess + heat_thresh (float): value of topk below this threshhold will be ignored + tag_thresh (float): coord's value sampled in tagmap below this threshold belong to same people for init + + inputs(list[heatmap]): the output list of modle, [heatmap, heatmap_maxpool, tagmap], heatmap_maxpool used to get topk + original_height, original_width (float): the original image size + ''' + + def __init__(self, max_num_people=30, heat_thresh=0.1, tag_thresh=1.): + self.max_num_people = max_num_people + self.heat_thresh = heat_thresh + self.tag_thresh = tag_thresh + + def lerp(self, j, y, x, heatmap): + H, W = heatmap.shape[-2:] + left = np.clip(x - 1, 0, W - 1) + right = np.clip(x + 1, 0, W - 1) + up = np.clip(y - 1, 0, H - 1) + down = np.clip(y + 1, 0, H - 1) + offset_y = np.where(heatmap[j, down, x] > heatmap[j, up, x], 0.25, + -0.25) + offset_x = np.where(heatmap[j, y, right] > heatmap[j, y, left], 0.25, + -0.25) + return offset_y + 0.5, offset_x + 0.5 + + def __call__(self, heatmap, tagmap, heat_k, inds_k, original_height, + original_width): + + N, J, H, W = heatmap.shape + assert N == 1, "only support batch size 1" + heatmap = heatmap[0].cpu().detach().numpy() + tagmap = tagmap[0].cpu().detach().numpy() + heats = heat_k[0].cpu().detach().numpy() + inds_np = inds_k[0].cpu().detach().numpy() + y = inds_np // W + x = inds_np % W + tags = tagmap[np.arange(J)[None, :].repeat(self.max_num_people), + y.flatten(), x.flatten()].reshape(J, -1, tagmap.shape[-1]) + coords = np.stack((y, x), axis=2) + # threshold + mask = heats > self.heat_thresh + # cluster + cluster = defaultdict(lambda: { + 'coords': np.zeros((J, 2), dtype=np.float32), + 'scores': np.zeros(J, dtype=np.float32), + 'tags': [] + }) + for jid, m in enumerate(mask): + num_valid = m.sum() + if num_valid == 0: + continue + valid_inds = np.where(m)[0] + valid_tags = tags[jid, m, :] + if len(cluster) == 0: # initialize + for i in valid_inds: + tag = tags[jid, i] + key = tag[0] + cluster[key]['tags'].append(tag) + cluster[key]['scores'][jid] = heats[jid, i] + cluster[key]['coords'][jid] = coords[jid, i] + continue + candidates = list(cluster.keys())[:self.max_num_people] + centroids = [ + np.mean( + cluster[k]['tags'], axis=0) for k in candidates + ] + num_clusters = len(centroids) + # shape is (num_valid, num_clusters, tag_dim) + dist = valid_tags[:, None, :] - np.array(centroids)[None, ...] + l2_dist = np.linalg.norm(dist, ord=2, axis=2) + # modulate dist with heat value, see `use_detection_val` + cost = np.round(l2_dist) * 100 - heats[jid, m, None] + # pad the cost matrix, otherwise new pose are ignored + if num_valid > num_clusters: + cost = np.pad(cost, ((0, 0), (0, num_valid - num_clusters)), + 'constant', + constant_values=((0, 0), (0, 1e-10))) + rows, cols = linear_sum_assignment(cost) + for y, x in zip(rows, cols): + tag = tags[jid, y] + if y < num_valid and x < num_clusters and \ + l2_dist[y, x] < self.tag_thresh: + key = candidates[x] # merge to cluster + else: + key = tag[0] # initialize new cluster + cluster[key]['tags'].append(tag) + cluster[key]['scores'][jid] = heats[jid, y] + cluster[key]['coords'][jid] = coords[jid, y] + + # shape is [k, J, 2] and [k, J] + pose_tags = np.array([cluster[k]['tags'] for k in cluster]) + pose_coords = np.array([cluster[k]['coords'] for k in cluster]) + pose_scores = np.array([cluster[k]['scores'] for k in cluster]) + valid = pose_scores > 0 + + pose_kpts = np.zeros((pose_scores.shape[0], J, 3), dtype=np.float32) + if valid.sum() == 0: + return pose_kpts, pose_kpts + + # refine coords + valid_coords = pose_coords[valid].astype(np.int32) + y = valid_coords[..., 0].flatten() + x = valid_coords[..., 1].flatten() + _, j = np.nonzero(valid) + offsets = self.lerp(j, y, x, heatmap) + pose_coords[valid, 0] += offsets[0] + pose_coords[valid, 1] += offsets[1] + + # mean score before salvage + mean_score = pose_scores.mean(axis=1) + pose_kpts[valid, 2] = pose_scores[valid] + + # salvage missing joints + if True: + for pid, coords in enumerate(pose_coords): + tag_mean = np.array(pose_tags[pid]).mean(axis=0) + norm = np.sum((tagmap - tag_mean)**2, axis=3)**0.5 + score = heatmap - np.round(norm) # (J, H, W) + flat_score = score.reshape(J, -1) + max_inds = np.argmax(flat_score, axis=1) + max_scores = np.max(flat_score, axis=1) + salvage_joints = (pose_scores[pid] == 0) & (max_scores > 0) + if salvage_joints.sum() == 0: + continue + y = max_inds[salvage_joints] // W + x = max_inds[salvage_joints] % W + offsets = self.lerp(salvage_joints.nonzero()[0], y, x, heatmap) + y = y.astype(np.float32) + offsets[0] + x = x.astype(np.float32) + offsets[1] + pose_coords[pid][salvage_joints, 0] = y + pose_coords[pid][salvage_joints, 1] = x + pose_kpts[pid][salvage_joints, 2] = max_scores[salvage_joints] + pose_kpts[..., :2] = transpred(pose_coords[..., :2][..., ::-1], + original_height, original_width, + min(H, W)) + return pose_kpts, mean_score diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrnet.py new file mode 100644 index 000000000..914bd043c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/keypoint_hrnet.py @@ -0,0 +1,267 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import numpy as np +import math +import cv2 +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch +from ..keypoint_utils import transform_preds +from .. import layers as L + +__all__ = ['TopDownHRNet'] + + +@register +class TopDownHRNet(BaseArch): + __category__ = 'architecture' + __inject__ = ['loss'] + + def __init__(self, + width, + num_joints, + backbone='HRNet', + loss='KeyPointMSELoss', + post_process='HRNetPostProcess', + flip_perm=None, + flip=True, + shift_heatmap=True, + use_dark=True): + """ + HRNet network, see https://arxiv.org/abs/1902.09212 + + Args: + backbone (nn.Layer): backbone instance + post_process (object): `HRNetPostProcess` instance + flip_perm (list): The left-right joints exchange order list + use_dark(bool): Whether to use DARK in post processing + """ + super(TopDownHRNet, self).__init__() + self.backbone = backbone + self.post_process = HRNetPostProcess(use_dark) + self.loss = loss + self.flip_perm = flip_perm + self.flip = flip + self.final_conv = L.Conv2d(width, num_joints, 1, 1, 0, bias=True) + self.shift_heatmap = shift_heatmap + self.deploy = False + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + + return {'backbone': backbone, } + + def _forward(self): + feats = self.backbone(self.inputs) + hrnet_outputs = self.final_conv(feats[0]) + + if self.training: + return self.loss(hrnet_outputs, self.inputs) + elif self.deploy: + outshape = hrnet_outputs.shape + max_idx = paddle.argmax( + hrnet_outputs.reshape( + (outshape[0], outshape[1], outshape[2] * outshape[3])), + axis=-1) + return hrnet_outputs, max_idx + else: + if self.flip: + self.inputs['image'] = self.inputs['image'].flip([3]) + feats = self.backbone(self.inputs) + output_flipped = self.final_conv(feats[0]) + output_flipped = self.flip_back(output_flipped.numpy(), + self.flip_perm) + output_flipped = paddle.to_tensor(output_flipped.copy()) + if self.shift_heatmap: + output_flipped[:, :, :, 1:] = output_flipped.clone( + )[:, :, :, 0:-1] + hrnet_outputs = (hrnet_outputs + output_flipped) * 0.5 + imshape = (self.inputs['im_shape'].numpy() + )[:, ::-1] if 'im_shape' in self.inputs else None + center = self.inputs['center'].numpy( + ) if 'center' in self.inputs else np.round(imshape / 2.) + scale = self.inputs['scale'].numpy( + ) if 'scale' in self.inputs else imshape / 200. + outputs = self.post_process(hrnet_outputs, center, scale) + return outputs + + def get_loss(self): + return self._forward() + + def get_pred(self): + res_lst = self._forward() + outputs = {'keypoint': res_lst} + return outputs + + def flip_back(self, output_flipped, matched_parts): + assert output_flipped.ndim == 4,\ + 'output_flipped should be [batch_size, num_joints, height, width]' + + output_flipped = output_flipped[:, :, :, ::-1] + + for pair in matched_parts: + tmp = output_flipped[:, pair[0], :, :].copy() + output_flipped[:, pair[0], :, :] = output_flipped[:, pair[1], :, :] + output_flipped[:, pair[1], :, :] = tmp + + return output_flipped + + +class HRNetPostProcess(object): + def __init__(self, use_dark=True): + self.use_dark = use_dark + + def get_max_preds(self, heatmaps): + '''get predictions from score maps + + Args: + heatmaps: numpy.ndarray([batch_size, num_joints, height, width]) + + Returns: + preds: numpy.ndarray([batch_size, num_joints, 2]), keypoints coords + maxvals: numpy.ndarray([batch_size, num_joints, 2]), the maximum confidence of the keypoints + ''' + assert isinstance(heatmaps, + np.ndarray), 'heatmaps should be numpy.ndarray' + assert heatmaps.ndim == 4, 'batch_images should be 4-ndim' + + batch_size = heatmaps.shape[0] + num_joints = heatmaps.shape[1] + width = heatmaps.shape[3] + heatmaps_reshaped = heatmaps.reshape((batch_size, num_joints, -1)) + idx = np.argmax(heatmaps_reshaped, 2) + maxvals = np.amax(heatmaps_reshaped, 2) + + maxvals = maxvals.reshape((batch_size, num_joints, 1)) + idx = idx.reshape((batch_size, num_joints, 1)) + + preds = np.tile(idx, (1, 1, 2)).astype(np.float32) + + preds[:, :, 0] = (preds[:, :, 0]) % width + preds[:, :, 1] = np.floor((preds[:, :, 1]) / width) + + pred_mask = np.tile(np.greater(maxvals, 0.0), (1, 1, 2)) + pred_mask = pred_mask.astype(np.float32) + + preds *= pred_mask + + return preds, maxvals + + def gaussian_blur(self, heatmap, kernel): + border = (kernel - 1) // 2 + batch_size = heatmap.shape[0] + num_joints = heatmap.shape[1] + height = heatmap.shape[2] + width = heatmap.shape[3] + for i in range(batch_size): + for j in range(num_joints): + origin_max = np.max(heatmap[i, j]) + dr = np.zeros((height + 2 * border, width + 2 * border)) + dr[border:-border, border:-border] = heatmap[i, j].copy() + dr = cv2.GaussianBlur(dr, (kernel, kernel), 0) + heatmap[i, j] = dr[border:-border, border:-border].copy() + heatmap[i, j] *= origin_max / np.max(heatmap[i, j]) + return heatmap + + def dark_parse(self, hm, coord): + heatmap_height = hm.shape[0] + heatmap_width = hm.shape[1] + px = int(coord[0]) + py = int(coord[1]) + if 1 < px < heatmap_width - 2 and 1 < py < heatmap_height - 2: + dx = 0.5 * (hm[py][px + 1] - hm[py][px - 1]) + dy = 0.5 * (hm[py + 1][px] - hm[py - 1][px]) + dxx = 0.25 * (hm[py][px + 2] - 2 * hm[py][px] + hm[py][px - 2]) + dxy = 0.25 * (hm[py+1][px+1] - hm[py-1][px+1] - hm[py+1][px-1] \ + + hm[py-1][px-1]) + dyy = 0.25 * ( + hm[py + 2 * 1][px] - 2 * hm[py][px] + hm[py - 2 * 1][px]) + derivative = np.matrix([[dx], [dy]]) + hessian = np.matrix([[dxx, dxy], [dxy, dyy]]) + if dxx * dyy - dxy**2 != 0: + hessianinv = hessian.I + offset = -hessianinv * derivative + offset = np.squeeze(np.array(offset.T), axis=0) + coord += offset + return coord + + def dark_postprocess(self, hm, coords, kernelsize): + '''DARK postpocessing, Zhang et al. Distribution-Aware Coordinate + Representation for Human Pose Estimation (CVPR 2020). + ''' + + hm = self.gaussian_blur(hm, kernelsize) + hm = np.maximum(hm, 1e-10) + hm = np.log(hm) + for n in range(coords.shape[0]): + for p in range(coords.shape[1]): + coords[n, p] = self.dark_parse(hm[n][p], coords[n][p]) + return coords + + def get_final_preds(self, heatmaps, center, scale, kernelsize=3): + """the highest heatvalue location with a quarter offset in the + direction from the highest response to the second highest response. + + Args: + heatmaps (numpy.ndarray): The predicted heatmaps + center (numpy.ndarray): The boxes center + scale (numpy.ndarray): The scale factor + + Returns: + preds: numpy.ndarray([batch_size, num_joints, 2]), keypoints coords + maxvals: numpy.ndarray([batch_size, num_joints, 1]), the maximum confidence of the keypoints + """ + coords, maxvals = self.get_max_preds(heatmaps) + + heatmap_height = heatmaps.shape[2] + heatmap_width = heatmaps.shape[3] + + if self.use_dark: + coords = self.dark_postprocess(heatmaps, coords, kernelsize) + else: + for n in range(coords.shape[0]): + for p in range(coords.shape[1]): + hm = heatmaps[n][p] + px = int(math.floor(coords[n][p][0] + 0.5)) + py = int(math.floor(coords[n][p][1] + 0.5)) + if 1 < px < heatmap_width - 1 and 1 < py < heatmap_height - 1: + diff = np.array([ + hm[py][px + 1] - hm[py][px - 1], + hm[py + 1][px] - hm[py - 1][px] + ]) + coords[n][p] += np.sign(diff) * .25 + preds = coords.copy() + + # Transform back + for i in range(coords.shape[0]): + preds[i] = transform_preds(coords[i], center[i], scale[i], + [heatmap_width, heatmap_height]) + + return preds, maxvals + + def __call__(self, output, center, scale): + preds, maxvals = self.get_final_preds(output.numpy(), center, scale) + outputs = [[ + np.concatenate( + (preds, maxvals), axis=-1), np.mean( + maxvals, axis=1) + ]] + return outputs diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/mask_rcnn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/mask_rcnn.py new file mode 100644 index 000000000..071a326f4 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/mask_rcnn.py @@ -0,0 +1,135 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['MaskRCNN'] + + +@register +class MaskRCNN(BaseArch): + """ + Mask R-CNN network, see https://arxiv.org/abs/1703.06870 + + Args: + backbone (object): backbone instance + rpn_head (object): `RPNHead` instance + bbox_head (object): `BBoxHead` instance + mask_head (object): `MaskHead` instance + bbox_post_process (object): `BBoxPostProcess` instance + mask_post_process (object): `MaskPostProcess` instance + neck (object): 'FPN' instance + """ + + __category__ = 'architecture' + __inject__ = [ + 'bbox_post_process', + 'mask_post_process', + ] + + def __init__(self, + backbone, + rpn_head, + bbox_head, + mask_head, + bbox_post_process, + mask_post_process, + neck=None): + super(MaskRCNN, self).__init__() + self.backbone = backbone + self.neck = neck + self.rpn_head = rpn_head + self.bbox_head = bbox_head + self.mask_head = mask_head + + self.bbox_post_process = bbox_post_process + self.mask_post_process = mask_post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + kwargs = {'input_shape': backbone.out_shape} + neck = cfg['neck'] and create(cfg['neck'], **kwargs) + + out_shape = neck and neck.out_shape or backbone.out_shape + kwargs = {'input_shape': out_shape} + rpn_head = create(cfg['rpn_head'], **kwargs) + bbox_head = create(cfg['bbox_head'], **kwargs) + + out_shape = neck and out_shape or bbox_head.get_head().out_shape + kwargs = {'input_shape': out_shape} + mask_head = create(cfg['mask_head'], **kwargs) + return { + 'backbone': backbone, + 'neck': neck, + "rpn_head": rpn_head, + "bbox_head": bbox_head, + "mask_head": mask_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + if self.neck is not None: + body_feats = self.neck(body_feats) + + if self.training: + rois, rois_num, rpn_loss = self.rpn_head(body_feats, self.inputs) + bbox_loss, bbox_feat = self.bbox_head(body_feats, rois, rois_num, + self.inputs) + rois, rois_num = self.bbox_head.get_assigned_rois() + bbox_targets = self.bbox_head.get_assigned_targets() + # Mask Head needs bbox_feat in Mask RCNN + mask_loss = self.mask_head(body_feats, rois, rois_num, self.inputs, + bbox_targets, bbox_feat) + return rpn_loss, bbox_loss, mask_loss + else: + rois, rois_num, _ = self.rpn_head(body_feats, self.inputs) + preds, feat_func = self.bbox_head(body_feats, rois, rois_num, None) + + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + + bbox, bbox_num = self.bbox_post_process(preds, (rois, rois_num), + im_shape, scale_factor) + mask_out = self.mask_head( + body_feats, bbox, bbox_num, self.inputs, feat_func=feat_func) + + # rescale the prediction back to origin image + bbox_pred = self.bbox_post_process.get_pred(bbox, bbox_num, + im_shape, scale_factor) + origin_shape = self.bbox_post_process.get_origin_shape() + mask_pred = self.mask_post_process(mask_out[:, 0, :, :], bbox_pred, + bbox_num, origin_shape) + return bbox_pred, bbox_num, mask_pred + + def get_loss(self, ): + bbox_loss, mask_loss, rpn_loss = self._forward() + loss = {} + loss.update(rpn_loss) + loss.update(bbox_loss) + loss.update(mask_loss) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num, mask_pred = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num, 'mask': mask_pred} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/meta_arch.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/meta_arch.py new file mode 100644 index 000000000..d9875e183 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/meta_arch.py @@ -0,0 +1,72 @@ +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +from ppdet.core.workspace import register + +__all__ = ['BaseArch'] + + +@register +class BaseArch(nn.Layer): + def __init__(self, data_format='NCHW'): + super(BaseArch, self).__init__() + self.data_format = data_format + self.inputs = {} + self.fuse_norm = False + + def load_meanstd(self, cfg_transform): + self.scale = 1. + self.mean = paddle.to_tensor([0.485, 0.456, 0.406]).reshape( + (1, 3, 1, 1)) + self.std = paddle.to_tensor([0.229, 0.224, 0.225]).reshape((1, 3, 1, 1)) + for item in cfg_transform: + if 'NormalizeImage' in item: + self.mean = paddle.to_tensor(item['NormalizeImage'][ + 'mean']).reshape((1, 3, 1, 1)) + self.std = paddle.to_tensor(item['NormalizeImage'][ + 'std']).reshape((1, 3, 1, 1)) + if item['NormalizeImage'].get('is_scale', True): + self.scale = 1. / 255. + break + if self.data_format == 'NHWC': + self.mean = self.mean.reshape(1, 1, 1, 3) + self.std = self.std.reshape(1, 1, 1, 3) + + def forward(self, inputs): + if self.data_format == 'NHWC': + image = inputs['image'] + inputs['image'] = paddle.transpose(image, [0, 2, 3, 1]) + + if self.fuse_norm: + image = inputs['image'] + self.inputs['image'] = (image * self.scale - self.mean) / self.std + self.inputs['im_shape'] = inputs['im_shape'] + self.inputs['scale_factor'] = inputs['scale_factor'] + else: + self.inputs = inputs + + self.model_arch() + + if self.training: + out = self.get_loss() + else: + out = self.get_pred() + return out + + def build_inputs(self, data, input_def): + inputs = {} + for i, k in enumerate(input_def): + inputs[k] = data[i] + return inputs + + def model_arch(self, ): + pass + + def get_loss(self, ): + raise NotImplementedError("Should implement get_loss method!") + + def get_pred(self, ): + raise NotImplementedError("Should implement get_pred method!") diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/picodet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/picodet.py new file mode 100644 index 000000000..cd807a9fa --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/picodet.py @@ -0,0 +1,91 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['PicoDet'] + + +@register +class PicoDet(BaseArch): + """ + Generalized Focal Loss network, see https://arxiv.org/abs/2006.04388 + + Args: + backbone (object): backbone instance + neck (object): 'FPN' instance + head (object): 'PicoHead' instance + """ + + __category__ = 'architecture' + + def __init__(self, backbone, neck, head='PicoHead'): + super(PicoDet, self).__init__() + self.backbone = backbone + self.neck = neck + self.head = head + self.deploy = False + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + head = create(cfg['head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "head": head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + fpn_feats = self.neck(body_feats) + head_outs = self.head(fpn_feats, self.deploy) + if self.training or self.deploy: + return head_outs, None + else: + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + bboxes, bbox_num = self.head.post_process(head_outs, im_shape, + scale_factor) + return bboxes, bbox_num + + def get_loss(self, ): + loss = {} + + head_outs, _ = self._forward() + loss_gfl = self.head.get_loss(head_outs, self.inputs) + loss.update(loss_gfl) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + if self.deploy: + return {'picodet': self._forward()[0]} + else: + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/s2anet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/s2anet.py new file mode 100644 index 000000000..ecfc987f9 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/s2anet.py @@ -0,0 +1,102 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['S2ANet'] + + +@register +class S2ANet(BaseArch): + __category__ = 'architecture' + __inject__ = [ + 's2anet_head', + 's2anet_bbox_post_process', + ] + + def __init__(self, backbone, neck, s2anet_head, s2anet_bbox_post_process): + """ + S2ANet, see https://arxiv.org/pdf/2008.09397.pdf + + Args: + backbone (object): backbone instance + neck (object): `FPN` instance + s2anet_head (object): `S2ANetHead` instance + s2anet_bbox_post_process (object): `S2ANetBBoxPostProcess` instance + """ + super(S2ANet, self).__init__() + self.backbone = backbone + self.neck = neck + self.s2anet_head = s2anet_head + self.s2anet_bbox_post_process = s2anet_bbox_post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + kwargs = {'input_shape': backbone.out_shape} + neck = cfg['neck'] and create(cfg['neck'], **kwargs) + + out_shape = neck and neck.out_shape or backbone.out_shape + kwargs = {'input_shape': out_shape} + s2anet_head = create(cfg['s2anet_head'], **kwargs) + s2anet_bbox_post_process = create(cfg['s2anet_bbox_post_process'], + **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "s2anet_head": s2anet_head, + "s2anet_bbox_post_process": s2anet_bbox_post_process, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + if self.neck is not None: + body_feats = self.neck(body_feats) + self.s2anet_head(body_feats) + if self.training: + loss = self.s2anet_head.get_loss(self.inputs) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + else: + im_shape = self.inputs['im_shape'] + scale_factor = self.inputs['scale_factor'] + nms_pre = self.s2anet_bbox_post_process.nms_pre + pred_scores, pred_bboxes = self.s2anet_head.get_prediction(nms_pre) + + # post_process + pred_bboxes, bbox_num = self.s2anet_bbox_post_process(pred_scores, + pred_bboxes) + # rescale the prediction back to origin image + pred_bboxes = self.s2anet_bbox_post_process.get_pred( + pred_bboxes, bbox_num, im_shape, scale_factor) + + # output + output = {'bbox': pred_bboxes, 'bbox_num': bbox_num} + return output + + def get_loss(self, ): + loss = self._forward() + return loss + + def get_pred(self): + output = self._forward() + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/solov2.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/solov2.py new file mode 100644 index 000000000..4e5fc2118 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/solov2.py @@ -0,0 +1,110 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['SOLOv2'] + + +@register +class SOLOv2(BaseArch): + """ + SOLOv2 network, see https://arxiv.org/abs/2003.10152 + + Args: + backbone (object): an backbone instance + solov2_head (object): an `SOLOv2Head` instance + mask_head (object): an `SOLOv2MaskHead` instance + neck (object): neck of network, such as feature pyramid network instance + """ + + __category__ = 'architecture' + + def __init__(self, backbone, solov2_head, mask_head, neck=None): + super(SOLOv2, self).__init__() + self.backbone = backbone + self.neck = neck + self.solov2_head = solov2_head + self.mask_head = mask_head + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + solov2_head = create(cfg['solov2_head'], **kwargs) + mask_head = create(cfg['mask_head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + 'solov2_head': solov2_head, + 'mask_head': mask_head, + } + + def model_arch(self): + body_feats = self.backbone(self.inputs) + + body_feats = self.neck(body_feats) + + self.seg_pred = self.mask_head(body_feats) + + self.cate_pred_list, self.kernel_pred_list = self.solov2_head( + body_feats) + + def get_loss(self, ): + loss = {} + # get gt_ins_labels, gt_cate_labels, etc. + gt_ins_labels, gt_cate_labels, gt_grid_orders = [], [], [] + fg_num = self.inputs['fg_num'] + for i in range(len(self.solov2_head.seg_num_grids)): + ins_label = 'ins_label{}'.format(i) + if ins_label in self.inputs: + gt_ins_labels.append(self.inputs[ins_label]) + cate_label = 'cate_label{}'.format(i) + if cate_label in self.inputs: + gt_cate_labels.append(self.inputs[cate_label]) + grid_order = 'grid_order{}'.format(i) + if grid_order in self.inputs: + gt_grid_orders.append(self.inputs[grid_order]) + + loss_solov2 = self.solov2_head.get_loss( + self.cate_pred_list, self.kernel_pred_list, self.seg_pred, + gt_ins_labels, gt_cate_labels, gt_grid_orders, fg_num) + loss.update(loss_solov2) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + seg_masks, cate_labels, cate_scores, bbox_num = self.solov2_head.get_prediction( + self.cate_pred_list, self.kernel_pred_list, self.seg_pred, + self.inputs['im_shape'], self.inputs['scale_factor']) + outs = { + "segm": seg_masks, + "bbox_num": bbox_num, + 'cate_label': cate_labels, + 'cate_score': cate_scores + } + return outs diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/sparse_rcnn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/sparse_rcnn.py new file mode 100644 index 000000000..34c29498b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/sparse_rcnn.py @@ -0,0 +1,99 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ["SparseRCNN"] + + +@register +class SparseRCNN(BaseArch): + __category__ = 'architecture' + __inject__ = ["postprocess"] + + def __init__(self, + backbone, + neck, + head="SparsercnnHead", + postprocess="SparsePostProcess"): + super(SparseRCNN, self).__init__() + self.backbone = backbone + self.neck = neck + self.head = head + self.postprocess = postprocess + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'roi_input_shape': neck.out_shape} + head = create(cfg['head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "head": head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + fpn_feats = self.neck(body_feats) + head_outs = self.head(fpn_feats, self.inputs["img_whwh"]) + + if not self.training: + bboxes = self.postprocess( + head_outs["pred_logits"], head_outs["pred_boxes"], + self.inputs["scale_factor_wh"], self.inputs["img_whwh"]) + return bboxes + else: + return head_outs + + def get_loss(self): + batch_gt_class = self.inputs["gt_class"] + batch_gt_box = self.inputs["gt_bbox"] + batch_whwh = self.inputs["img_whwh"] + targets = [] + + for i in range(len(batch_gt_class)): + boxes = batch_gt_box[i] + labels = batch_gt_class[i].squeeze(-1) + img_whwh = batch_whwh[i] + img_whwh_tgt = img_whwh.unsqueeze(0).tile([int(boxes.shape[0]), 1]) + targets.append({ + "boxes": boxes, + "labels": labels, + "img_whwh": img_whwh, + "img_whwh_tgt": img_whwh_tgt + }) + + outputs = self._forward() + loss_dict = self.head.get_loss(outputs, targets) + acc = loss_dict["acc"] + loss_dict.pop("acc") + total_loss = sum(loss_dict.values()) + loss_dict.update({"loss": total_loss, "acc": acc}) + return loss_dict + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ssd.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ssd.py new file mode 100644 index 000000000..34bf24108 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ssd.py @@ -0,0 +1,92 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['SSD'] + + +@register +class SSD(BaseArch): + """ + Single Shot MultiBox Detector, see https://arxiv.org/abs/1512.02325 + + Args: + backbone (nn.Layer): backbone instance + ssd_head (nn.Layer): `SSDHead` instance + post_process (object): `BBoxPostProcess` instance + """ + + __category__ = 'architecture' + __inject__ = ['post_process'] + + def __init__(self, backbone, ssd_head, post_process, r34_backbone=False): + super(SSD, self).__init__() + self.backbone = backbone + self.ssd_head = ssd_head + self.post_process = post_process + self.r34_backbone = r34_backbone + if self.r34_backbone: + from ppdet.modeling.backbones.resnet import ResNet + assert isinstance(self.backbone, ResNet) and \ + self.backbone.depth == 34, \ + "If you set r34_backbone=True, please use ResNet-34 as backbone." + self.backbone.res_layers[2].blocks[0].branch2a.conv._stride = [1, 1] + self.backbone.res_layers[2].blocks[0].short.conv._stride = [1, 1] + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + + # head + kwargs = {'input_shape': backbone.out_shape} + ssd_head = create(cfg['ssd_head'], **kwargs) + + return { + 'backbone': backbone, + "ssd_head": ssd_head, + } + + def _forward(self): + # Backbone + body_feats = self.backbone(self.inputs) + + # SSD Head + if self.training: + return self.ssd_head(body_feats, self.inputs['image'], + self.inputs['gt_bbox'], + self.inputs['gt_class']) + else: + preds, anchors = self.ssd_head(body_feats, self.inputs['image']) + bbox, bbox_num = self.post_process(preds, anchors, + self.inputs['im_shape'], + self.inputs['scale_factor']) + return bbox, bbox_num + + def get_loss(self, ): + return {"loss": self._forward()} + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = { + "bbox": bbox_pred, + "bbox_num": bbox_num, + } + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/tood.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/tood.py new file mode 100644 index 000000000..157ec6f3a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/tood.py @@ -0,0 +1,77 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['TOOD'] + + +@register +class TOOD(BaseArch): + """ + TOOD: Task-aligned One-stage Object Detection, see https://arxiv.org/abs/2108.07755 + Args: + backbone (nn.Layer): backbone instance + neck (nn.Layer): 'FPN' instance + head (nn.Layer): 'TOODHead' instance + """ + + __category__ = 'architecture' + + def __init__(self, backbone, neck, head): + super(TOOD, self).__init__() + self.backbone = backbone + self.neck = neck + self.head = head + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + head = create(cfg['head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "head": head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + fpn_feats = self.neck(body_feats) + head_outs = self.head(fpn_feats) + if not self.training: + bboxes, bbox_num = self.head.post_process( + head_outs, self.inputs['im_shape'], self.inputs['scale_factor']) + return bboxes, bbox_num + else: + loss = self.head.get_loss(head_outs, self.inputs) + return loss + + def get_loss(self): + return self._forward() + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = {'bbox': bbox_pred, 'bbox_num': bbox_num} + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ttfnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ttfnet.py new file mode 100644 index 000000000..c3eb61c87 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/ttfnet.py @@ -0,0 +1,98 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch + +__all__ = ['TTFNet'] + + +@register +class TTFNet(BaseArch): + """ + TTFNet network, see https://arxiv.org/abs/1909.00700 + + Args: + backbone (object): backbone instance + neck (object): 'TTFFPN' instance + ttf_head (object): 'TTFHead' instance + post_process (object): 'BBoxPostProcess' instance + """ + + __category__ = 'architecture' + __inject__ = ['post_process'] + + def __init__(self, + backbone='DarkNet', + neck='TTFFPN', + ttf_head='TTFHead', + post_process='BBoxPostProcess'): + super(TTFNet, self).__init__() + self.backbone = backbone + self.neck = neck + self.ttf_head = ttf_head + self.post_process = post_process + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + backbone = create(cfg['backbone']) + + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + kwargs = {'input_shape': neck.out_shape} + ttf_head = create(cfg['ttf_head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "ttf_head": ttf_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + body_feats = self.neck(body_feats) + hm, wh = self.ttf_head(body_feats) + if self.training: + return hm, wh + else: + bbox, bbox_num = self.post_process(hm, wh, self.inputs['im_shape'], + self.inputs['scale_factor']) + return bbox, bbox_num + + def get_loss(self, ): + loss = {} + heatmap = self.inputs['ttf_heatmap'] + box_target = self.inputs['ttf_box_target'] + reg_weight = self.inputs['ttf_reg_weight'] + hm, wh = self._forward() + head_loss = self.ttf_head.get_loss(hm, wh, heatmap, box_target, + reg_weight) + loss.update(head_loss) + total_loss = paddle.add_n(list(loss.values())) + loss.update({'loss': total_loss}) + return loss + + def get_pred(self): + bbox_pred, bbox_num = self._forward() + output = { + "bbox": bbox_pred, + "bbox_num": bbox_num, + } + return output diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/yolo.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/yolo.py new file mode 100644 index 000000000..d5979e695 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/architectures/yolo.py @@ -0,0 +1,124 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from ppdet.core.workspace import register, create +from .meta_arch import BaseArch +from ..post_process import JDEBBoxPostProcess + +__all__ = ['YOLOv3'] + + +@register +class YOLOv3(BaseArch): + __category__ = 'architecture' + __shared__ = ['data_format'] + __inject__ = ['post_process'] + + def __init__(self, + backbone='DarkNet', + neck='YOLOv3FPN', + yolo_head='YOLOv3Head', + post_process='BBoxPostProcess', + data_format='NCHW', + for_mot=False): + """ + YOLOv3 network, see https://arxiv.org/abs/1804.02767 + + Args: + backbone (nn.Layer): backbone instance + neck (nn.Layer): neck instance + yolo_head (nn.Layer): anchor_head instance + bbox_post_process (object): `BBoxPostProcess` instance + data_format (str): data format, NCHW or NHWC + for_mot (bool): whether return other features for multi-object tracking + models, default False in pure object detection models. + """ + super(YOLOv3, self).__init__(data_format=data_format) + self.backbone = backbone + self.neck = neck + self.yolo_head = yolo_head + self.post_process = post_process + self.for_mot = for_mot + self.return_idx = isinstance(post_process, JDEBBoxPostProcess) + + @classmethod + def from_config(cls, cfg, *args, **kwargs): + # backbone + backbone = create(cfg['backbone']) + + # fpn + kwargs = {'input_shape': backbone.out_shape} + neck = create(cfg['neck'], **kwargs) + + # head + kwargs = {'input_shape': neck.out_shape} + yolo_head = create(cfg['yolo_head'], **kwargs) + + return { + 'backbone': backbone, + 'neck': neck, + "yolo_head": yolo_head, + } + + def _forward(self): + body_feats = self.backbone(self.inputs) + neck_feats = self.neck(body_feats, self.for_mot) + + if isinstance(neck_feats, dict): + assert self.for_mot == True + emb_feats = neck_feats['emb_feats'] + neck_feats = neck_feats['yolo_feats'] + + if self.training: + yolo_losses = self.yolo_head(neck_feats, self.inputs) + + if self.for_mot: + return {'det_losses': yolo_losses, 'emb_feats': emb_feats} + else: + return yolo_losses + + else: + yolo_head_outs = self.yolo_head(neck_feats) + + if self.for_mot: + boxes_idx, bbox, bbox_num, nms_keep_idx = self.post_process( + yolo_head_outs, self.yolo_head.mask_anchors) + output = { + 'bbox': bbox, + 'bbox_num': bbox_num, + 'boxes_idx': boxes_idx, + 'nms_keep_idx': nms_keep_idx, + 'emb_feats': emb_feats, + } + else: + if self.return_idx: + _, bbox, bbox_num, _ = self.post_process( + yolo_head_outs, self.yolo_head.mask_anchors) + else: + bbox, bbox_num = self.post_process( + yolo_head_outs, self.yolo_head.mask_anchors, + self.inputs['im_shape'], self.inputs['scale_factor']) + output = {'bbox': bbox, 'bbox_num': bbox_num} + + return output + + def get_loss(self): + return self._forward() + + def get_pred(self): + return self._forward() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__init__.py new file mode 100644 index 000000000..be5bb04d3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__init__.py @@ -0,0 +1,23 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import utils +from . import task_aligned_assigner +from . import atss_assigner +from . import simota_assigner + +from .utils import * +from .task_aligned_assigner import * +from .atss_assigner import * +from .simota_assigner import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..ea40821c5 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/atss_assigner.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/atss_assigner.cpython-37.pyc new file mode 100644 index 000000000..1cbaefccc Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/atss_assigner.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/simota_assigner.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/simota_assigner.cpython-37.pyc new file mode 100644 index 000000000..359a50c2e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/simota_assigner.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/task_aligned_assigner.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/task_aligned_assigner.cpython-37.pyc new file mode 100644 index 000000000..55652436d Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/task_aligned_assigner.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/utils.cpython-37.pyc new file mode 100644 index 000000000..e31eae2d2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/__pycache__/utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/atss_assigner.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/atss_assigner.py new file mode 100644 index 000000000..43e6ae2ab --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/atss_assigner.py @@ -0,0 +1,209 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ppdet.core.workspace import register +from ..ops import iou_similarity +from ..bbox_utils import bbox_center +from .utils import (pad_gt, check_points_inside_bboxes, compute_max_iou_anchor, + compute_max_iou_gt) + + +@register +class ATSSAssigner(nn.Layer): + """Bridging the Gap Between Anchor-based and Anchor-free Detection + via Adaptive Training Sample Selection + """ + __shared__ = ['num_classes'] + + def __init__(self, + topk=9, + num_classes=80, + force_gt_matching=False, + eps=1e-9): + super(ATSSAssigner, self).__init__() + self.topk = topk + self.num_classes = num_classes + self.force_gt_matching = force_gt_matching + self.eps = eps + + def _gather_topk_pyramid(self, gt2anchor_distances, num_anchors_list, + pad_gt_mask): + pad_gt_mask = pad_gt_mask.tile([1, 1, self.topk]).astype(paddle.bool) + gt2anchor_distances_list = paddle.split( + gt2anchor_distances, num_anchors_list, axis=-1) + num_anchors_index = np.cumsum(num_anchors_list).tolist() + num_anchors_index = [0, ] + num_anchors_index[:-1] + is_in_topk_list = [] + topk_idxs_list = [] + for distances, anchors_index in zip(gt2anchor_distances_list, + num_anchors_index): + num_anchors = distances.shape[-1] + topk_metrics, topk_idxs = paddle.topk( + distances, self.topk, axis=-1, largest=False) + topk_idxs_list.append(topk_idxs + anchors_index) + topk_idxs = paddle.where(pad_gt_mask, topk_idxs, + paddle.zeros_like(topk_idxs)) + is_in_topk = F.one_hot(topk_idxs, num_anchors).sum(axis=-2) + is_in_topk = paddle.where(is_in_topk > 1, + paddle.zeros_like(is_in_topk), is_in_topk) + is_in_topk_list.append(is_in_topk.astype(gt2anchor_distances.dtype)) + is_in_topk_list = paddle.concat(is_in_topk_list, axis=-1) + topk_idxs_list = paddle.concat(topk_idxs_list, axis=-1) + return is_in_topk_list, topk_idxs_list + + @paddle.no_grad() + def forward(self, + anchor_bboxes, + num_anchors_list, + gt_labels, + gt_bboxes, + bg_index, + gt_scores=None): + r"""This code is based on + https://github.com/fcjian/TOOD/blob/master/mmdet/core/bbox/assigners/atss_assigner.py + + The assignment is done in following steps + 1. compute iou between all bbox (bbox of all pyramid levels) and gt + 2. compute center distance between all bbox and gt + 3. on each pyramid level, for each gt, select k bbox whose center + are closest to the gt center, so we total select k*l bbox as + candidates for each gt + 4. get corresponding iou for the these candidates, and compute the + mean and std, set mean + std as the iou threshold + 5. select these candidates whose iou are greater than or equal to + the threshold as positive + 6. limit the positive sample's center in gt + 7. if an anchor box is assigned to multiple gts, the one with the + highest iou will be selected. + Args: + anchor_bboxes (Tensor, float32): pre-defined anchors, shape(L, 4), + "xmin, xmax, ymin, ymax" format + num_anchors_list (List): num of anchors in each level + gt_labels (Tensor|List[Tensor], int64): Label of gt_bboxes, shape(B, n, 1) + gt_bboxes (Tensor|List[Tensor], float32): Ground truth bboxes, shape(B, n, 4) + bg_index (int): background index + gt_scores (Tensor|List[Tensor]|None, float32) Score of gt_bboxes, + shape(B, n, 1), if None, then it will initialize with one_hot label + Returns: + assigned_labels (Tensor): (B, L) + assigned_bboxes (Tensor): (B, L, 4) + assigned_scores (Tensor): (B, L, C) + """ + gt_labels, gt_bboxes, pad_gt_scores, pad_gt_mask = pad_gt( + gt_labels, gt_bboxes, gt_scores) + assert gt_labels.ndim == gt_bboxes.ndim and \ + gt_bboxes.ndim == 3 + + num_anchors, _ = anchor_bboxes.shape + batch_size, num_max_boxes, _ = gt_bboxes.shape + + # negative batch + if num_max_boxes == 0: + assigned_labels = paddle.full([batch_size, num_anchors], bg_index) + assigned_bboxes = paddle.zeros([batch_size, num_anchors, 4]) + assigned_scores = paddle.zeros( + [batch_size, num_anchors, self.num_classes]) + return assigned_labels, assigned_bboxes, assigned_scores + + # 1. compute iou between gt and anchor bbox, [B, n, L] + ious = iou_similarity(gt_bboxes.reshape([-1, 4]), anchor_bboxes) + ious = ious.reshape([batch_size, -1, num_anchors]) + + # 2. compute center distance between all anchors and gt, [B, n, L] + gt_centers = bbox_center(gt_bboxes.reshape([-1, 4])).unsqueeze(1) + anchor_centers = bbox_center(anchor_bboxes) + gt2anchor_distances = (gt_centers - anchor_centers.unsqueeze(0)) \ + .norm(2, axis=-1).reshape([batch_size, -1, num_anchors]) + + # 3. on each pyramid level, selecting topk closest candidates + # based on the center distance, [B, n, L] + is_in_topk, topk_idxs = self._gather_topk_pyramid( + gt2anchor_distances, num_anchors_list, pad_gt_mask) + + # 4. get corresponding iou for the these candidates, and compute the + # mean and std, 5. set mean + std as the iou threshold + iou_candidates = ious * is_in_topk + iou_threshold = paddle.index_sample( + iou_candidates.flatten(stop_axis=-2), + topk_idxs.flatten(stop_axis=-2)) + iou_threshold = iou_threshold.reshape([batch_size, num_max_boxes, -1]) + iou_threshold = iou_threshold.mean(axis=-1, keepdim=True) + \ + iou_threshold.std(axis=-1, keepdim=True) + is_in_topk = paddle.where( + iou_candidates > iou_threshold.tile([1, 1, num_anchors]), + is_in_topk, paddle.zeros_like(is_in_topk)) + + # 6. check the positive sample's center in gt, [B, n, L] + is_in_gts = check_points_inside_bboxes(anchor_centers, gt_bboxes) + + # select positive sample, [B, n, L] + mask_positive = is_in_topk * is_in_gts * pad_gt_mask + + # 7. if an anchor box is assigned to multiple gts, + # the one with the highest iou will be selected. + mask_positive_sum = mask_positive.sum(axis=-2) + if mask_positive_sum.max() > 1: + mask_multiple_gts = (mask_positive_sum.unsqueeze(1) > 1).tile( + [1, num_max_boxes, 1]) + is_max_iou = compute_max_iou_anchor(ious) + mask_positive = paddle.where(mask_multiple_gts, is_max_iou, + mask_positive) + mask_positive_sum = mask_positive.sum(axis=-2) + # 8. make sure every gt_bbox matches the anchor + if self.force_gt_matching: + is_max_iou = compute_max_iou_gt(ious) * pad_gt_mask + mask_max_iou = (is_max_iou.sum(-2, keepdim=True) == 1).tile( + [1, num_max_boxes, 1]) + mask_positive = paddle.where(mask_max_iou, is_max_iou, + mask_positive) + mask_positive_sum = mask_positive.sum(axis=-2) + assigned_gt_index = mask_positive.argmax(axis=-2) + assert mask_positive_sum.max() == 1, \ + ("one anchor just assign one gt, but received not equals 1. " + "Received: %f" % mask_positive_sum.max().item()) + + # assigned target + batch_ind = paddle.arange( + end=batch_size, dtype=gt_labels.dtype).unsqueeze(-1) + assigned_gt_index = assigned_gt_index + batch_ind * num_max_boxes + assigned_labels = paddle.gather( + gt_labels.flatten(), assigned_gt_index.flatten(), axis=0) + assigned_labels = assigned_labels.reshape([batch_size, num_anchors]) + assigned_labels = paddle.where( + mask_positive_sum > 0, assigned_labels, + paddle.full_like(assigned_labels, bg_index)) + + assigned_bboxes = paddle.gather( + gt_bboxes.reshape([-1, 4]), assigned_gt_index.flatten(), axis=0) + assigned_bboxes = assigned_bboxes.reshape([batch_size, num_anchors, 4]) + + assigned_scores = F.one_hot(assigned_labels, self.num_classes) + if gt_scores is not None: + gather_scores = paddle.gather( + pad_gt_scores.flatten(), assigned_gt_index.flatten(), axis=0) + gather_scores = gather_scores.reshape([batch_size, num_anchors]) + gather_scores = paddle.where(mask_positive_sum > 0, gather_scores, + paddle.zeros_like(gather_scores)) + assigned_scores *= gather_scores.unsqueeze(-1) + + return assigned_labels, assigned_bboxes, assigned_scores diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/simota_assigner.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/simota_assigner.py new file mode 100644 index 000000000..4b34027e3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/simota_assigner.py @@ -0,0 +1,262 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/core/bbox/assigners/sim_ota_assigner.py + +import paddle +import numpy as np +import paddle.nn.functional as F + +from ppdet.modeling.losses.varifocal_loss import varifocal_loss +from ppdet.modeling.bbox_utils import batch_bbox_overlaps +from ppdet.core.workspace import register + + +@register +class SimOTAAssigner(object): + """Computes matching between predictions and ground truth. + Args: + center_radius (int | float, optional): Ground truth center size + to judge whether a prior is in center. Default 2.5. + candidate_topk (int, optional): The candidate top-k which used to + get top-k ious to calculate dynamic-k. Default 10. + iou_weight (int | float, optional): The scale factor for regression + iou cost. Default 3.0. + cls_weight (int | float, optional): The scale factor for classification + cost. Default 1.0. + num_classes (int): The num_classes of dataset. + use_vfl (int): Whether to use varifocal_loss when calculating the cost matrix. + """ + __shared__ = ['num_classes'] + + def __init__(self, + center_radius=2.5, + candidate_topk=10, + iou_weight=3.0, + cls_weight=1.0, + num_classes=80, + use_vfl=True): + self.center_radius = center_radius + self.candidate_topk = candidate_topk + self.iou_weight = iou_weight + self.cls_weight = cls_weight + self.num_classes = num_classes + self.use_vfl = use_vfl + + def get_in_gt_and_in_center_info(self, flatten_center_and_stride, + gt_bboxes): + num_gt = gt_bboxes.shape[0] + + flatten_x = flatten_center_and_stride[:, 0].unsqueeze(1).tile( + [1, num_gt]) + flatten_y = flatten_center_and_stride[:, 1].unsqueeze(1).tile( + [1, num_gt]) + flatten_stride_x = flatten_center_and_stride[:, 2].unsqueeze(1).tile( + [1, num_gt]) + flatten_stride_y = flatten_center_and_stride[:, 3].unsqueeze(1).tile( + [1, num_gt]) + + # is prior centers in gt bboxes, shape: [n_center, n_gt] + l_ = flatten_x - gt_bboxes[:, 0] + t_ = flatten_y - gt_bboxes[:, 1] + r_ = gt_bboxes[:, 2] - flatten_x + b_ = gt_bboxes[:, 3] - flatten_y + + deltas = paddle.stack([l_, t_, r_, b_], axis=1) + is_in_gts = deltas.min(axis=1) > 0 + is_in_gts_all = is_in_gts.sum(axis=1) > 0 + + # is prior centers in gt centers + gt_center_xs = (gt_bboxes[:, 0] + gt_bboxes[:, 2]) / 2.0 + gt_center_ys = (gt_bboxes[:, 1] + gt_bboxes[:, 3]) / 2.0 + ct_bound_l = gt_center_xs - self.center_radius * flatten_stride_x + ct_bound_t = gt_center_ys - self.center_radius * flatten_stride_y + ct_bound_r = gt_center_xs + self.center_radius * flatten_stride_x + ct_bound_b = gt_center_ys + self.center_radius * flatten_stride_y + + cl_ = flatten_x - ct_bound_l + ct_ = flatten_y - ct_bound_t + cr_ = ct_bound_r - flatten_x + cb_ = ct_bound_b - flatten_y + + ct_deltas = paddle.stack([cl_, ct_, cr_, cb_], axis=1) + is_in_cts = ct_deltas.min(axis=1) > 0 + is_in_cts_all = is_in_cts.sum(axis=1) > 0 + + # in any of gts or gt centers, shape: [n_center] + is_in_gts_or_centers_all = paddle.logical_or(is_in_gts_all, + is_in_cts_all) + + is_in_gts_or_centers_all_inds = paddle.nonzero( + is_in_gts_or_centers_all).squeeze(1) + + # both in gts and gt centers, shape: [num_fg, num_gt] + is_in_gts_and_centers = paddle.logical_and( + paddle.gather( + is_in_gts.cast('int'), is_in_gts_or_centers_all_inds, + axis=0).cast('bool'), + paddle.gather( + is_in_cts.cast('int'), is_in_gts_or_centers_all_inds, + axis=0).cast('bool')) + return is_in_gts_or_centers_all, is_in_gts_or_centers_all_inds, is_in_gts_and_centers + + def dynamic_k_matching(self, cost_matrix, pairwise_ious, num_gt): + match_matrix = np.zeros_like(cost_matrix.numpy()) + # select candidate topk ious for dynamic-k calculation + topk_ious, _ = paddle.topk(pairwise_ious, self.candidate_topk, axis=0) + # calculate dynamic k for each gt + dynamic_ks = paddle.clip(topk_ious.sum(0).cast('int'), min=1) + for gt_idx in range(num_gt): + _, pos_idx = paddle.topk( + cost_matrix[:, gt_idx], k=dynamic_ks[gt_idx], largest=False) + match_matrix[:, gt_idx][pos_idx.numpy()] = 1.0 + + del topk_ious, dynamic_ks, pos_idx + + # match points more than two gts + extra_match_gts_mask = match_matrix.sum(1) > 1 + if extra_match_gts_mask.sum() > 0: + cost_matrix = cost_matrix.numpy() + cost_argmin = np.argmin( + cost_matrix[extra_match_gts_mask, :], axis=1) + match_matrix[extra_match_gts_mask, :] *= 0.0 + match_matrix[extra_match_gts_mask, cost_argmin] = 1.0 + # get foreground mask + match_fg_mask_inmatrix = match_matrix.sum(1) > 0 + match_gt_inds_to_fg = match_matrix[match_fg_mask_inmatrix, :].argmax(1) + + return match_gt_inds_to_fg, match_fg_mask_inmatrix + + def get_sample(self, assign_gt_inds, gt_bboxes): + pos_inds = np.unique(np.nonzero(assign_gt_inds > 0)[0]) + neg_inds = np.unique(np.nonzero(assign_gt_inds == 0)[0]) + pos_assigned_gt_inds = assign_gt_inds[pos_inds] - 1 + + if gt_bboxes.size == 0: + # hack for index error case + assert pos_assigned_gt_inds.size == 0 + pos_gt_bboxes = np.empty_like(gt_bboxes).reshape(-1, 4) + else: + if len(gt_bboxes.shape) < 2: + gt_bboxes = gt_bboxes.resize(-1, 4) + pos_gt_bboxes = gt_bboxes[pos_assigned_gt_inds, :] + return pos_inds, neg_inds, pos_gt_bboxes, pos_assigned_gt_inds + + def __call__(self, + flatten_cls_pred_scores, + flatten_center_and_stride, + flatten_bboxes, + gt_bboxes, + gt_labels, + eps=1e-7): + """Assign gt to priors using SimOTA. + TODO: add comment. + Returns: + assign_result: The assigned result. + """ + num_gt = gt_bboxes.shape[0] + num_bboxes = flatten_bboxes.shape[0] + + if num_gt == 0 or num_bboxes == 0: + # No ground truth or boxes + label = np.ones([num_bboxes], dtype=np.int64) * self.num_classes + label_weight = np.ones([num_bboxes], dtype=np.float32) + bbox_target = np.zeros_like(flatten_center_and_stride) + return 0, label, label_weight, bbox_target + + is_in_gts_or_centers_all, is_in_gts_or_centers_all_inds, is_in_boxes_and_center = self.get_in_gt_and_in_center_info( + flatten_center_and_stride, gt_bboxes) + + # bboxes and scores to calculate matrix + valid_flatten_bboxes = flatten_bboxes[is_in_gts_or_centers_all_inds] + valid_cls_pred_scores = flatten_cls_pred_scores[ + is_in_gts_or_centers_all_inds] + num_valid_bboxes = valid_flatten_bboxes.shape[0] + + pairwise_ious = batch_bbox_overlaps(valid_flatten_bboxes, + gt_bboxes) # [num_points,num_gts] + if self.use_vfl: + gt_vfl_labels = gt_labels.squeeze(-1).unsqueeze(0).tile( + [num_valid_bboxes, 1]).reshape([-1]) + valid_pred_scores = valid_cls_pred_scores.unsqueeze(1).tile( + [1, num_gt, 1]).reshape([-1, self.num_classes]) + vfl_score = np.zeros(valid_pred_scores.shape) + vfl_score[np.arange(0, vfl_score.shape[0]), gt_vfl_labels.numpy( + )] = pairwise_ious.reshape([-1]) + vfl_score = paddle.to_tensor(vfl_score) + losses_vfl = varifocal_loss( + valid_pred_scores, vfl_score, + use_sigmoid=False).reshape([num_valid_bboxes, num_gt]) + losses_giou = batch_bbox_overlaps( + valid_flatten_bboxes, gt_bboxes, mode='giou') + cost_matrix = ( + losses_vfl * self.cls_weight + losses_giou * self.iou_weight + + paddle.logical_not(is_in_boxes_and_center).cast('float32') * + 100000000) + else: + iou_cost = -paddle.log(pairwise_ious + eps) + gt_onehot_label = (F.one_hot( + gt_labels.squeeze(-1).cast(paddle.int64), + flatten_cls_pred_scores.shape[-1]).cast('float32').unsqueeze(0) + .tile([num_valid_bboxes, 1, 1])) + + valid_pred_scores = valid_cls_pred_scores.unsqueeze(1).tile( + [1, num_gt, 1]) + cls_cost = F.binary_cross_entropy( + valid_pred_scores, gt_onehot_label, reduction='none').sum(-1) + + cost_matrix = ( + cls_cost * self.cls_weight + iou_cost * self.iou_weight + + paddle.logical_not(is_in_boxes_and_center).cast('float32') * + 100000000) + + match_gt_inds_to_fg, match_fg_mask_inmatrix = \ + self.dynamic_k_matching( + cost_matrix, pairwise_ious, num_gt) + + # sample and assign results + assigned_gt_inds = np.zeros([num_bboxes], dtype=np.int64) + match_fg_mask_inall = np.zeros_like(assigned_gt_inds) + match_fg_mask_inall[is_in_gts_or_centers_all.numpy( + )] = match_fg_mask_inmatrix + + assigned_gt_inds[match_fg_mask_inall.astype( + np.bool)] = match_gt_inds_to_fg + 1 + + pos_inds, neg_inds, pos_gt_bboxes, pos_assigned_gt_inds \ + = self.get_sample(assigned_gt_inds, gt_bboxes.numpy()) + + bbox_target = np.zeros_like(flatten_bboxes) + bbox_weight = np.zeros_like(flatten_bboxes) + label = np.ones([num_bboxes], dtype=np.int64) * self.num_classes + label_weight = np.zeros([num_bboxes], dtype=np.float32) + + if len(pos_inds) > 0: + gt_labels = gt_labels.numpy() + pos_bbox_targets = pos_gt_bboxes + bbox_target[pos_inds, :] = pos_bbox_targets + bbox_weight[pos_inds, :] = 1.0 + if not np.any(gt_labels): + label[pos_inds] = 0 + else: + label[pos_inds] = gt_labels.squeeze(-1)[pos_assigned_gt_inds] + + label_weight[pos_inds] = 1.0 + if len(neg_inds) > 0: + label_weight[neg_inds] = 1.0 + + pos_num = max(pos_inds.size, 1) + + return pos_num, label, label_weight, bbox_target diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/task_aligned_assigner.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/task_aligned_assigner.py new file mode 100644 index 000000000..7e31c8afc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/task_aligned_assigner.py @@ -0,0 +1,158 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ppdet.core.workspace import register +from ..bbox_utils import iou_similarity +from .utils import (pad_gt, gather_topk_anchors, check_points_inside_bboxes, + compute_max_iou_anchor) + + +@register +class TaskAlignedAssigner(nn.Layer): + """TOOD: Task-aligned One-stage Object Detection + """ + + def __init__(self, topk=13, alpha=1.0, beta=6.0, eps=1e-9): + super(TaskAlignedAssigner, self).__init__() + self.topk = topk + self.alpha = alpha + self.beta = beta + self.eps = eps + + @paddle.no_grad() + def forward(self, + pred_scores, + pred_bboxes, + anchor_points, + gt_labels, + gt_bboxes, + bg_index, + gt_scores=None): + r"""This code is based on + https://github.com/fcjian/TOOD/blob/master/mmdet/core/bbox/assigners/task_aligned_assigner.py + + The assignment is done in following steps + 1. compute alignment metric between all bbox (bbox of all pyramid levels) and gt + 2. select top-k bbox as candidates for each gt + 3. limit the positive sample's center in gt (because the anchor-free detector + only can predict positive distance) + 4. if an anchor box is assigned to multiple gts, the one with the + highest iou will be selected. + Args: + pred_scores (Tensor, float32): predicted class probability, shape(B, L, C) + pred_bboxes (Tensor, float32): predicted bounding boxes, shape(B, L, 4) + anchor_points (Tensor, float32): pre-defined anchors, shape(L, 2), "cxcy" format + gt_labels (Tensor|List[Tensor], int64): Label of gt_bboxes, shape(B, n, 1) + gt_bboxes (Tensor|List[Tensor], float32): Ground truth bboxes, shape(B, n, 4) + bg_index (int): background index + gt_scores (Tensor|List[Tensor]|None, float32) Score of gt_bboxes, + shape(B, n, 1), if None, then it will initialize with one_hot label + Returns: + assigned_labels (Tensor): (B, L) + assigned_bboxes (Tensor): (B, L, 4) + assigned_scores (Tensor): (B, L, C) + """ + assert pred_scores.ndim == pred_bboxes.ndim + + gt_labels, gt_bboxes, pad_gt_scores, pad_gt_mask = pad_gt( + gt_labels, gt_bboxes, gt_scores) + assert gt_labels.ndim == gt_bboxes.ndim and \ + gt_bboxes.ndim == 3 + + batch_size, num_anchors, num_classes = pred_scores.shape + _, num_max_boxes, _ = gt_bboxes.shape + + # negative batch + if num_max_boxes == 0: + assigned_labels = paddle.full([batch_size, num_anchors], bg_index) + assigned_bboxes = paddle.zeros([batch_size, num_anchors, 4]) + assigned_scores = paddle.zeros( + [batch_size, num_anchors, num_classes]) + return assigned_labels, assigned_bboxes, assigned_scores + + # compute iou between gt and pred bbox, [B, n, L] + ious = iou_similarity(gt_bboxes, pred_bboxes) + # gather pred bboxes class score + pred_scores = pred_scores.transpose([0, 2, 1]) + batch_ind = paddle.arange( + end=batch_size, dtype=gt_labels.dtype).unsqueeze(-1) + gt_labels_ind = paddle.stack( + [batch_ind.tile([1, num_max_boxes]), gt_labels.squeeze(-1)], + axis=-1) + bbox_cls_scores = paddle.gather_nd(pred_scores, gt_labels_ind) + # compute alignment metrics, [B, n, L] + alignment_metrics = bbox_cls_scores.pow(self.alpha) * ious.pow( + self.beta) + + # check the positive sample's center in gt, [B, n, L] + is_in_gts = check_points_inside_bboxes(anchor_points, gt_bboxes) + + # select topk largest alignment metrics pred bbox as candidates + # for each gt, [B, n, L] + is_in_topk = gather_topk_anchors( + alignment_metrics * is_in_gts, + self.topk, + topk_mask=pad_gt_mask.tile([1, 1, self.topk]).astype(paddle.bool)) + + # select positive sample, [B, n, L] + mask_positive = is_in_topk * is_in_gts * pad_gt_mask + + # if an anchor box is assigned to multiple gts, + # the one with the highest iou will be selected, [B, n, L] + mask_positive_sum = mask_positive.sum(axis=-2) + if mask_positive_sum.max() > 1: + mask_multiple_gts = (mask_positive_sum.unsqueeze(1) > 1).tile( + [1, num_max_boxes, 1]) + is_max_iou = compute_max_iou_anchor(ious) + mask_positive = paddle.where(mask_multiple_gts, is_max_iou, + mask_positive) + mask_positive_sum = mask_positive.sum(axis=-2) + assigned_gt_index = mask_positive.argmax(axis=-2) + assert mask_positive_sum.max() == 1, \ + ("one anchor just assign one gt, but received not equals 1. " + "Received: %f" % mask_positive_sum.max().item()) + + # assigned target + assigned_gt_index = assigned_gt_index + batch_ind * num_max_boxes + assigned_labels = paddle.gather( + gt_labels.flatten(), assigned_gt_index.flatten(), axis=0) + assigned_labels = assigned_labels.reshape([batch_size, num_anchors]) + assigned_labels = paddle.where( + mask_positive_sum > 0, assigned_labels, + paddle.full_like(assigned_labels, bg_index)) + + assigned_bboxes = paddle.gather( + gt_bboxes.reshape([-1, 4]), assigned_gt_index.flatten(), axis=0) + assigned_bboxes = assigned_bboxes.reshape([batch_size, num_anchors, 4]) + + assigned_scores = F.one_hot(assigned_labels, num_classes) + # rescale alignment metrics + alignment_metrics *= mask_positive + max_metrics_per_instance = alignment_metrics.max(axis=-1, keepdim=True) + max_ious_per_instance = (ious * mask_positive).max(axis=-1, + keepdim=True) + alignment_metrics = alignment_metrics / ( + max_metrics_per_instance + self.eps) * max_ious_per_instance + alignment_metrics = alignment_metrics.max(-2).unsqueeze(-1) + assigned_scores = assigned_scores * alignment_metrics + + return assigned_labels, assigned_bboxes, assigned_scores diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/utils.py new file mode 100644 index 000000000..3448d9d8a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/assigners/utils.py @@ -0,0 +1,149 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn.functional as F + + +def pad_gt(gt_labels, gt_bboxes, gt_scores=None): + r""" Pad 0 in gt_labels and gt_bboxes. + Args: + gt_labels (Tensor|List[Tensor], int64): Label of gt_bboxes, + shape is [B, n, 1] or [[n_1, 1], [n_2, 1], ...], here n = sum(n_i) + gt_bboxes (Tensor|List[Tensor], float32): Ground truth bboxes, + shape is [B, n, 4] or [[n_1, 4], [n_2, 4], ...], here n = sum(n_i) + gt_scores (Tensor|List[Tensor]|None, float32): Score of gt_bboxes, + shape is [B, n, 1] or [[n_1, 4], [n_2, 4], ...], here n = sum(n_i) + Returns: + pad_gt_labels (Tensor, int64): shape[B, n, 1] + pad_gt_bboxes (Tensor, float32): shape[B, n, 4] + pad_gt_scores (Tensor, float32): shape[B, n, 1] + pad_gt_mask (Tensor, float32): shape[B, n, 1], 1 means bbox, 0 means no bbox + """ + if isinstance(gt_labels, paddle.Tensor) and isinstance(gt_bboxes, + paddle.Tensor): + assert gt_labels.ndim == gt_bboxes.ndim and \ + gt_bboxes.ndim == 3 + pad_gt_mask = ( + gt_bboxes.sum(axis=-1, keepdim=True) > 0).astype(gt_bboxes.dtype) + if gt_scores is None: + gt_scores = pad_gt_mask.clone() + assert gt_labels.ndim == gt_scores.ndim + + return gt_labels, gt_bboxes, gt_scores, pad_gt_mask + elif isinstance(gt_labels, list) and isinstance(gt_bboxes, list): + assert len(gt_labels) == len(gt_bboxes), \ + 'The number of `gt_labels` and `gt_bboxes` is not equal. ' + num_max_boxes = max([len(a) for a in gt_bboxes]) + batch_size = len(gt_bboxes) + # pad label and bbox + pad_gt_labels = paddle.zeros( + [batch_size, num_max_boxes, 1], dtype=gt_labels[0].dtype) + pad_gt_bboxes = paddle.zeros( + [batch_size, num_max_boxes, 4], dtype=gt_bboxes[0].dtype) + pad_gt_scores = paddle.zeros( + [batch_size, num_max_boxes, 1], dtype=gt_bboxes[0].dtype) + pad_gt_mask = paddle.zeros( + [batch_size, num_max_boxes, 1], dtype=gt_bboxes[0].dtype) + for i, (label, bbox) in enumerate(zip(gt_labels, gt_bboxes)): + if len(label) > 0 and len(bbox) > 0: + pad_gt_labels[i, :len(label)] = label + pad_gt_bboxes[i, :len(bbox)] = bbox + pad_gt_mask[i, :len(bbox)] = 1. + if gt_scores is not None: + pad_gt_scores[i, :len(gt_scores[i])] = gt_scores[i] + if gt_scores is None: + pad_gt_scores = pad_gt_mask.clone() + return pad_gt_labels, pad_gt_bboxes, pad_gt_scores, pad_gt_mask + else: + raise ValueError('The input `gt_labels` or `gt_bboxes` is invalid! ') + + +def gather_topk_anchors(metrics, topk, largest=True, topk_mask=None, eps=1e-9): + r""" + Args: + metrics (Tensor, float32): shape[B, n, L], n: num_gts, L: num_anchors + topk (int): The number of top elements to look for along the axis. + largest (bool) : largest is a flag, if set to true, + algorithm will sort by descending order, otherwise sort by + ascending order. Default: True + topk_mask (Tensor, bool|None): shape[B, n, topk], ignore bbox mask, + Default: None + eps (float): Default: 1e-9 + Returns: + is_in_topk (Tensor, float32): shape[B, n, L], value=1. means selected + """ + num_anchors = metrics.shape[-1] + topk_metrics, topk_idxs = paddle.topk( + metrics, topk, axis=-1, largest=largest) + if topk_mask is None: + topk_mask = (topk_metrics.max(axis=-1, keepdim=True) > eps).tile( + [1, 1, topk]) + topk_idxs = paddle.where(topk_mask, topk_idxs, paddle.zeros_like(topk_idxs)) + is_in_topk = F.one_hot(topk_idxs, num_anchors).sum(axis=-2) + is_in_topk = paddle.where(is_in_topk > 1, + paddle.zeros_like(is_in_topk), is_in_topk) + return is_in_topk.astype(metrics.dtype) + + +def check_points_inside_bboxes(points, bboxes, eps=1e-9): + r""" + Args: + points (Tensor, float32): shape[L, 2], "xy" format, L: num_anchors + bboxes (Tensor, float32): shape[B, n, 4], "xmin, ymin, xmax, ymax" format + eps (float): Default: 1e-9 + Returns: + is_in_bboxes (Tensor, float32): shape[B, n, L], value=1. means selected + """ + points = points.unsqueeze([0, 1]) + x, y = points.chunk(2, axis=-1) + xmin, ymin, xmax, ymax = bboxes.unsqueeze(2).chunk(4, axis=-1) + l = x - xmin + t = y - ymin + r = xmax - x + b = ymax - y + bbox_ltrb = paddle.concat([l, t, r, b], axis=-1) + return (bbox_ltrb.min(axis=-1) > eps).astype(bboxes.dtype) + + +def compute_max_iou_anchor(ious): + r""" + For each anchor, find the GT with the largest IOU. + Args: + ious (Tensor, float32): shape[B, n, L], n: num_gts, L: num_anchors + Returns: + is_max_iou (Tensor, float32): shape[B, n, L], value=1. means selected + """ + num_max_boxes = ious.shape[-2] + max_iou_index = ious.argmax(axis=-2) + is_max_iou = F.one_hot(max_iou_index, num_max_boxes).transpose([0, 2, 1]) + return is_max_iou.astype(ious.dtype) + + +def compute_max_iou_gt(ious): + r""" + For each GT, find the anchor with the largest IOU. + Args: + ious (Tensor, float32): shape[B, n, L], n: num_gts, L: num_anchors + Returns: + is_max_iou (Tensor, float32): shape[B, n, L], value=1. means selected + """ + num_anchors = ious.shape[-1] + max_iou_index = ious.argmax(axis=-1) + is_max_iou = F.one_hot(max_iou_index, num_anchors) + return is_max_iou.astype(ious.dtype) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__init__.py new file mode 100644 index 000000000..3f415e6a5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__init__.py @@ -0,0 +1,49 @@ +# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import vgg +from . import resnet +from . import darknet +from . import mobilenet_v1 +from . import mobilenet_v3 +from . import hrnet +from . import lite_hrnet +from . import blazenet +from . import ghostnet +from . import senet +from . import res2net +from . import dla +from . import shufflenet_v2 +from . import swin_transformer +from . import lcnet +from . import hardnet +from . import esnet + +from .vgg import * +from .resnet import * +from .darknet import * +from .mobilenet_v1 import * +from .mobilenet_v3 import * +from .hrnet import * +from .lite_hrnet import * +from .blazenet import * +from .ghostnet import * +from .senet import * +from .res2net import * +from .dla import * +from .shufflenet_v2 import * +from .swin_transformer import * +from .lcnet import * +from .hardnet import * +from .esnet import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..065c620ad Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/blazenet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/blazenet.cpython-37.pyc new file mode 100644 index 000000000..afc60e84c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/blazenet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/darknet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/darknet.cpython-37.pyc new file mode 100644 index 000000000..626ca3c65 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/darknet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/dla.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/dla.cpython-37.pyc new file mode 100644 index 000000000..2319927be Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/dla.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/esnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/esnet.cpython-37.pyc new file mode 100644 index 000000000..dd0b7cd16 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/esnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/ghostnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/ghostnet.cpython-37.pyc new file mode 100644 index 000000000..630894564 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/ghostnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hardnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hardnet.cpython-37.pyc new file mode 100644 index 000000000..085823a44 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hardnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hrnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hrnet.cpython-37.pyc new file mode 100644 index 000000000..c610b5d31 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/hrnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lcnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lcnet.cpython-37.pyc new file mode 100644 index 000000000..c4294309c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lcnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lite_hrnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lite_hrnet.cpython-37.pyc new file mode 100644 index 000000000..abbfe056c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/lite_hrnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v1.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v1.cpython-37.pyc new file mode 100644 index 000000000..5c53f4dd3 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v1.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v3.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v3.cpython-37.pyc new file mode 100644 index 000000000..4b589de4f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/mobilenet_v3.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/name_adapter.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/name_adapter.cpython-37.pyc new file mode 100644 index 000000000..8654f5ebe Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/name_adapter.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/res2net.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/res2net.cpython-37.pyc new file mode 100644 index 000000000..a418868ba Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/res2net.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/resnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/resnet.cpython-37.pyc new file mode 100644 index 000000000..ba70f23d8 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/resnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/senet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/senet.cpython-37.pyc new file mode 100644 index 000000000..4fe1193c7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/senet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/shufflenet_v2.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/shufflenet_v2.cpython-37.pyc new file mode 100644 index 000000000..7fa8928fe Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/shufflenet_v2.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/swin_transformer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/swin_transformer.cpython-37.pyc new file mode 100644 index 000000000..c0bbf2d3e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/swin_transformer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/vgg.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/vgg.cpython-37.pyc new file mode 100644 index 000000000..ff8f56ec1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/__pycache__/vgg.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/blazenet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/blazenet.py new file mode 100644 index 000000000..425f2a86e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/blazenet.py @@ -0,0 +1,322 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import KaimingNormal +from ppdet.core.workspace import register, serializable +from ..shape_spec import ShapeSpec + +__all__ = ['BlazeNet'] + + +def hard_swish(x): + return x * F.relu6(x + 3) / 6. + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size, + stride, + padding, + num_groups=1, + act='relu', + conv_lr=0.1, + conv_decay=0., + norm_decay=0., + norm_type='bn', + name=None): + super(ConvBNLayer, self).__init__() + self.act = act + self._conv = nn.Conv2D( + in_channels, + out_channels, + kernel_size=kernel_size, + stride=stride, + padding=padding, + groups=num_groups, + weight_attr=ParamAttr( + learning_rate=conv_lr, initializer=KaimingNormal()), + bias_attr=False) + + if norm_type == 'sync_bn': + self._batch_norm = nn.SyncBatchNorm(out_channels) + else: + self._batch_norm = nn.BatchNorm( + out_channels, act=None, use_global_stats=False) + + def forward(self, x): + x = self._conv(x) + x = self._batch_norm(x) + if self.act == "relu": + x = F.relu(x) + elif self.act == "relu6": + x = F.relu6(x) + elif self.act == 'leaky': + x = F.leaky_relu(x) + elif self.act == 'hard_swish': + x = hard_swish(x) + return x + + +class BlazeBlock(nn.Layer): + def __init__(self, + in_channels, + out_channels1, + out_channels2, + double_channels=None, + stride=1, + use_5x5kernel=True, + act='relu', + name=None): + super(BlazeBlock, self).__init__() + assert stride in [1, 2] + self.use_pool = not stride == 1 + self.use_double_block = double_channels is not None + self.conv_dw = [] + if use_5x5kernel: + self.conv_dw.append( + self.add_sublayer( + name + "1_dw", + ConvBNLayer( + in_channels=in_channels, + out_channels=out_channels1, + kernel_size=5, + stride=stride, + padding=2, + num_groups=out_channels1, + name=name + "1_dw"))) + else: + self.conv_dw.append( + self.add_sublayer( + name + "1_dw_1", + ConvBNLayer( + in_channels=in_channels, + out_channels=out_channels1, + kernel_size=3, + stride=1, + padding=1, + num_groups=out_channels1, + name=name + "1_dw_1"))) + self.conv_dw.append( + self.add_sublayer( + name + "1_dw_2", + ConvBNLayer( + in_channels=out_channels1, + out_channels=out_channels1, + kernel_size=3, + stride=stride, + padding=1, + num_groups=out_channels1, + name=name + "1_dw_2"))) + self.act = act if self.use_double_block else None + self.conv_pw = ConvBNLayer( + in_channels=out_channels1, + out_channels=out_channels2, + kernel_size=1, + stride=1, + padding=0, + act=self.act, + name=name + "1_sep") + if self.use_double_block: + self.conv_dw2 = [] + if use_5x5kernel: + self.conv_dw2.append( + self.add_sublayer( + name + "2_dw", + ConvBNLayer( + in_channels=out_channels2, + out_channels=out_channels2, + kernel_size=5, + stride=1, + padding=2, + num_groups=out_channels2, + name=name + "2_dw"))) + else: + self.conv_dw2.append( + self.add_sublayer( + name + "2_dw_1", + ConvBNLayer( + in_channels=out_channels2, + out_channels=out_channels2, + kernel_size=3, + stride=1, + padding=1, + num_groups=out_channels2, + name=name + "1_dw_1"))) + self.conv_dw2.append( + self.add_sublayer( + name + "2_dw_2", + ConvBNLayer( + in_channels=out_channels2, + out_channels=out_channels2, + kernel_size=3, + stride=1, + padding=1, + num_groups=out_channels2, + name=name + "2_dw_2"))) + self.conv_pw2 = ConvBNLayer( + in_channels=out_channels2, + out_channels=double_channels, + kernel_size=1, + stride=1, + padding=0, + name=name + "2_sep") + # shortcut + if self.use_pool: + shortcut_channel = double_channels or out_channels2 + self._shortcut = [] + self._shortcut.append( + self.add_sublayer( + name + '_shortcut_pool', + nn.MaxPool2D( + kernel_size=stride, stride=stride, ceil_mode=True))) + self._shortcut.append( + self.add_sublayer( + name + '_shortcut_conv', + ConvBNLayer( + in_channels=in_channels, + out_channels=shortcut_channel, + kernel_size=1, + stride=1, + padding=0, + name="shortcut" + name))) + + def forward(self, x): + y = x + for conv_dw_block in self.conv_dw: + y = conv_dw_block(y) + y = self.conv_pw(y) + if self.use_double_block: + for conv_dw2_block in self.conv_dw2: + y = conv_dw2_block(y) + y = self.conv_pw2(y) + if self.use_pool: + for shortcut in self._shortcut: + x = shortcut(x) + return F.relu(paddle.add(x, y)) + + +@register +@serializable +class BlazeNet(nn.Layer): + """ + BlazeFace, see https://arxiv.org/abs/1907.05047 + + Args: + blaze_filters (list): number of filter for each blaze block. + double_blaze_filters (list): number of filter for each double_blaze block. + use_5x5kernel (bool): whether or not filter size is 5x5 in depth-wise conv. + """ + + def __init__( + self, + blaze_filters=[[24, 24], [24, 24], [24, 48, 2], [48, 48], [48, 48]], + double_blaze_filters=[[48, 24, 96, 2], [96, 24, 96], [96, 24, 96], + [96, 24, 96, 2], [96, 24, 96], [96, 24, 96]], + use_5x5kernel=True, + act=None): + super(BlazeNet, self).__init__() + conv1_num_filters = blaze_filters[0][0] + self.conv1 = ConvBNLayer( + in_channels=3, + out_channels=conv1_num_filters, + kernel_size=3, + stride=2, + padding=1, + name="conv1") + in_channels = conv1_num_filters + self.blaze_block = [] + self._out_channels = [] + for k, v in enumerate(blaze_filters): + assert len(v) in [2, 3], \ + "blaze_filters {} not in [2, 3]" + if len(v) == 2: + self.blaze_block.append( + self.add_sublayer( + 'blaze_{}'.format(k), + BlazeBlock( + in_channels, + v[0], + v[1], + use_5x5kernel=use_5x5kernel, + act=act, + name='blaze_{}'.format(k)))) + elif len(v) == 3: + self.blaze_block.append( + self.add_sublayer( + 'blaze_{}'.format(k), + BlazeBlock( + in_channels, + v[0], + v[1], + stride=v[2], + use_5x5kernel=use_5x5kernel, + act=act, + name='blaze_{}'.format(k)))) + in_channels = v[1] + + for k, v in enumerate(double_blaze_filters): + assert len(v) in [3, 4], \ + "blaze_filters {} not in [3, 4]" + if len(v) == 3: + self.blaze_block.append( + self.add_sublayer( + 'double_blaze_{}'.format(k), + BlazeBlock( + in_channels, + v[0], + v[1], + double_channels=v[2], + use_5x5kernel=use_5x5kernel, + act=act, + name='double_blaze_{}'.format(k)))) + elif len(v) == 4: + self.blaze_block.append( + self.add_sublayer( + 'double_blaze_{}'.format(k), + BlazeBlock( + in_channels, + v[0], + v[1], + double_channels=v[2], + stride=v[3], + use_5x5kernel=use_5x5kernel, + act=act, + name='double_blaze_{}'.format(k)))) + in_channels = v[2] + self._out_channels.append(in_channels) + + def forward(self, inputs): + outs = [] + y = self.conv1(inputs['image']) + for block in self.blaze_block: + y = block(y) + outs.append(y) + return [outs[-4], outs[-1]] + + @property + def out_shape(self): + return [ + ShapeSpec(channels=c) + for c in [self._out_channels[-4], self._out_channels[-1]] + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/darknet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/darknet.py new file mode 100644 index 000000000..246529699 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/darknet.py @@ -0,0 +1,340 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ppdet.core.workspace import register, serializable +from ppdet.modeling.ops import batch_norm, mish +from ..shape_spec import ShapeSpec + +__all__ = ['DarkNet', 'ConvBNLayer'] + + +class ConvBNLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size=3, + stride=1, + groups=1, + padding=0, + norm_type='bn', + norm_decay=0., + act="leaky", + freeze_norm=False, + data_format='NCHW', + name=''): + """ + conv + bn + activation layer + + Args: + ch_in (int): input channel + ch_out (int): output channel + filter_size (int): filter size, default 3 + stride (int): stride, default 1 + groups (int): number of groups of conv layer, default 1 + padding (int): padding size, default 0 + norm_type (str): batch norm type, default bn + norm_decay (str): decay for weight and bias of batch norm layer, default 0. + act (str): activation function type, default 'leaky', which means leaky_relu + freeze_norm (bool): whether to freeze norm, default False + data_format (str): data format, NCHW or NHWC + """ + super(ConvBNLayer, self).__init__() + + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=padding, + groups=groups, + data_format=data_format, + bias_attr=False) + self.batch_norm = batch_norm( + ch_out, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + self.act = act + + def forward(self, inputs): + out = self.conv(inputs) + out = self.batch_norm(out) + if self.act == 'leaky': + out = F.leaky_relu(out, 0.1) + elif self.act == 'mish': + out = mish(out) + return out + + +class DownSample(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size=3, + stride=2, + padding=1, + norm_type='bn', + norm_decay=0., + freeze_norm=False, + data_format='NCHW'): + """ + downsample layer + + Args: + ch_in (int): input channel + ch_out (int): output channel + filter_size (int): filter size, default 3 + stride (int): stride, default 2 + padding (int): padding size, default 1 + norm_type (str): batch norm type, default bn + norm_decay (str): decay for weight and bias of batch norm layer, default 0. + freeze_norm (bool): whether to freeze norm, default False + data_format (str): data format, NCHW or NHWC + """ + + super(DownSample, self).__init__() + + self.conv_bn_layer = ConvBNLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=filter_size, + stride=stride, + padding=padding, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + self.ch_out = ch_out + + def forward(self, inputs): + out = self.conv_bn_layer(inputs) + return out + + +class BasicBlock(nn.Layer): + def __init__(self, + ch_in, + ch_out, + norm_type='bn', + norm_decay=0., + freeze_norm=False, + data_format='NCHW'): + """ + BasicBlock layer of DarkNet + + Args: + ch_in (int): input channel + ch_out (int): output channel + norm_type (str): batch norm type, default bn + norm_decay (str): decay for weight and bias of batch norm layer, default 0. + freeze_norm (bool): whether to freeze norm, default False + data_format (str): data format, NCHW or NHWC + """ + + super(BasicBlock, self).__init__() + + self.conv1 = ConvBNLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=1, + padding=0, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + self.conv2 = ConvBNLayer( + ch_in=ch_out, + ch_out=ch_out * 2, + filter_size=3, + stride=1, + padding=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + + def forward(self, inputs): + conv1 = self.conv1(inputs) + conv2 = self.conv2(conv1) + out = paddle.add(x=inputs, y=conv2) + return out + + +class Blocks(nn.Layer): + def __init__(self, + ch_in, + ch_out, + count, + norm_type='bn', + norm_decay=0., + freeze_norm=False, + name=None, + data_format='NCHW'): + """ + Blocks layer, which consist of some BaickBlock layers + + Args: + ch_in (int): input channel + ch_out (int): output channel + count (int): number of BasicBlock layer + norm_type (str): batch norm type, default bn + norm_decay (str): decay for weight and bias of batch norm layer, default 0. + freeze_norm (bool): whether to freeze norm, default False + name (str): layer name + data_format (str): data format, NCHW or NHWC + """ + super(Blocks, self).__init__() + + self.basicblock0 = BasicBlock( + ch_in, + ch_out, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + self.res_out_list = [] + for i in range(1, count): + block_name = '{}.{}'.format(name, i) + res_out = self.add_sublayer( + block_name, + BasicBlock( + ch_out * 2, + ch_out, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format)) + self.res_out_list.append(res_out) + self.ch_out = ch_out + + def forward(self, inputs): + y = self.basicblock0(inputs) + for basic_block_i in self.res_out_list: + y = basic_block_i(y) + return y + + +DarkNet_cfg = {53: ([1, 2, 8, 8, 4])} + + +@register +@serializable +class DarkNet(nn.Layer): + __shared__ = ['norm_type', 'data_format'] + + def __init__(self, + depth=53, + freeze_at=-1, + return_idx=[2, 3, 4], + num_stages=5, + norm_type='bn', + norm_decay=0., + freeze_norm=False, + data_format='NCHW'): + """ + Darknet, see https://pjreddie.com/darknet/yolo/ + + Args: + depth (int): depth of network + freeze_at (int): freeze the backbone at which stage + filter_size (int): filter size, default 3 + return_idx (list): index of stages whose feature maps are returned + norm_type (str): batch norm type, default bn + norm_decay (str): decay for weight and bias of batch norm layer, default 0. + data_format (str): data format, NCHW or NHWC + """ + super(DarkNet, self).__init__() + self.depth = depth + self.freeze_at = freeze_at + self.return_idx = return_idx + self.num_stages = num_stages + self.stages = DarkNet_cfg[self.depth][0:num_stages] + + self.conv0 = ConvBNLayer( + ch_in=3, + ch_out=32, + filter_size=3, + stride=1, + padding=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + + self.downsample0 = DownSample( + ch_in=32, + ch_out=32 * 2, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format) + + self._out_channels = [] + self.darknet_conv_block_list = [] + self.downsample_list = [] + ch_in = [64, 128, 256, 512, 1024] + for i, stage in enumerate(self.stages): + name = 'stage.{}'.format(i) + conv_block = self.add_sublayer( + name, + Blocks( + int(ch_in[i]), + 32 * (2**i), + stage, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format, + name=name)) + self.darknet_conv_block_list.append(conv_block) + if i in return_idx: + self._out_channels.append(64 * (2**i)) + for i in range(num_stages - 1): + down_name = 'stage.{}.downsample'.format(i) + downsample = self.add_sublayer( + down_name, + DownSample( + ch_in=32 * (2**(i + 1)), + ch_out=32 * (2**(i + 2)), + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + data_format=data_format)) + self.downsample_list.append(downsample) + + def forward(self, inputs): + x = inputs['image'] + + out = self.conv0(x) + out = self.downsample0(out) + blocks = [] + for i, conv_block_i in enumerate(self.darknet_conv_block_list): + out = conv_block_i(out) + if i == self.freeze_at: + out.stop_gradient = True + if i in self.return_idx: + blocks.append(out) + if i < self.num_stages - 1: + out = self.downsample_list[i](out) + return blocks + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/dla.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/dla.py new file mode 100644 index 000000000..4ab06ab7f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/dla.py @@ -0,0 +1,243 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ppdet.modeling.layers import ConvNormLayer +from ..shape_spec import ShapeSpec + +DLA_cfg = {34: ([1, 1, 1, 2, 2, 1], [16, 32, 64, 128, 256, 512])} + + +class BasicBlock(nn.Layer): + def __init__(self, ch_in, ch_out, stride=1): + super(BasicBlock, self).__init__() + self.conv1 = ConvNormLayer( + ch_in, + ch_out, + filter_size=3, + stride=stride, + bias_on=False, + norm_decay=None) + self.conv2 = ConvNormLayer( + ch_out, + ch_out, + filter_size=3, + stride=1, + bias_on=False, + norm_decay=None) + + def forward(self, inputs, residual=None): + if residual is None: + residual = inputs + + out = self.conv1(inputs) + out = F.relu(out) + + out = self.conv2(out) + + out = paddle.add(x=out, y=residual) + out = F.relu(out) + + return out + + +class Root(nn.Layer): + def __init__(self, ch_in, ch_out, kernel_size, residual): + super(Root, self).__init__() + self.conv = ConvNormLayer( + ch_in, + ch_out, + filter_size=1, + stride=1, + bias_on=False, + norm_decay=None) + self.residual = residual + + def forward(self, inputs): + children = inputs + out = self.conv(paddle.concat(inputs, axis=1)) + if self.residual: + out = paddle.add(x=out, y=children[0]) + out = F.relu(out) + + return out + + +class Tree(nn.Layer): + def __init__(self, + level, + block, + ch_in, + ch_out, + stride=1, + level_root=False, + root_dim=0, + root_kernel_size=1, + root_residual=False): + super(Tree, self).__init__() + if root_dim == 0: + root_dim = 2 * ch_out + if level_root: + root_dim += ch_in + if level == 1: + self.tree1 = block(ch_in, ch_out, stride) + self.tree2 = block(ch_out, ch_out, 1) + else: + self.tree1 = Tree( + level - 1, + block, + ch_in, + ch_out, + stride, + root_dim=0, + root_kernel_size=root_kernel_size, + root_residual=root_residual) + self.tree2 = Tree( + level - 1, + block, + ch_out, + ch_out, + 1, + root_dim=root_dim + ch_out, + root_kernel_size=root_kernel_size, + root_residual=root_residual) + + if level == 1: + self.root = Root(root_dim, ch_out, root_kernel_size, root_residual) + self.level_root = level_root + self.root_dim = root_dim + self.downsample = None + self.project = None + self.level = level + if stride > 1: + self.downsample = nn.MaxPool2D(stride, stride=stride) + if ch_in != ch_out: + self.project = ConvNormLayer( + ch_in, + ch_out, + filter_size=1, + stride=1, + bias_on=False, + norm_decay=None) + + def forward(self, x, residual=None, children=None): + children = [] if children is None else children + bottom = self.downsample(x) if self.downsample else x + residual = self.project(bottom) if self.project else bottom + if self.level_root: + children.append(bottom) + x1 = self.tree1(x, residual) + if self.level == 1: + x2 = self.tree2(x1) + x = self.root([x2, x1] + children) + else: + children.append(x1) + x = self.tree2(x1, children=children) + return x + + +@register +@serializable +class DLA(nn.Layer): + """ + DLA, see https://arxiv.org/pdf/1707.06484.pdf + + Args: + depth (int): DLA depth, should be 34. + residual_root (bool): whether use a reidual layer in the root block + + """ + + def __init__(self, depth=34, residual_root=False): + super(DLA, self).__init__() + levels, channels = DLA_cfg[depth] + if depth == 34: + block = BasicBlock + self.channels = channels + self.base_layer = nn.Sequential( + ConvNormLayer( + 3, + channels[0], + filter_size=7, + stride=1, + bias_on=False, + norm_decay=None), + nn.ReLU()) + self.level0 = self._make_conv_level(channels[0], channels[0], levels[0]) + self.level1 = self._make_conv_level( + channels[0], channels[1], levels[1], stride=2) + self.level2 = Tree( + levels[2], + block, + channels[1], + channels[2], + 2, + level_root=False, + root_residual=residual_root) + self.level3 = Tree( + levels[3], + block, + channels[2], + channels[3], + 2, + level_root=True, + root_residual=residual_root) + self.level4 = Tree( + levels[4], + block, + channels[3], + channels[4], + 2, + level_root=True, + root_residual=residual_root) + self.level5 = Tree( + levels[5], + block, + channels[4], + channels[5], + 2, + level_root=True, + root_residual=residual_root) + + def _make_conv_level(self, ch_in, ch_out, conv_num, stride=1): + modules = [] + for i in range(conv_num): + modules.extend([ + ConvNormLayer( + ch_in, + ch_out, + filter_size=3, + stride=stride if i == 0 else 1, + bias_on=False, + norm_decay=None), nn.ReLU() + ]) + ch_in = ch_out + return nn.Sequential(*modules) + + @property + def out_shape(self): + return [ShapeSpec(channels=self.channels[i]) for i in range(6)] + + def forward(self, inputs): + outs = [] + im = inputs['image'] + feats = self.base_layer(im) + for i in range(6): + feats = getattr(self, 'level{}'.format(i))(feats) + outs.append(feats) + + return outs diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/esnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/esnet.py new file mode 100644 index 000000000..2b3f3c54a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/esnet.py @@ -0,0 +1,290 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn import Conv2D, MaxPool2D, AdaptiveAvgPool2D, BatchNorm +from paddle.nn.initializer import KaimingNormal +from paddle.regularizer import L2Decay + +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec +from ppdet.modeling.ops import channel_shuffle +from ppdet.modeling.backbones.shufflenet_v2 import ConvBNLayer + +__all__ = ['ESNet'] + + +def make_divisible(v, divisor=16, min_value=None): + if min_value is None: + min_value = divisor + new_v = max(min_value, int(v + divisor / 2) // divisor * divisor) + if new_v < 0.9 * v: + new_v += divisor + return new_v + + +class SEModule(nn.Layer): + def __init__(self, channel, reduction=4): + super(SEModule, self).__init__() + self.avg_pool = AdaptiveAvgPool2D(1) + self.conv1 = Conv2D( + in_channels=channel, + out_channels=channel // reduction, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr(), + bias_attr=ParamAttr()) + self.conv2 = Conv2D( + in_channels=channel // reduction, + out_channels=channel, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr(), + bias_attr=ParamAttr()) + + def forward(self, inputs): + outputs = self.avg_pool(inputs) + outputs = self.conv1(outputs) + outputs = F.relu(outputs) + outputs = self.conv2(outputs) + outputs = F.hardsigmoid(outputs) + return paddle.multiply(x=inputs, y=outputs) + + +class InvertedResidual(nn.Layer): + def __init__(self, + in_channels, + mid_channels, + out_channels, + stride, + act="relu"): + super(InvertedResidual, self).__init__() + self._conv_pw = ConvBNLayer( + in_channels=in_channels // 2, + out_channels=mid_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + self._conv_dw = ConvBNLayer( + in_channels=mid_channels // 2, + out_channels=mid_channels // 2, + kernel_size=3, + stride=stride, + padding=1, + groups=mid_channels // 2, + act=None) + self._se = SEModule(mid_channels) + + self._conv_linear = ConvBNLayer( + in_channels=mid_channels, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + + def forward(self, inputs): + x1, x2 = paddle.split( + inputs, + num_or_sections=[inputs.shape[1] // 2, inputs.shape[1] // 2], + axis=1) + x2 = self._conv_pw(x2) + x3 = self._conv_dw(x2) + x3 = paddle.concat([x2, x3], axis=1) + x3 = self._se(x3) + x3 = self._conv_linear(x3) + out = paddle.concat([x1, x3], axis=1) + return channel_shuffle(out, 2) + + +class InvertedResidualDS(nn.Layer): + def __init__(self, + in_channels, + mid_channels, + out_channels, + stride, + act="relu"): + super(InvertedResidualDS, self).__init__() + + # branch1 + self._conv_dw_1 = ConvBNLayer( + in_channels=in_channels, + out_channels=in_channels, + kernel_size=3, + stride=stride, + padding=1, + groups=in_channels, + act=None) + self._conv_linear_1 = ConvBNLayer( + in_channels=in_channels, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + # branch2 + self._conv_pw_2 = ConvBNLayer( + in_channels=in_channels, + out_channels=mid_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + self._conv_dw_2 = ConvBNLayer( + in_channels=mid_channels // 2, + out_channels=mid_channels // 2, + kernel_size=3, + stride=stride, + padding=1, + groups=mid_channels // 2, + act=None) + self._se = SEModule(mid_channels // 2) + self._conv_linear_2 = ConvBNLayer( + in_channels=mid_channels // 2, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + self._conv_dw_mv1 = ConvBNLayer( + in_channels=out_channels, + out_channels=out_channels, + kernel_size=3, + stride=1, + padding=1, + groups=out_channels, + act="hard_swish") + self._conv_pw_mv1 = ConvBNLayer( + in_channels=out_channels, + out_channels=out_channels, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act="hard_swish") + + def forward(self, inputs): + x1 = self._conv_dw_1(inputs) + x1 = self._conv_linear_1(x1) + x2 = self._conv_pw_2(inputs) + x2 = self._conv_dw_2(x2) + x2 = self._se(x2) + x2 = self._conv_linear_2(x2) + out = paddle.concat([x1, x2], axis=1) + out = self._conv_dw_mv1(out) + out = self._conv_pw_mv1(out) + + return out + + +@register +@serializable +class ESNet(nn.Layer): + def __init__(self, + scale=1.0, + act="hard_swish", + feature_maps=[4, 11, 14], + channel_ratio=[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]): + super(ESNet, self).__init__() + self.scale = scale + if isinstance(feature_maps, Integral): + feature_maps = [feature_maps] + self.feature_maps = feature_maps + stage_repeats = [3, 7, 3] + + stage_out_channels = [ + -1, 24, make_divisible(128 * scale), make_divisible(256 * scale), + make_divisible(512 * scale), 1024 + ] + + self._out_channels = [] + self._feature_idx = 0 + # 1. conv1 + self._conv1 = ConvBNLayer( + in_channels=3, + out_channels=stage_out_channels[1], + kernel_size=3, + stride=2, + padding=1, + act=act) + self._max_pool = MaxPool2D(kernel_size=3, stride=2, padding=1) + self._feature_idx += 1 + + # 2. bottleneck sequences + self._block_list = [] + arch_idx = 0 + for stage_id, num_repeat in enumerate(stage_repeats): + for i in range(num_repeat): + channels_scales = channel_ratio[arch_idx] + mid_c = make_divisible( + int(stage_out_channels[stage_id + 2] * channels_scales), + divisor=8) + if i == 0: + block = self.add_sublayer( + name=str(stage_id + 2) + '_' + str(i + 1), + sublayer=InvertedResidualDS( + in_channels=stage_out_channels[stage_id + 1], + mid_channels=mid_c, + out_channels=stage_out_channels[stage_id + 2], + stride=2, + act=act)) + else: + block = self.add_sublayer( + name=str(stage_id + 2) + '_' + str(i + 1), + sublayer=InvertedResidual( + in_channels=stage_out_channels[stage_id + 2], + mid_channels=mid_c, + out_channels=stage_out_channels[stage_id + 2], + stride=1, + act=act)) + self._block_list.append(block) + arch_idx += 1 + self._feature_idx += 1 + self._update_out_channels(stage_out_channels[stage_id + 2], + self._feature_idx, self.feature_maps) + + def _update_out_channels(self, channel, feature_idx, feature_maps): + if feature_idx in feature_maps: + self._out_channels.append(channel) + + def forward(self, inputs): + y = self._conv1(inputs['image']) + y = self._max_pool(y) + outs = [] + for i, inv in enumerate(self._block_list): + y = inv(y) + if i + 2 in self.feature_maps: + outs.append(y) + + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/ghostnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/ghostnet.py new file mode 100644 index 000000000..cd333b4fe --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/ghostnet.py @@ -0,0 +1,470 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import math +import paddle +from paddle import ParamAttr +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn import AdaptiveAvgPool2D, Linear +from paddle.nn.initializer import Uniform + +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec +from .mobilenet_v3 import make_divisible, ConvBNLayer + +__all__ = ['GhostNet'] + + +class ExtraBlockDW(nn.Layer): + def __init__(self, + in_c, + ch_1, + ch_2, + stride, + lr_mult, + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + name=None): + super(ExtraBlockDW, self).__init__() + self.pointwise_conv = ConvBNLayer( + in_c=in_c, + out_c=ch_1, + filter_size=1, + stride=1, + padding=0, + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra1") + self.depthwise_conv = ConvBNLayer( + in_c=ch_1, + out_c=ch_2, + filter_size=3, + stride=stride, + padding=1, # + num_groups=int(ch_1), + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra2_dw") + self.normal_conv = ConvBNLayer( + in_c=ch_2, + out_c=ch_2, + filter_size=1, + stride=1, + padding=0, + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra2_sep") + + def forward(self, inputs): + x = self.pointwise_conv(inputs) + x = self.depthwise_conv(x) + x = self.normal_conv(x) + return x + + +class SEBlock(nn.Layer): + def __init__(self, num_channels, lr_mult, reduction_ratio=4, name=None): + super(SEBlock, self).__init__() + self.pool2d_gap = AdaptiveAvgPool2D(1) + self._num_channels = num_channels + stdv = 1.0 / math.sqrt(num_channels * 1.0) + med_ch = num_channels // reduction_ratio + self.squeeze = Linear( + num_channels, + med_ch, + weight_attr=ParamAttr( + learning_rate=lr_mult, initializer=Uniform(-stdv, stdv)), + bias_attr=ParamAttr(learning_rate=lr_mult)) + stdv = 1.0 / math.sqrt(med_ch * 1.0) + self.excitation = Linear( + med_ch, + num_channels, + weight_attr=ParamAttr( + learning_rate=lr_mult, initializer=Uniform(-stdv, stdv)), + bias_attr=ParamAttr(learning_rate=lr_mult)) + + def forward(self, inputs): + pool = self.pool2d_gap(inputs) + pool = paddle.squeeze(pool, axis=[2, 3]) + squeeze = self.squeeze(pool) + squeeze = F.relu(squeeze) + excitation = self.excitation(squeeze) + excitation = paddle.clip(x=excitation, min=0, max=1) + excitation = paddle.unsqueeze(excitation, axis=[2, 3]) + out = paddle.multiply(inputs, excitation) + return out + + +class GhostModule(nn.Layer): + def __init__(self, + in_channels, + output_channels, + kernel_size=1, + ratio=2, + dw_size=3, + stride=1, + relu=True, + lr_mult=1., + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + name=None): + super(GhostModule, self).__init__() + init_channels = int(math.ceil(output_channels / ratio)) + new_channels = int(init_channels * (ratio - 1)) + self.primary_conv = ConvBNLayer( + in_c=in_channels, + out_c=init_channels, + filter_size=kernel_size, + stride=stride, + padding=int((kernel_size - 1) // 2), + num_groups=1, + act="relu" if relu else None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_primary_conv") + self.cheap_operation = ConvBNLayer( + in_c=init_channels, + out_c=new_channels, + filter_size=dw_size, + stride=1, + padding=int((dw_size - 1) // 2), + num_groups=init_channels, + act="relu" if relu else None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_cheap_operation") + + def forward(self, inputs): + x = self.primary_conv(inputs) + y = self.cheap_operation(x) + out = paddle.concat([x, y], axis=1) + return out + + +class GhostBottleneck(nn.Layer): + def __init__(self, + in_channels, + hidden_dim, + output_channels, + kernel_size, + stride, + use_se, + lr_mult, + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + return_list=False, + name=None): + super(GhostBottleneck, self).__init__() + self._stride = stride + self._use_se = use_se + self._num_channels = in_channels + self._output_channels = output_channels + self.return_list = return_list + + self.ghost_module_1 = GhostModule( + in_channels=in_channels, + output_channels=hidden_dim, + kernel_size=1, + stride=1, + relu=True, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_ghost_module_1") + if stride == 2: + self.depthwise_conv = ConvBNLayer( + in_c=hidden_dim, + out_c=hidden_dim, + filter_size=kernel_size, + stride=stride, + padding=int((kernel_size - 1) // 2), + num_groups=hidden_dim, + act=None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + + "_depthwise_depthwise" # looks strange due to an old typo, will be fixed later. + ) + if use_se: + self.se_block = SEBlock(hidden_dim, lr_mult, name=name + "_se") + self.ghost_module_2 = GhostModule( + in_channels=hidden_dim, + output_channels=output_channels, + kernel_size=1, + relu=False, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_ghost_module_2") + if stride != 1 or in_channels != output_channels: + self.shortcut_depthwise = ConvBNLayer( + in_c=in_channels, + out_c=in_channels, + filter_size=kernel_size, + stride=stride, + padding=int((kernel_size - 1) // 2), + num_groups=in_channels, + act=None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + + "_shortcut_depthwise_depthwise" # looks strange due to an old typo, will be fixed later. + ) + self.shortcut_conv = ConvBNLayer( + in_c=in_channels, + out_c=output_channels, + filter_size=1, + stride=1, + padding=0, + num_groups=1, + act=None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_shortcut_conv") + + def forward(self, inputs): + y = self.ghost_module_1(inputs) + x = y + if self._stride == 2: + x = self.depthwise_conv(x) + if self._use_se: + x = self.se_block(x) + x = self.ghost_module_2(x) + + if self._stride == 1 and self._num_channels == self._output_channels: + shortcut = inputs + else: + shortcut = self.shortcut_depthwise(inputs) + shortcut = self.shortcut_conv(shortcut) + x = paddle.add(x=x, y=shortcut) + + if self.return_list: + return [y, x] + else: + return x + + +@register +@serializable +class GhostNet(nn.Layer): + __shared__ = ['norm_type'] + + def __init__( + self, + scale=1.3, + feature_maps=[6, 12, 15], + with_extra_blocks=False, + extra_block_filters=[[256, 512], [128, 256], [128, 256], [64, 128]], + lr_mult_list=[1.0, 1.0, 1.0, 1.0, 1.0], + conv_decay=0., + norm_type='bn', + norm_decay=0.0, + freeze_norm=False): + super(GhostNet, self).__init__() + if isinstance(feature_maps, Integral): + feature_maps = [feature_maps] + if norm_type == 'sync_bn' and freeze_norm: + raise ValueError( + "The norm_type should not be sync_bn when freeze_norm is True") + self.feature_maps = feature_maps + self.with_extra_blocks = with_extra_blocks + self.extra_block_filters = extra_block_filters + + inplanes = 16 + self.cfgs = [ + # k, t, c, SE, s + [3, 16, 16, 0, 1], + [3, 48, 24, 0, 2], + [3, 72, 24, 0, 1], + [5, 72, 40, 1, 2], + [5, 120, 40, 1, 1], + [3, 240, 80, 0, 2], + [3, 200, 80, 0, 1], + [3, 184, 80, 0, 1], + [3, 184, 80, 0, 1], + [3, 480, 112, 1, 1], + [3, 672, 112, 1, 1], + [5, 672, 160, 1, 2], # SSDLite output + [5, 960, 160, 0, 1], + [5, 960, 160, 1, 1], + [5, 960, 160, 0, 1], + [5, 960, 160, 1, 1] + ] + self.scale = scale + conv1_out_ch = int(make_divisible(inplanes * self.scale, 4)) + self.conv1 = ConvBNLayer( + in_c=3, + out_c=conv1_out_ch, + filter_size=3, + stride=2, + padding=1, + num_groups=1, + act="relu", + lr_mult=1., + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="conv1") + + # build inverted residual blocks + self._out_channels = [] + self.ghost_bottleneck_list = [] + idx = 0 + inplanes = conv1_out_ch + for k, exp_size, c, use_se, s in self.cfgs: + lr_idx = min(idx // 3, len(lr_mult_list) - 1) + lr_mult = lr_mult_list[lr_idx] + + # for SSD/SSDLite, first head input is after ResidualUnit expand_conv + return_list = self.with_extra_blocks and idx + 2 in self.feature_maps + + ghost_bottleneck = self.add_sublayer( + "_ghostbottleneck_" + str(idx), + sublayer=GhostBottleneck( + in_channels=inplanes, + hidden_dim=int(make_divisible(exp_size * self.scale, 4)), + output_channels=int(make_divisible(c * self.scale, 4)), + kernel_size=k, + stride=s, + use_se=use_se, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + return_list=return_list, + name="_ghostbottleneck_" + str(idx))) + self.ghost_bottleneck_list.append(ghost_bottleneck) + inplanes = int(make_divisible(c * self.scale, 4)) + idx += 1 + self._update_out_channels( + int(make_divisible(exp_size * self.scale, 4)) + if return_list else inplanes, idx + 1, feature_maps) + + if self.with_extra_blocks: + self.extra_block_list = [] + extra_out_c = int(make_divisible(self.scale * self.cfgs[-1][1], 4)) + lr_idx = min(idx // 3, len(lr_mult_list) - 1) + lr_mult = lr_mult_list[lr_idx] + + conv_extra = self.add_sublayer( + "conv" + str(idx + 2), + sublayer=ConvBNLayer( + in_c=inplanes, + out_c=extra_out_c, + filter_size=1, + stride=1, + padding=0, + num_groups=1, + act="relu6", + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="conv" + str(idx + 2))) + self.extra_block_list.append(conv_extra) + idx += 1 + self._update_out_channels(extra_out_c, idx + 1, feature_maps) + + for j, block_filter in enumerate(self.extra_block_filters): + in_c = extra_out_c if j == 0 else self.extra_block_filters[j - + 1][1] + conv_extra = self.add_sublayer( + "conv" + str(idx + 2), + sublayer=ExtraBlockDW( + in_c, + block_filter[0], + block_filter[1], + stride=2, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name='conv' + str(idx + 2))) + self.extra_block_list.append(conv_extra) + idx += 1 + self._update_out_channels(block_filter[1], idx + 1, + feature_maps) + + def _update_out_channels(self, channel, feature_idx, feature_maps): + if feature_idx in feature_maps: + self._out_channels.append(channel) + + def forward(self, inputs): + x = self.conv1(inputs['image']) + outs = [] + for idx, ghost_bottleneck in enumerate(self.ghost_bottleneck_list): + x = ghost_bottleneck(x) + if idx + 2 in self.feature_maps: + if isinstance(x, list): + outs.append(x[0]) + x = x[1] + else: + outs.append(x) + + if not self.with_extra_blocks: + return outs + + for i, block in enumerate(self.extra_block_list): + idx = i + len(self.ghost_bottleneck_list) + x = block(x) + if idx + 2 in self.feature_maps: + outs.append(x) + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hardnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hardnet.py new file mode 100644 index 000000000..14a1599df --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hardnet.py @@ -0,0 +1,224 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +from ppdet.core.workspace import register +from ..shape_spec import ShapeSpec + +__all__ = ['HarDNet'] + + +def ConvLayer(in_channels, + out_channels, + kernel_size=3, + stride=1, + bias_attr=False): + layer = nn.Sequential( + ('conv', nn.Conv2D( + in_channels, + out_channels, + kernel_size=kernel_size, + stride=stride, + padding=kernel_size // 2, + groups=1, + bias_attr=bias_attr)), ('norm', nn.BatchNorm2D(out_channels)), + ('relu', nn.ReLU6())) + return layer + + +def DWConvLayer(in_channels, + out_channels, + kernel_size=3, + stride=1, + bias_attr=False): + layer = nn.Sequential( + ('dwconv', nn.Conv2D( + in_channels, + out_channels, + kernel_size=kernel_size, + stride=stride, + padding=1, + groups=out_channels, + bias_attr=bias_attr)), ('norm', nn.BatchNorm2D(out_channels))) + return layer + + +def CombConvLayer(in_channels, out_channels, kernel_size=1, stride=1): + layer = nn.Sequential( + ('layer1', ConvLayer( + in_channels, out_channels, kernel_size=kernel_size)), + ('layer2', DWConvLayer( + out_channels, out_channels, stride=stride))) + return layer + + +class HarDBlock(nn.Layer): + def __init__(self, + in_channels, + growth_rate, + grmul, + n_layers, + keepBase=False, + residual_out=False, + dwconv=False): + super().__init__() + self.keepBase = keepBase + self.links = [] + layers_ = [] + self.out_channels = 0 + for i in range(n_layers): + outch, inch, link = self.get_link(i + 1, in_channels, growth_rate, + grmul) + self.links.append(link) + if dwconv: + layers_.append(CombConvLayer(inch, outch)) + else: + layers_.append(ConvLayer(inch, outch)) + + if (i % 2 == 0) or (i == n_layers - 1): + self.out_channels += outch + self.layers = nn.LayerList(layers_) + + def get_out_ch(self): + return self.out_channels + + def get_link(self, layer, base_ch, growth_rate, grmul): + if layer == 0: + return base_ch, 0, [] + out_channels = growth_rate + + link = [] + for i in range(10): + dv = 2**i + if layer % dv == 0: + k = layer - dv + link.append(k) + if i > 0: + out_channels *= grmul + + out_channels = int(int(out_channels + 1) / 2) * 2 + in_channels = 0 + + for i in link: + ch, _, _ = self.get_link(i, base_ch, growth_rate, grmul) + in_channels += ch + + return out_channels, in_channels, link + + def forward(self, x): + layers_ = [x] + + for layer in range(len(self.layers)): + link = self.links[layer] + tin = [] + for i in link: + tin.append(layers_[i]) + if len(tin) > 1: + x = paddle.concat(tin, 1) + else: + x = tin[0] + out = self.layers[layer](x) + layers_.append(out) + + t = len(layers_) + out_ = [] + for i in range(t): + if (i == 0 and self.keepBase) or (i == t - 1) or (i % 2 == 1): + out_.append(layers_[i]) + out = paddle.concat(out_, 1) + + return out + + +@register +class HarDNet(nn.Layer): + def __init__(self, depth_wise=False, return_idx=[1, 3, 8, 13], arch=85): + super(HarDNet, self).__init__() + assert arch in [39, 68, 85], "HarDNet-{} not support.".format(arch) + if arch == 85: + first_ch = [48, 96] + second_kernel = 3 + ch_list = [192, 256, 320, 480, 720] + grmul = 1.7 + gr = [24, 24, 28, 36, 48] + n_layers = [8, 16, 16, 16, 16] + elif arch == 68: + first_ch = [32, 64] + second_kernel = 3 + ch_list = [128, 256, 320, 640] + grmul = 1.7 + gr = [14, 16, 20, 40] + n_layers = [8, 16, 16, 16] + + self.return_idx = return_idx + self._out_channels = [96, 214, 458, 784] + + avg_pool = True + if depth_wise: + second_kernel = 1 + avg_pool = False + + blks = len(n_layers) + self.base = nn.LayerList([]) + + # First Layer: Standard Conv3x3, Stride=2 + self.base.append( + ConvLayer( + in_channels=3, + out_channels=first_ch[0], + kernel_size=3, + stride=2, + bias_attr=False)) + + # Second Layer + self.base.append( + ConvLayer( + first_ch[0], first_ch[1], kernel_size=second_kernel)) + + # Avgpooling or DWConv3x3 downsampling + if avg_pool: + self.base.append(nn.AvgPool2D(kernel_size=3, stride=2, padding=1)) + else: + self.base.append(DWConvLayer(first_ch[1], first_ch[1], stride=2)) + + # Build all HarDNet blocks + ch = first_ch[1] + for i in range(blks): + blk = HarDBlock(ch, gr[i], grmul, n_layers[i], dwconv=depth_wise) + ch = blk.out_channels + self.base.append(blk) + + if i != blks - 1: + self.base.append(ConvLayer(ch, ch_list[i], kernel_size=1)) + ch = ch_list[i] + if i == 0: + self.base.append( + nn.AvgPool2D( + kernel_size=2, stride=2, ceil_mode=True)) + elif i != blks - 1 and i != 1 and i != 3: + self.base.append(nn.AvgPool2D(kernel_size=2, stride=2)) + + def forward(self, inputs): + x = inputs['image'] + outs = [] + for i, layer in enumerate(self.base): + x = layer(x) + if i in self.return_idx: + outs.append(x) + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=self._out_channels[i]) for i in range(4)] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hrnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hrnet.py new file mode 100644 index 000000000..d92aa95f5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/hrnet.py @@ -0,0 +1,727 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn import AdaptiveAvgPool2D, Linear +from paddle.regularizer import L2Decay +from paddle import ParamAttr +from paddle.nn.initializer import Normal, Uniform +from numbers import Integral +import math + +from ppdet.core.workspace import register +from ..shape_spec import ShapeSpec + +__all__ = ['HRNet'] + + +class ConvNormLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + stride=1, + norm_type='bn', + norm_groups=32, + use_dcn=False, + norm_decay=0., + freeze_norm=False, + act=None, + name=None): + super(ConvNormLayer, self).__init__() + assert norm_type in ['bn', 'sync_bn', 'gn'] + + self.act = act + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=False) + + norm_lr = 0. if freeze_norm else 1. + + param_attr = ParamAttr( + learning_rate=norm_lr, regularizer=L2Decay(norm_decay)) + bias_attr = ParamAttr( + learning_rate=norm_lr, regularizer=L2Decay(norm_decay)) + global_stats = True if freeze_norm else False + if norm_type in ['bn', 'sync_bn']: + self.norm = nn.BatchNorm( + ch_out, + param_attr=param_attr, + bias_attr=bias_attr, + use_global_stats=global_stats) + elif norm_type == 'gn': + self.norm = nn.GroupNorm( + num_groups=norm_groups, + num_channels=ch_out, + weight_attr=param_attr, + bias_attr=bias_attr) + norm_params = self.norm.parameters() + if freeze_norm: + for param in norm_params: + param.stop_gradient = True + + def forward(self, inputs): + out = self.conv(inputs) + out = self.norm(out) + + if self.act == 'relu': + out = F.relu(out) + return out + + +class Layer1(nn.Layer): + def __init__(self, + num_channels, + has_se=False, + norm_decay=0., + freeze_norm=True, + name=None): + super(Layer1, self).__init__() + + self.bottleneck_block_list = [] + + for i in range(4): + bottleneck_block = self.add_sublayer( + "block_{}_{}".format(name, i + 1), + BottleneckBlock( + num_channels=num_channels if i == 0 else 256, + num_filters=64, + has_se=has_se, + stride=1, + downsample=True if i == 0 else False, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + '_' + str(i + 1))) + self.bottleneck_block_list.append(bottleneck_block) + + def forward(self, input): + conv = input + for block_func in self.bottleneck_block_list: + conv = block_func(conv) + return conv + + +class TransitionLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + norm_decay=0., + freeze_norm=True, + name=None): + super(TransitionLayer, self).__init__() + + num_in = len(in_channels) + num_out = len(out_channels) + out = [] + self.conv_bn_func_list = [] + for i in range(num_out): + residual = None + if i < num_in: + if in_channels[i] != out_channels[i]: + residual = self.add_sublayer( + "transition_{}_layer_{}".format(name, i + 1), + ConvNormLayer( + ch_in=in_channels[i], + ch_out=out_channels[i], + filter_size=3, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act='relu', + name=name + '_layer_' + str(i + 1))) + else: + residual = self.add_sublayer( + "transition_{}_layer_{}".format(name, i + 1), + ConvNormLayer( + ch_in=in_channels[-1], + ch_out=out_channels[i], + filter_size=3, + stride=2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act='relu', + name=name + '_layer_' + str(i + 1))) + self.conv_bn_func_list.append(residual) + + def forward(self, input): + outs = [] + for idx, conv_bn_func in enumerate(self.conv_bn_func_list): + if conv_bn_func is None: + outs.append(input[idx]) + else: + if idx < len(input): + outs.append(conv_bn_func(input[idx])) + else: + outs.append(conv_bn_func(input[-1])) + return outs + + +class Branches(nn.Layer): + def __init__(self, + block_num, + in_channels, + out_channels, + has_se=False, + norm_decay=0., + freeze_norm=True, + name=None): + super(Branches, self).__init__() + + self.basic_block_list = [] + for i in range(len(out_channels)): + self.basic_block_list.append([]) + for j in range(block_num): + in_ch = in_channels[i] if j == 0 else out_channels[i] + basic_block_func = self.add_sublayer( + "bb_{}_branch_layer_{}_{}".format(name, i + 1, j + 1), + BasicBlock( + num_channels=in_ch, + num_filters=out_channels[i], + has_se=has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + '_branch_layer_' + str(i + 1) + '_' + + str(j + 1))) + self.basic_block_list[i].append(basic_block_func) + + def forward(self, inputs): + outs = [] + for idx, input in enumerate(inputs): + conv = input + basic_block_list = self.basic_block_list[idx] + for basic_block_func in basic_block_list: + conv = basic_block_func(conv) + outs.append(conv) + return outs + + +class BottleneckBlock(nn.Layer): + def __init__(self, + num_channels, + num_filters, + has_se, + stride=1, + downsample=False, + norm_decay=0., + freeze_norm=True, + name=None): + super(BottleneckBlock, self).__init__() + + self.has_se = has_se + self.downsample = downsample + + self.conv1 = ConvNormLayer( + ch_in=num_channels, + ch_out=num_filters, + filter_size=1, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act="relu", + name=name + "_conv1") + self.conv2 = ConvNormLayer( + ch_in=num_filters, + ch_out=num_filters, + filter_size=3, + stride=stride, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act="relu", + name=name + "_conv2") + self.conv3 = ConvNormLayer( + ch_in=num_filters, + ch_out=num_filters * 4, + filter_size=1, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act=None, + name=name + "_conv3") + + if self.downsample: + self.conv_down = ConvNormLayer( + ch_in=num_channels, + ch_out=num_filters * 4, + filter_size=1, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act=None, + name=name + "_downsample") + + if self.has_se: + self.se = SELayer( + num_channels=num_filters * 4, + num_filters=num_filters * 4, + reduction_ratio=16, + name='fc' + name) + + def forward(self, input): + residual = input + conv1 = self.conv1(input) + conv2 = self.conv2(conv1) + conv3 = self.conv3(conv2) + + if self.downsample: + residual = self.conv_down(input) + + if self.has_se: + conv3 = self.se(conv3) + + y = paddle.add(x=residual, y=conv3) + y = F.relu(y) + return y + + +class BasicBlock(nn.Layer): + def __init__(self, + num_channels, + num_filters, + stride=1, + has_se=False, + downsample=False, + norm_decay=0., + freeze_norm=True, + name=None): + super(BasicBlock, self).__init__() + + self.has_se = has_se + self.downsample = downsample + self.conv1 = ConvNormLayer( + ch_in=num_channels, + ch_out=num_filters, + filter_size=3, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + stride=stride, + act="relu", + name=name + "_conv1") + self.conv2 = ConvNormLayer( + ch_in=num_filters, + ch_out=num_filters, + filter_size=3, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + stride=1, + act=None, + name=name + "_conv2") + + if self.downsample: + self.conv_down = ConvNormLayer( + ch_in=num_channels, + ch_out=num_filters * 4, + filter_size=1, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act=None, + name=name + "_downsample") + + if self.has_se: + self.se = SELayer( + num_channels=num_filters, + num_filters=num_filters, + reduction_ratio=16, + name='fc' + name) + + def forward(self, input): + residual = input + conv1 = self.conv1(input) + conv2 = self.conv2(conv1) + + if self.downsample: + residual = self.conv_down(input) + + if self.has_se: + conv2 = self.se(conv2) + + y = paddle.add(x=residual, y=conv2) + y = F.relu(y) + return y + + +class SELayer(nn.Layer): + def __init__(self, num_channels, num_filters, reduction_ratio, name=None): + super(SELayer, self).__init__() + + self.pool2d_gap = AdaptiveAvgPool2D(1) + + self._num_channels = num_channels + + med_ch = int(num_channels / reduction_ratio) + stdv = 1.0 / math.sqrt(num_channels * 1.0) + self.squeeze = Linear( + num_channels, + med_ch, + weight_attr=ParamAttr(initializer=Uniform(-stdv, stdv))) + + stdv = 1.0 / math.sqrt(med_ch * 1.0) + self.excitation = Linear( + med_ch, + num_filters, + weight_attr=ParamAttr(initializer=Uniform(-stdv, stdv))) + + def forward(self, input): + pool = self.pool2d_gap(input) + pool = paddle.squeeze(pool, axis=[2, 3]) + squeeze = self.squeeze(pool) + squeeze = F.relu(squeeze) + excitation = self.excitation(squeeze) + excitation = F.sigmoid(excitation) + excitation = paddle.unsqueeze(excitation, axis=[2, 3]) + out = input * excitation + return out + + +class Stage(nn.Layer): + def __init__(self, + num_channels, + num_modules, + num_filters, + has_se=False, + norm_decay=0., + freeze_norm=True, + multi_scale_output=True, + name=None): + super(Stage, self).__init__() + + self._num_modules = num_modules + self.stage_func_list = [] + for i in range(num_modules): + if i == num_modules - 1 and not multi_scale_output: + stage_func = self.add_sublayer( + "stage_{}_{}".format(name, i + 1), + HighResolutionModule( + num_channels=num_channels, + num_filters=num_filters, + has_se=has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + multi_scale_output=False, + name=name + '_' + str(i + 1))) + else: + stage_func = self.add_sublayer( + "stage_{}_{}".format(name, i + 1), + HighResolutionModule( + num_channels=num_channels, + num_filters=num_filters, + has_se=has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + '_' + str(i + 1))) + + self.stage_func_list.append(stage_func) + + def forward(self, input): + out = input + for idx in range(self._num_modules): + out = self.stage_func_list[idx](out) + return out + + +class HighResolutionModule(nn.Layer): + def __init__(self, + num_channels, + num_filters, + has_se=False, + multi_scale_output=True, + norm_decay=0., + freeze_norm=True, + name=None): + super(HighResolutionModule, self).__init__() + self.branches_func = Branches( + block_num=4, + in_channels=num_channels, + out_channels=num_filters, + has_se=has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name) + + self.fuse_func = FuseLayers( + in_channels=num_filters, + out_channels=num_filters, + multi_scale_output=multi_scale_output, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name) + + def forward(self, input): + out = self.branches_func(input) + out = self.fuse_func(out) + return out + + +class FuseLayers(nn.Layer): + def __init__(self, + in_channels, + out_channels, + multi_scale_output=True, + norm_decay=0., + freeze_norm=True, + name=None): + super(FuseLayers, self).__init__() + + self._actual_ch = len(in_channels) if multi_scale_output else 1 + self._in_channels = in_channels + + self.residual_func_list = [] + for i in range(self._actual_ch): + for j in range(len(in_channels)): + residual_func = None + if j > i: + residual_func = self.add_sublayer( + "residual_{}_layer_{}_{}".format(name, i + 1, j + 1), + ConvNormLayer( + ch_in=in_channels[j], + ch_out=out_channels[i], + filter_size=1, + stride=1, + act=None, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + '_layer_' + str(i + 1) + '_' + + str(j + 1))) + self.residual_func_list.append(residual_func) + elif j < i: + pre_num_filters = in_channels[j] + for k in range(i - j): + if k == i - j - 1: + residual_func = self.add_sublayer( + "residual_{}_layer_{}_{}_{}".format( + name, i + 1, j + 1, k + 1), + ConvNormLayer( + ch_in=pre_num_filters, + ch_out=out_channels[i], + filter_size=3, + stride=2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act=None, + name=name + '_layer_' + str(i + 1) + '_' + + str(j + 1) + '_' + str(k + 1))) + pre_num_filters = out_channels[i] + else: + residual_func = self.add_sublayer( + "residual_{}_layer_{}_{}_{}".format( + name, i + 1, j + 1, k + 1), + ConvNormLayer( + ch_in=pre_num_filters, + ch_out=out_channels[j], + filter_size=3, + stride=2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act="relu", + name=name + '_layer_' + str(i + 1) + '_' + + str(j + 1) + '_' + str(k + 1))) + pre_num_filters = out_channels[j] + self.residual_func_list.append(residual_func) + + def forward(self, input): + outs = [] + residual_func_idx = 0 + for i in range(self._actual_ch): + residual = input[i] + for j in range(len(self._in_channels)): + if j > i: + y = self.residual_func_list[residual_func_idx](input[j]) + residual_func_idx += 1 + y = F.interpolate(y, scale_factor=2**(j - i)) + residual = paddle.add(x=residual, y=y) + elif j < i: + y = input[j] + for k in range(i - j): + y = self.residual_func_list[residual_func_idx](y) + residual_func_idx += 1 + + residual = paddle.add(x=residual, y=y) + residual = F.relu(residual) + outs.append(residual) + + return outs + + +@register +class HRNet(nn.Layer): + """ + HRNet, see https://arxiv.org/abs/1908.07919 + + Args: + width (int): the width of HRNet + has_se (bool): whether to add SE block for each stage + freeze_at (int): the stage to freeze + freeze_norm (bool): whether to freeze norm in HRNet + norm_decay (float): weight decay for normalization layer weights + return_idx (List): the stage to return + upsample (bool): whether to upsample and concat the backbone feats + """ + + def __init__(self, + width=18, + has_se=False, + freeze_at=0, + freeze_norm=True, + norm_decay=0., + return_idx=[0, 1, 2, 3], + upsample=False): + super(HRNet, self).__init__() + + self.width = width + self.has_se = has_se + if isinstance(return_idx, Integral): + return_idx = [return_idx] + + assert len(return_idx) > 0, "need one or more return index" + self.freeze_at = freeze_at + self.return_idx = return_idx + self.upsample = upsample + + self.channels = { + 18: [[18, 36], [18, 36, 72], [18, 36, 72, 144]], + 30: [[30, 60], [30, 60, 120], [30, 60, 120, 240]], + 32: [[32, 64], [32, 64, 128], [32, 64, 128, 256]], + 40: [[40, 80], [40, 80, 160], [40, 80, 160, 320]], + 44: [[44, 88], [44, 88, 176], [44, 88, 176, 352]], + 48: [[48, 96], [48, 96, 192], [48, 96, 192, 384]], + 60: [[60, 120], [60, 120, 240], [60, 120, 240, 480]], + 64: [[64, 128], [64, 128, 256], [64, 128, 256, 512]] + } + + channels_2, channels_3, channels_4 = self.channels[width] + num_modules_2, num_modules_3, num_modules_4 = 1, 4, 3 + self._out_channels = [sum(channels_4)] if self.upsample else channels_4 + self._out_strides = [4] if self.upsample else [4, 8, 16, 32] + + self.conv_layer1_1 = ConvNormLayer( + ch_in=3, + ch_out=64, + filter_size=3, + stride=2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act='relu', + name="layer1_1") + + self.conv_layer1_2 = ConvNormLayer( + ch_in=64, + ch_out=64, + filter_size=3, + stride=2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + act='relu', + name="layer1_2") + + self.la1 = Layer1( + num_channels=64, + has_se=has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="layer2") + + self.tr1 = TransitionLayer( + in_channels=[256], + out_channels=channels_2, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="tr1") + + self.st2 = Stage( + num_channels=channels_2, + num_modules=num_modules_2, + num_filters=channels_2, + has_se=self.has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="st2") + + self.tr2 = TransitionLayer( + in_channels=channels_2, + out_channels=channels_3, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="tr2") + + self.st3 = Stage( + num_channels=channels_3, + num_modules=num_modules_3, + num_filters=channels_3, + has_se=self.has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="st3") + + self.tr3 = TransitionLayer( + in_channels=channels_3, + out_channels=channels_4, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="tr3") + self.st4 = Stage( + num_channels=channels_4, + num_modules=num_modules_4, + num_filters=channels_4, + has_se=self.has_se, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + multi_scale_output=len(return_idx) > 1, + name="st4") + + def forward(self, inputs): + x = inputs['image'] + conv1 = self.conv_layer1_1(x) + conv2 = self.conv_layer1_2(conv1) + + la1 = self.la1(conv2) + tr1 = self.tr1([la1]) + st2 = self.st2(tr1) + tr2 = self.tr2(st2) + + st3 = self.st3(tr2) + tr3 = self.tr3(st3) + + st4 = self.st4(tr3) + + if self.upsample: + # Upsampling + x0_h, x0_w = st4[0].shape[2:4] + x1 = F.upsample(st4[1], size=(x0_h, x0_w), mode='bilinear') + x2 = F.upsample(st4[2], size=(x0_h, x0_w), mode='bilinear') + x3 = F.upsample(st4[3], size=(x0_h, x0_w), mode='bilinear') + x = paddle.concat([st4[0], x1, x2, x3], 1) + return x + + res = [] + for i, layer in enumerate(st4): + if i == self.freeze_at: + layer.stop_gradient = True + if i in self.return_idx: + res.append(layer) + + return res + + @property + def out_shape(self): + if self.upsample: + self.return_idx = [0] + return [ + ShapeSpec( + channels=self._out_channels[i], stride=self._out_strides[i]) + for i in self.return_idx + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lcnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lcnet.py new file mode 100644 index 000000000..fd8ad4e46 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lcnet.py @@ -0,0 +1,258 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +from paddle import ParamAttr +from paddle.nn import AdaptiveAvgPool2D, BatchNorm, Conv2D, Dropout, Linear +from paddle.regularizer import L2Decay +from paddle.nn.initializer import KaimingNormal + +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec + +__all__ = ['LCNet'] + +NET_CONFIG = { + "blocks2": + #k, in_c, out_c, s, use_se + [[3, 16, 32, 1, False], ], + "blocks3": [ + [3, 32, 64, 2, False], + [3, 64, 64, 1, False], + ], + "blocks4": [ + [3, 64, 128, 2, False], + [3, 128, 128, 1, False], + ], + "blocks5": [ + [3, 128, 256, 2, False], + [5, 256, 256, 1, False], + [5, 256, 256, 1, False], + [5, 256, 256, 1, False], + [5, 256, 256, 1, False], + [5, 256, 256, 1, False], + ], + "blocks6": [[5, 256, 512, 2, True], [5, 512, 512, 1, True]] +} + + +def make_divisible(v, divisor=8, min_value=None): + if min_value is None: + min_value = divisor + new_v = max(min_value, int(v + divisor / 2) // divisor * divisor) + if new_v < 0.9 * v: + new_v += divisor + return new_v + + +class ConvBNLayer(nn.Layer): + def __init__(self, + num_channels, + filter_size, + num_filters, + stride, + num_groups=1): + super().__init__() + + self.conv = Conv2D( + in_channels=num_channels, + out_channels=num_filters, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=num_groups, + weight_attr=ParamAttr(initializer=KaimingNormal()), + bias_attr=False) + + self.bn = BatchNorm( + num_filters, + param_attr=ParamAttr(regularizer=L2Decay(0.0)), + bias_attr=ParamAttr(regularizer=L2Decay(0.0))) + self.hardswish = nn.Hardswish() + + def forward(self, x): + x = self.conv(x) + x = self.bn(x) + x = self.hardswish(x) + return x + + +class DepthwiseSeparable(nn.Layer): + def __init__(self, + num_channels, + num_filters, + stride, + dw_size=3, + use_se=False): + super().__init__() + self.use_se = use_se + self.dw_conv = ConvBNLayer( + num_channels=num_channels, + num_filters=num_channels, + filter_size=dw_size, + stride=stride, + num_groups=num_channels) + if use_se: + self.se = SEModule(num_channels) + self.pw_conv = ConvBNLayer( + num_channels=num_channels, + filter_size=1, + num_filters=num_filters, + stride=1) + + def forward(self, x): + x = self.dw_conv(x) + if self.use_se: + x = self.se(x) + x = self.pw_conv(x) + return x + + +class SEModule(nn.Layer): + def __init__(self, channel, reduction=4): + super().__init__() + self.avg_pool = AdaptiveAvgPool2D(1) + self.conv1 = Conv2D( + in_channels=channel, + out_channels=channel // reduction, + kernel_size=1, + stride=1, + padding=0) + self.relu = nn.ReLU() + self.conv2 = Conv2D( + in_channels=channel // reduction, + out_channels=channel, + kernel_size=1, + stride=1, + padding=0) + self.hardsigmoid = nn.Hardsigmoid() + + def forward(self, x): + identity = x + x = self.avg_pool(x) + x = self.conv1(x) + x = self.relu(x) + x = self.conv2(x) + x = self.hardsigmoid(x) + x = paddle.multiply(x=identity, y=x) + return x + + +@register +@serializable +class LCNet(nn.Layer): + def __init__(self, scale=1.0, feature_maps=[3, 4, 5]): + super().__init__() + self.scale = scale + self.feature_maps = feature_maps + + out_channels = [] + + self.conv1 = ConvBNLayer( + num_channels=3, + filter_size=3, + num_filters=make_divisible(16 * scale), + stride=2) + + self.blocks2 = nn.Sequential(* [ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks2"]) + ]) + + self.blocks3 = nn.Sequential(* [ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks3"]) + ]) + + out_channels.append( + make_divisible(NET_CONFIG["blocks3"][-1][2] * scale)) + + self.blocks4 = nn.Sequential(* [ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks4"]) + ]) + + out_channels.append( + make_divisible(NET_CONFIG["blocks4"][-1][2] * scale)) + + self.blocks5 = nn.Sequential(* [ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks5"]) + ]) + + out_channels.append( + make_divisible(NET_CONFIG["blocks5"][-1][2] * scale)) + + self.blocks6 = nn.Sequential(* [ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks6"]) + ]) + + out_channels.append( + make_divisible(NET_CONFIG["blocks6"][-1][2] * scale)) + self._out_channels = [ + ch for idx, ch in enumerate(out_channels) if idx + 2 in feature_maps + ] + + def forward(self, inputs): + x = inputs['image'] + outs = [] + + x = self.conv1(x) + x = self.blocks2(x) + x = self.blocks3(x) + outs.append(x) + x = self.blocks4(x) + outs.append(x) + x = self.blocks5(x) + outs.append(x) + x = self.blocks6(x) + outs.append(x) + outs = [o for i, o in enumerate(outs) if i + 2 in self.feature_maps] + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lite_hrnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lite_hrnet.py new file mode 100644 index 000000000..f14aae8e2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/lite_hrnet.py @@ -0,0 +1,881 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from numbers import Integral +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from paddle.nn.initializer import Normal, Constant +from ppdet.core.workspace import register +from ppdet.modeling.shape_spec import ShapeSpec +from ppdet.modeling.ops import channel_shuffle +from .. import layers as L + +__all__ = ['LiteHRNet'] + + +class ConvNormLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + stride=1, + groups=1, + norm_type=None, + norm_groups=32, + norm_decay=0., + freeze_norm=False, + act=None): + super(ConvNormLayer, self).__init__() + self.act = act + norm_lr = 0. if freeze_norm else 1. + if norm_type is not None: + assert norm_type in ['bn', 'sync_bn', 'gn'],\ + "norm_type should be one of ['bn', 'sync_bn', 'gn'], but got {}".format(norm_type) + param_attr = ParamAttr( + initializer=Constant(1.0), + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), ) + bias_attr = ParamAttr( + learning_rate=norm_lr, regularizer=L2Decay(norm_decay)) + global_stats = True if freeze_norm else False + if norm_type in ['bn', 'sync_bn']: + self.norm = nn.BatchNorm( + ch_out, + param_attr=param_attr, + bias_attr=bias_attr, + use_global_stats=global_stats, ) + elif norm_type == 'gn': + self.norm = nn.GroupNorm( + num_groups=norm_groups, + num_channels=ch_out, + weight_attr=param_attr, + bias_attr=bias_attr) + norm_params = self.norm.parameters() + if freeze_norm: + for param in norm_params: + param.stop_gradient = True + conv_bias_attr = False + else: + conv_bias_attr = True + self.norm = None + + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=groups, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.001)), + bias_attr=conv_bias_attr) + + def forward(self, inputs): + out = self.conv(inputs) + if self.norm is not None: + out = self.norm(out) + + if self.act == 'relu': + out = F.relu(out) + elif self.act == 'sigmoid': + out = F.sigmoid(out) + return out + + +class DepthWiseSeparableConvNormLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + stride=1, + dw_norm_type=None, + pw_norm_type=None, + norm_decay=0., + freeze_norm=False, + dw_act=None, + pw_act=None): + super(DepthWiseSeparableConvNormLayer, self).__init__() + self.depthwise_conv = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_in, + filter_size=filter_size, + stride=stride, + groups=ch_in, + norm_type=dw_norm_type, + act=dw_act, + norm_decay=norm_decay, + freeze_norm=freeze_norm, ) + self.pointwise_conv = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=1, + norm_type=pw_norm_type, + act=pw_act, + norm_decay=norm_decay, + freeze_norm=freeze_norm, ) + + def forward(self, x): + x = self.depthwise_conv(x) + x = self.pointwise_conv(x) + return x + + +class CrossResolutionWeightingModule(nn.Layer): + def __init__(self, + channels, + ratio=16, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(CrossResolutionWeightingModule, self).__init__() + self.channels = channels + total_channel = sum(channels) + self.conv1 = ConvNormLayer( + ch_in=total_channel, + ch_out=total_channel // ratio, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + self.conv2 = ConvNormLayer( + ch_in=total_channel // ratio, + ch_out=total_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='sigmoid', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + + def forward(self, x): + mini_size = x[-1].shape[-2:] + out = [F.adaptive_avg_pool2d(s, mini_size) for s in x[:-1]] + [x[-1]] + out = paddle.concat(out, 1) + out = self.conv1(out) + out = self.conv2(out) + out = paddle.split(out, self.channels, 1) + out = [ + s * F.interpolate( + a, s.shape[-2:], mode='nearest') for s, a in zip(x, out) + ] + return out + + +class SpatialWeightingModule(nn.Layer): + def __init__(self, in_channel, ratio=16, freeze_norm=False, norm_decay=0.): + super(SpatialWeightingModule, self).__init__() + self.global_avgpooling = nn.AdaptiveAvgPool2D(1) + self.conv1 = ConvNormLayer( + ch_in=in_channel, + ch_out=in_channel // ratio, + filter_size=1, + stride=1, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + self.conv2 = ConvNormLayer( + ch_in=in_channel // ratio, + ch_out=in_channel, + filter_size=1, + stride=1, + act='sigmoid', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + + def forward(self, x): + out = self.global_avgpooling(x) + out = self.conv1(out) + out = self.conv2(out) + return x * out + + +class ConditionalChannelWeightingBlock(nn.Layer): + def __init__(self, + in_channels, + stride, + reduce_ratio, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(ConditionalChannelWeightingBlock, self).__init__() + assert stride in [1, 2] + branch_channels = [channel // 2 for channel in in_channels] + + self.cross_resolution_weighting = CrossResolutionWeightingModule( + branch_channels, + ratio=reduce_ratio, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay) + self.depthwise_convs = nn.LayerList([ + ConvNormLayer( + channel, + channel, + filter_size=3, + stride=stride, + groups=channel, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay) for channel in branch_channels + ]) + + self.spatial_weighting = nn.LayerList([ + SpatialWeightingModule( + channel, + ratio=4, + freeze_norm=freeze_norm, + norm_decay=norm_decay) for channel in branch_channels + ]) + + def forward(self, x): + x = [s.chunk(2, axis=1) for s in x] + x1 = [s[0] for s in x] + x2 = [s[1] for s in x] + + x2 = self.cross_resolution_weighting(x2) + x2 = [dw(s) for s, dw in zip(x2, self.depthwise_convs)] + x2 = [sw(s) for s, sw in zip(x2, self.spatial_weighting)] + + out = [paddle.concat([s1, s2], axis=1) for s1, s2 in zip(x1, x2)] + out = [channel_shuffle(s, groups=2) for s in out] + return out + + +class ShuffleUnit(nn.Layer): + def __init__(self, + in_channel, + out_channel, + stride, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(ShuffleUnit, self).__init__() + branch_channel = out_channel // 2 + self.stride = stride + if self.stride == 1: + assert in_channel == branch_channel * 2,\ + "when stride=1, in_channel {} should equal to branch_channel*2 {}".format(in_channel, branch_channel * 2) + if stride > 1: + self.branch1 = nn.Sequential( + ConvNormLayer( + ch_in=in_channel, + ch_out=in_channel, + filter_size=3, + stride=self.stride, + groups=in_channel, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay), + ConvNormLayer( + ch_in=in_channel, + ch_out=branch_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay), ) + self.branch2 = nn.Sequential( + ConvNormLayer( + ch_in=branch_channel if stride == 1 else in_channel, + ch_out=branch_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay), + ConvNormLayer( + ch_in=branch_channel, + ch_out=branch_channel, + filter_size=3, + stride=self.stride, + groups=branch_channel, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay), + ConvNormLayer( + ch_in=branch_channel, + ch_out=branch_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay), ) + + def forward(self, x): + if self.stride > 1: + x1 = self.branch1(x) + x2 = self.branch2(x) + else: + x1, x2 = x.chunk(2, axis=1) + x2 = self.branch2(x2) + out = paddle.concat([x1, x2], axis=1) + out = channel_shuffle(out, groups=2) + return out + + +class IterativeHead(nn.Layer): + def __init__(self, + in_channels, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(IterativeHead, self).__init__() + num_branches = len(in_channels) + self.in_channels = in_channels[::-1] + + projects = [] + for i in range(num_branches): + if i != num_branches - 1: + projects.append( + DepthWiseSeparableConvNormLayer( + ch_in=self.in_channels[i], + ch_out=self.in_channels[i + 1], + filter_size=3, + stride=1, + dw_act=None, + pw_act='relu', + dw_norm_type=norm_type, + pw_norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay)) + else: + projects.append( + DepthWiseSeparableConvNormLayer( + ch_in=self.in_channels[i], + ch_out=self.in_channels[i], + filter_size=3, + stride=1, + dw_act=None, + pw_act='relu', + dw_norm_type=norm_type, + pw_norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay)) + self.projects = nn.LayerList(projects) + + def forward(self, x): + x = x[::-1] + y = [] + last_x = None + for i, s in enumerate(x): + if last_x is not None: + last_x = F.interpolate( + last_x, + size=s.shape[-2:], + mode='bilinear', + align_corners=True) + s = s + last_x + s = self.projects[i](s) + y.append(s) + last_x = s + + return y[::-1] + + +class Stem(nn.Layer): + def __init__(self, + in_channel, + stem_channel, + out_channel, + expand_ratio, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(Stem, self).__init__() + self.conv1 = ConvNormLayer( + in_channel, + stem_channel, + filter_size=3, + stride=2, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + mid_channel = int(round(stem_channel * expand_ratio)) + branch_channel = stem_channel // 2 + if stem_channel == out_channel: + inc_channel = out_channel - branch_channel + else: + inc_channel = out_channel - stem_channel + self.branch1 = nn.Sequential( + ConvNormLayer( + ch_in=branch_channel, + ch_out=branch_channel, + filter_size=3, + stride=2, + groups=branch_channel, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay), + ConvNormLayer( + ch_in=branch_channel, + ch_out=inc_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay), ) + self.expand_conv = ConvNormLayer( + ch_in=branch_channel, + ch_out=mid_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + self.depthwise_conv = ConvNormLayer( + ch_in=mid_channel, + ch_out=mid_channel, + filter_size=3, + stride=2, + groups=mid_channel, + norm_type=norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay) + self.linear_conv = ConvNormLayer( + ch_in=mid_channel, + ch_out=branch_channel + if stem_channel == out_channel else stem_channel, + filter_size=1, + stride=1, + norm_type=norm_type, + act='relu', + freeze_norm=freeze_norm, + norm_decay=norm_decay) + + def forward(self, x): + x = self.conv1(x) + x1, x2 = x.chunk(2, axis=1) + x1 = self.branch1(x1) + x2 = self.expand_conv(x2) + x2 = self.depthwise_conv(x2) + x2 = self.linear_conv(x2) + out = paddle.concat([x1, x2], axis=1) + out = channel_shuffle(out, groups=2) + + return out + + +class LiteHRNetModule(nn.Layer): + def __init__(self, + num_branches, + num_blocks, + in_channels, + reduce_ratio, + module_type, + multiscale_output=False, + with_fuse=True, + norm_type='bn', + freeze_norm=False, + norm_decay=0.): + super(LiteHRNetModule, self).__init__() + assert num_branches == len(in_channels),\ + "num_branches {} should equal to num_in_channels {}".format(num_branches, len(in_channels)) + assert module_type in ['LITE', 'NAIVE'],\ + "module_type should be one of ['LITE', 'NAIVE']" + self.num_branches = num_branches + self.in_channels = in_channels + self.multiscale_output = multiscale_output + self.with_fuse = with_fuse + self.norm_type = 'bn' + self.module_type = module_type + + if self.module_type == 'LITE': + self.layers = self._make_weighting_blocks( + num_blocks, + reduce_ratio, + freeze_norm=freeze_norm, + norm_decay=norm_decay) + elif self.module_type == 'NAIVE': + self.layers = self._make_naive_branches( + num_branches, + num_blocks, + freeze_norm=freeze_norm, + norm_decay=norm_decay) + + if self.with_fuse: + self.fuse_layers = self._make_fuse_layers( + freeze_norm=freeze_norm, norm_decay=norm_decay) + self.relu = nn.ReLU() + + def _make_weighting_blocks(self, + num_blocks, + reduce_ratio, + stride=1, + freeze_norm=False, + norm_decay=0.): + layers = [] + for i in range(num_blocks): + layers.append( + ConditionalChannelWeightingBlock( + self.in_channels, + stride=stride, + reduce_ratio=reduce_ratio, + norm_type=self.norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay)) + return nn.Sequential(*layers) + + def _make_naive_branches(self, + num_branches, + num_blocks, + freeze_norm=False, + norm_decay=0.): + branches = [] + for branch_idx in range(num_branches): + layers = [] + for i in range(num_blocks): + layers.append( + ShuffleUnit( + self.in_channels[branch_idx], + self.in_channels[branch_idx], + stride=1, + norm_type=self.norm_type, + freeze_norm=freeze_norm, + norm_decay=norm_decay)) + branches.append(nn.Sequential(*layers)) + return nn.LayerList(branches) + + def _make_fuse_layers(self, freeze_norm=False, norm_decay=0.): + if self.num_branches == 1: + return None + fuse_layers = [] + num_out_branches = self.num_branches if self.multiscale_output else 1 + for i in range(num_out_branches): + fuse_layer = [] + for j in range(self.num_branches): + if j > i: + fuse_layer.append( + nn.Sequential( + L.Conv2d( + self.in_channels[j], + self.in_channels[i], + kernel_size=1, + stride=1, + padding=0, + bias=False, ), + nn.BatchNorm(self.in_channels[i]), + nn.Upsample( + scale_factor=2**(j - i), mode='nearest'))) + elif j == i: + fuse_layer.append(None) + else: + conv_downsamples = [] + for k in range(i - j): + if k == i - j - 1: + conv_downsamples.append( + nn.Sequential( + L.Conv2d( + self.in_channels[j], + self.in_channels[j], + kernel_size=3, + stride=2, + padding=1, + groups=self.in_channels[j], + bias=False, ), + nn.BatchNorm(self.in_channels[j]), + L.Conv2d( + self.in_channels[j], + self.in_channels[i], + kernel_size=1, + stride=1, + padding=0, + bias=False, ), + nn.BatchNorm(self.in_channels[i]))) + else: + conv_downsamples.append( + nn.Sequential( + L.Conv2d( + self.in_channels[j], + self.in_channels[j], + kernel_size=3, + stride=2, + padding=1, + groups=self.in_channels[j], + bias=False, ), + nn.BatchNorm(self.in_channels[j]), + L.Conv2d( + self.in_channels[j], + self.in_channels[j], + kernel_size=1, + stride=1, + padding=0, + bias=False, ), + nn.BatchNorm(self.in_channels[j]), + nn.ReLU())) + + fuse_layer.append(nn.Sequential(*conv_downsamples)) + fuse_layers.append(nn.LayerList(fuse_layer)) + + return nn.LayerList(fuse_layers) + + def forward(self, x): + if self.num_branches == 1: + return [self.layers[0](x[0])] + if self.module_type == 'LITE': + out = self.layers(x) + elif self.module_type == 'NAIVE': + for i in range(self.num_branches): + x[i] = self.layers[i](x[i]) + out = x + if self.with_fuse: + out_fuse = [] + for i in range(len(self.fuse_layers)): + y = out[0] if i == 0 else self.fuse_layers[i][0](out[0]) + for j in range(self.num_branches): + if j == 0: + y += y + elif i == j: + y += out[j] + else: + y += self.fuse_layers[i][j](out[j]) + if i == 0: + out[i] = y + out_fuse.append(self.relu(y)) + out = out_fuse + elif not self.multiscale_output: + out = [out[0]] + return out + + +@register +class LiteHRNet(nn.Layer): + """ + @inproceedings{Yulitehrnet21, + title={Lite-HRNet: A Lightweight High-Resolution Network}, + author={Yu, Changqian and Xiao, Bin and Gao, Changxin and Yuan, Lu and Zhang, Lei and Sang, Nong and Wang, Jingdong}, + booktitle={CVPR},year={2021} + } + Args: + network_type (str): the network_type should be one of ["lite_18", "lite_30", "naive", "wider_naive"], + "naive": Simply combining the shuffle block in ShuffleNet and the highresolution design pattern in HRNet. + "wider_naive": Naive network with wider channels in each block. + "lite_18": Lite-HRNet-18, which replaces the pointwise convolution in a shuffle block by conditional channel weighting. + "lite_30": Lite-HRNet-30, with more blocks compared with Lite-HRNet-18. + freeze_at (int): the stage to freeze + freeze_norm (bool): whether to freeze norm in HRNet + norm_decay (float): weight decay for normalization layer weights + return_idx (List): the stage to return + """ + + def __init__(self, + network_type, + freeze_at=0, + freeze_norm=True, + norm_decay=0., + return_idx=[0, 1, 2, 3]): + super(LiteHRNet, self).__init__() + if isinstance(return_idx, Integral): + return_idx = [return_idx] + assert network_type in ["lite_18", "lite_30", "naive", "wider_naive"],\ + "the network_type should be one of [lite_18, lite_30, naive, wider_naive]" + assert len(return_idx) > 0, "need one or more return index" + self.freeze_at = freeze_at + self.freeze_norm = freeze_norm + self.norm_decay = norm_decay + self.return_idx = return_idx + self.norm_type = 'bn' + + self.module_configs = { + "lite_18": { + "num_modules": [2, 4, 2], + "num_branches": [2, 3, 4], + "num_blocks": [2, 2, 2], + "module_type": ["LITE", "LITE", "LITE"], + "reduce_ratios": [8, 8, 8], + "num_channels": [[40, 80], [40, 80, 160], [40, 80, 160, 320]], + }, + "lite_30": { + "num_modules": [3, 8, 3], + "num_branches": [2, 3, 4], + "num_blocks": [2, 2, 2], + "module_type": ["LITE", "LITE", "LITE"], + "reduce_ratios": [8, 8, 8], + "num_channels": [[40, 80], [40, 80, 160], [40, 80, 160, 320]], + }, + "naive": { + "num_modules": [2, 4, 2], + "num_branches": [2, 3, 4], + "num_blocks": [2, 2, 2], + "module_type": ["NAIVE", "NAIVE", "NAIVE"], + "reduce_ratios": [1, 1, 1], + "num_channels": [[30, 60], [30, 60, 120], [30, 60, 120, 240]], + }, + "wider_naive": { + "num_modules": [2, 4, 2], + "num_branches": [2, 3, 4], + "num_blocks": [2, 2, 2], + "module_type": ["NAIVE", "NAIVE", "NAIVE"], + "reduce_ratios": [1, 1, 1], + "num_channels": [[40, 80], [40, 80, 160], [40, 80, 160, 320]], + }, + } + + self.stages_config = self.module_configs[network_type] + + self.stem = Stem(3, 32, 32, 1) + num_channels_pre_layer = [32] + for stage_idx in range(3): + num_channels = self.stages_config["num_channels"][stage_idx] + setattr(self, 'transition{}'.format(stage_idx), + self._make_transition_layer(num_channels_pre_layer, + num_channels, self.freeze_norm, + self.norm_decay)) + stage, num_channels_pre_layer = self._make_stage( + self.stages_config, stage_idx, num_channels, True, + self.freeze_norm, self.norm_decay) + setattr(self, 'stage{}'.format(stage_idx), stage) + self.head_layer = IterativeHead(num_channels_pre_layer, 'bn', + self.freeze_norm, self.norm_decay) + + def _make_transition_layer(self, + num_channels_pre_layer, + num_channels_cur_layer, + freeze_norm=False, + norm_decay=0.): + num_branches_pre = len(num_channels_pre_layer) + num_branches_cur = len(num_channels_cur_layer) + transition_layers = [] + for i in range(num_branches_cur): + if i < num_branches_pre: + if num_channels_cur_layer[i] != num_channels_pre_layer[i]: + transition_layers.append( + nn.Sequential( + L.Conv2d( + num_channels_pre_layer[i], + num_channels_pre_layer[i], + kernel_size=3, + stride=1, + padding=1, + groups=num_channels_pre_layer[i], + bias=False), + nn.BatchNorm(num_channels_pre_layer[i]), + L.Conv2d( + num_channels_pre_layer[i], + num_channels_cur_layer[i], + kernel_size=1, + stride=1, + padding=0, + bias=False, ), + nn.BatchNorm(num_channels_cur_layer[i]), + nn.ReLU())) + else: + transition_layers.append(None) + else: + conv_downsamples = [] + for j in range(i + 1 - num_branches_pre): + conv_downsamples.append( + nn.Sequential( + L.Conv2d( + num_channels_pre_layer[-1], + num_channels_pre_layer[-1], + groups=num_channels_pre_layer[-1], + kernel_size=3, + stride=2, + padding=1, + bias=False, ), + nn.BatchNorm(num_channels_pre_layer[-1]), + L.Conv2d( + num_channels_pre_layer[-1], + num_channels_cur_layer[i] + if j == i - num_branches_pre else + num_channels_pre_layer[-1], + kernel_size=1, + stride=1, + padding=0, + bias=False, ), + nn.BatchNorm(num_channels_cur_layer[i] + if j == i - num_branches_pre else + num_channels_pre_layer[-1]), + nn.ReLU())) + transition_layers.append(nn.Sequential(*conv_downsamples)) + return nn.LayerList(transition_layers) + + def _make_stage(self, + stages_config, + stage_idx, + in_channels, + multiscale_output, + freeze_norm=False, + norm_decay=0.): + num_modules = stages_config["num_modules"][stage_idx] + num_branches = stages_config["num_branches"][stage_idx] + num_blocks = stages_config["num_blocks"][stage_idx] + reduce_ratio = stages_config['reduce_ratios'][stage_idx] + module_type = stages_config['module_type'][stage_idx] + + modules = [] + for i in range(num_modules): + if not multiscale_output and i == num_modules - 1: + reset_multiscale_output = False + else: + reset_multiscale_output = True + modules.append( + LiteHRNetModule( + num_branches, + num_blocks, + in_channels, + reduce_ratio, + module_type, + multiscale_output=reset_multiscale_output, + with_fuse=True, + freeze_norm=freeze_norm, + norm_decay=norm_decay)) + in_channels = modules[-1].in_channels + return nn.Sequential(*modules), in_channels + + def forward(self, inputs): + x = inputs['image'] + x = self.stem(x) + y_list = [x] + for stage_idx in range(3): + x_list = [] + transition = getattr(self, 'transition{}'.format(stage_idx)) + for j in range(self.stages_config["num_branches"][stage_idx]): + if transition[j] is not None: + if j >= len(y_list): + x_list.append(transition[j](y_list[-1])) + else: + x_list.append(transition[j](y_list[j])) + else: + x_list.append(y_list[j]) + y_list = getattr(self, 'stage{}'.format(stage_idx))(x_list) + x = self.head_layer(y_list) + res = [] + for i, layer in enumerate(x): + if i == self.freeze_at: + layer.stop_gradient = True + if i in self.return_idx: + res.append(layer) + return res + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self._out_channels[i], stride=self._out_strides[i]) + for i in self.return_idx + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v1.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v1.py new file mode 100644 index 000000000..8cf602832 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v1.py @@ -0,0 +1,409 @@ +# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from paddle.nn.initializer import KaimingNormal +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec + +__all__ = ['MobileNet'] + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size, + stride, + padding, + num_groups=1, + act='relu', + conv_lr=1., + conv_decay=0., + norm_decay=0., + norm_type='bn', + name=None): + super(ConvBNLayer, self).__init__() + self.act = act + self._conv = nn.Conv2D( + in_channels, + out_channels, + kernel_size=kernel_size, + stride=stride, + padding=padding, + groups=num_groups, + weight_attr=ParamAttr( + learning_rate=conv_lr, + initializer=KaimingNormal(), + regularizer=L2Decay(conv_decay)), + bias_attr=False) + + param_attr = ParamAttr(regularizer=L2Decay(norm_decay)) + bias_attr = ParamAttr(regularizer=L2Decay(norm_decay)) + if norm_type == 'sync_bn': + self._batch_norm = nn.SyncBatchNorm( + out_channels, weight_attr=param_attr, bias_attr=bias_attr) + else: + self._batch_norm = nn.BatchNorm( + out_channels, + act=None, + param_attr=param_attr, + bias_attr=bias_attr, + use_global_stats=False) + + def forward(self, x): + x = self._conv(x) + x = self._batch_norm(x) + if self.act == "relu": + x = F.relu(x) + elif self.act == "relu6": + x = F.relu6(x) + return x + + +class DepthwiseSeparable(nn.Layer): + def __init__(self, + in_channels, + out_channels1, + out_channels2, + num_groups, + stride, + scale, + conv_lr=1., + conv_decay=0., + norm_decay=0., + norm_type='bn', + name=None): + super(DepthwiseSeparable, self).__init__() + + self._depthwise_conv = ConvBNLayer( + in_channels, + int(out_channels1 * scale), + kernel_size=3, + stride=stride, + padding=1, + num_groups=int(num_groups * scale), + conv_lr=conv_lr, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name=name + "_dw") + + self._pointwise_conv = ConvBNLayer( + int(out_channels1 * scale), + int(out_channels2 * scale), + kernel_size=1, + stride=1, + padding=0, + conv_lr=conv_lr, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name=name + "_sep") + + def forward(self, x): + x = self._depthwise_conv(x) + x = self._pointwise_conv(x) + return x + + +class ExtraBlock(nn.Layer): + def __init__(self, + in_channels, + out_channels1, + out_channels2, + num_groups=1, + stride=2, + conv_lr=1., + conv_decay=0., + norm_decay=0., + norm_type='bn', + name=None): + super(ExtraBlock, self).__init__() + + self.pointwise_conv = ConvBNLayer( + in_channels, + int(out_channels1), + kernel_size=1, + stride=1, + padding=0, + num_groups=int(num_groups), + act='relu6', + conv_lr=conv_lr, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name=name + "_extra1") + + self.normal_conv = ConvBNLayer( + int(out_channels1), + int(out_channels2), + kernel_size=3, + stride=stride, + padding=1, + num_groups=int(num_groups), + act='relu6', + conv_lr=conv_lr, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name=name + "_extra2") + + def forward(self, x): + x = self.pointwise_conv(x) + x = self.normal_conv(x) + return x + + +@register +@serializable +class MobileNet(nn.Layer): + __shared__ = ['norm_type'] + + def __init__(self, + norm_type='bn', + norm_decay=0., + conv_decay=0., + scale=1, + conv_learning_rate=1.0, + feature_maps=[4, 6, 13], + with_extra_blocks=False, + extra_block_filters=[[256, 512], [128, 256], [128, 256], + [64, 128]]): + super(MobileNet, self).__init__() + if isinstance(feature_maps, Integral): + feature_maps = [feature_maps] + self.feature_maps = feature_maps + self.with_extra_blocks = with_extra_blocks + self.extra_block_filters = extra_block_filters + + self._out_channels = [] + + self.conv1 = ConvBNLayer( + in_channels=3, + out_channels=int(32 * scale), + kernel_size=3, + stride=2, + padding=1, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv1") + + self.dwsl = [] + dws21 = self.add_sublayer( + "conv2_1", + sublayer=DepthwiseSeparable( + in_channels=int(32 * scale), + out_channels1=32, + out_channels2=64, + num_groups=32, + stride=1, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv2_1")) + self.dwsl.append(dws21) + self._update_out_channels(64, len(self.dwsl), feature_maps) + dws22 = self.add_sublayer( + "conv2_2", + sublayer=DepthwiseSeparable( + in_channels=int(64 * scale), + out_channels1=64, + out_channels2=128, + num_groups=64, + stride=2, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv2_2")) + self.dwsl.append(dws22) + self._update_out_channels(128, len(self.dwsl), feature_maps) + # 1/4 + dws31 = self.add_sublayer( + "conv3_1", + sublayer=DepthwiseSeparable( + in_channels=int(128 * scale), + out_channels1=128, + out_channels2=128, + num_groups=128, + stride=1, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv3_1")) + self.dwsl.append(dws31) + self._update_out_channels(128, len(self.dwsl), feature_maps) + dws32 = self.add_sublayer( + "conv3_2", + sublayer=DepthwiseSeparable( + in_channels=int(128 * scale), + out_channels1=128, + out_channels2=256, + num_groups=128, + stride=2, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv3_2")) + self.dwsl.append(dws32) + self._update_out_channels(256, len(self.dwsl), feature_maps) + # 1/8 + dws41 = self.add_sublayer( + "conv4_1", + sublayer=DepthwiseSeparable( + in_channels=int(256 * scale), + out_channels1=256, + out_channels2=256, + num_groups=256, + stride=1, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv4_1")) + self.dwsl.append(dws41) + self._update_out_channels(256, len(self.dwsl), feature_maps) + dws42 = self.add_sublayer( + "conv4_2", + sublayer=DepthwiseSeparable( + in_channels=int(256 * scale), + out_channels1=256, + out_channels2=512, + num_groups=256, + stride=2, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv4_2")) + self.dwsl.append(dws42) + self._update_out_channels(512, len(self.dwsl), feature_maps) + # 1/16 + for i in range(5): + tmp = self.add_sublayer( + "conv5_" + str(i + 1), + sublayer=DepthwiseSeparable( + in_channels=512, + out_channels1=512, + out_channels2=512, + num_groups=512, + stride=1, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv5_" + str(i + 1))) + self.dwsl.append(tmp) + self._update_out_channels(512, len(self.dwsl), feature_maps) + dws56 = self.add_sublayer( + "conv5_6", + sublayer=DepthwiseSeparable( + in_channels=int(512 * scale), + out_channels1=512, + out_channels2=1024, + num_groups=512, + stride=2, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv5_6")) + self.dwsl.append(dws56) + self._update_out_channels(1024, len(self.dwsl), feature_maps) + # 1/32 + dws6 = self.add_sublayer( + "conv6", + sublayer=DepthwiseSeparable( + in_channels=int(1024 * scale), + out_channels1=1024, + out_channels2=1024, + num_groups=1024, + stride=1, + scale=scale, + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv6")) + self.dwsl.append(dws6) + self._update_out_channels(1024, len(self.dwsl), feature_maps) + + if self.with_extra_blocks: + self.extra_blocks = [] + for i, block_filter in enumerate(self.extra_block_filters): + in_c = 1024 if i == 0 else self.extra_block_filters[i - 1][1] + conv_extra = self.add_sublayer( + "conv7_" + str(i + 1), + sublayer=ExtraBlock( + in_c, + block_filter[0], + block_filter[1], + conv_lr=conv_learning_rate, + conv_decay=conv_decay, + norm_decay=norm_decay, + norm_type=norm_type, + name="conv7_" + str(i + 1))) + self.extra_blocks.append(conv_extra) + self._update_out_channels( + block_filter[1], + len(self.dwsl) + len(self.extra_blocks), feature_maps) + + def _update_out_channels(self, channel, feature_idx, feature_maps): + if feature_idx in feature_maps: + self._out_channels.append(channel) + + def forward(self, inputs): + outs = [] + y = self.conv1(inputs['image']) + for i, block in enumerate(self.dwsl): + y = block(y) + if i + 1 in self.feature_maps: + outs.append(y) + + if not self.with_extra_blocks: + return outs + + y = outs[-1] + for i, block in enumerate(self.extra_blocks): + idx = i + len(self.dwsl) + y = block(y) + if idx + 1 in self.feature_maps: + outs.append(y) + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v3.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v3.py new file mode 100644 index 000000000..02021e87c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/mobilenet_v3.py @@ -0,0 +1,482 @@ +# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec + +__all__ = ['MobileNetV3'] + + +def make_divisible(v, divisor=8, min_value=None): + if min_value is None: + min_value = divisor + new_v = max(min_value, int(v + divisor / 2) // divisor * divisor) + if new_v < 0.9 * v: + new_v += divisor + return new_v + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_c, + out_c, + filter_size, + stride, + padding, + num_groups=1, + act=None, + lr_mult=1., + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + name=""): + super(ConvBNLayer, self).__init__() + self.act = act + self.conv = nn.Conv2D( + in_channels=in_c, + out_channels=out_c, + kernel_size=filter_size, + stride=stride, + padding=padding, + groups=num_groups, + weight_attr=ParamAttr( + learning_rate=lr_mult, regularizer=L2Decay(conv_decay)), + bias_attr=False) + + norm_lr = 0. if freeze_norm else lr_mult + param_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + bias_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + global_stats = True if freeze_norm else False + if norm_type == 'sync_bn': + self.bn = nn.SyncBatchNorm( + out_c, weight_attr=param_attr, bias_attr=bias_attr) + else: + self.bn = nn.BatchNorm( + out_c, + act=None, + param_attr=param_attr, + bias_attr=bias_attr, + use_global_stats=global_stats) + norm_params = self.bn.parameters() + if freeze_norm: + for param in norm_params: + param.stop_gradient = True + + def forward(self, x): + x = self.conv(x) + x = self.bn(x) + if self.act is not None: + if self.act == "relu": + x = F.relu(x) + elif self.act == "relu6": + x = F.relu6(x) + elif self.act == "hard_swish": + x = F.hardswish(x) + else: + raise NotImplementedError( + "The activation function is selected incorrectly.") + return x + + +class ResidualUnit(nn.Layer): + def __init__(self, + in_c, + mid_c, + out_c, + filter_size, + stride, + use_se, + lr_mult, + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + act=None, + return_list=False, + name=''): + super(ResidualUnit, self).__init__() + self.if_shortcut = stride == 1 and in_c == out_c + self.use_se = use_se + self.return_list = return_list + + self.expand_conv = ConvBNLayer( + in_c=in_c, + out_c=mid_c, + filter_size=1, + stride=1, + padding=0, + act=act, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_expand") + self.bottleneck_conv = ConvBNLayer( + in_c=mid_c, + out_c=mid_c, + filter_size=filter_size, + stride=stride, + padding=int((filter_size - 1) // 2), + num_groups=mid_c, + act=act, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_depthwise") + if self.use_se: + self.mid_se = SEModule( + mid_c, lr_mult, conv_decay, name=name + "_se") + self.linear_conv = ConvBNLayer( + in_c=mid_c, + out_c=out_c, + filter_size=1, + stride=1, + padding=0, + act=None, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_linear") + + def forward(self, inputs): + y = self.expand_conv(inputs) + x = self.bottleneck_conv(y) + if self.use_se: + x = self.mid_se(x) + x = self.linear_conv(x) + if self.if_shortcut: + x = paddle.add(inputs, x) + if self.return_list: + return [y, x] + else: + return x + + +class SEModule(nn.Layer): + def __init__(self, channel, lr_mult, conv_decay, reduction=4, name=""): + super(SEModule, self).__init__() + self.avg_pool = nn.AdaptiveAvgPool2D(1) + mid_channels = int(channel // reduction) + self.conv1 = nn.Conv2D( + in_channels=channel, + out_channels=mid_channels, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr( + learning_rate=lr_mult, regularizer=L2Decay(conv_decay)), + bias_attr=ParamAttr( + learning_rate=lr_mult, regularizer=L2Decay(conv_decay))) + self.conv2 = nn.Conv2D( + in_channels=mid_channels, + out_channels=channel, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr( + learning_rate=lr_mult, regularizer=L2Decay(conv_decay)), + bias_attr=ParamAttr( + learning_rate=lr_mult, regularizer=L2Decay(conv_decay))) + + def forward(self, inputs): + outputs = self.avg_pool(inputs) + outputs = self.conv1(outputs) + outputs = F.relu(outputs) + outputs = self.conv2(outputs) + outputs = F.hardsigmoid(outputs, slope=0.2, offset=0.5) + return paddle.multiply(x=inputs, y=outputs) + + +class ExtraBlockDW(nn.Layer): + def __init__(self, + in_c, + ch_1, + ch_2, + stride, + lr_mult, + conv_decay=0., + norm_type='bn', + norm_decay=0., + freeze_norm=False, + name=None): + super(ExtraBlockDW, self).__init__() + self.pointwise_conv = ConvBNLayer( + in_c=in_c, + out_c=ch_1, + filter_size=1, + stride=1, + padding='SAME', + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra1") + self.depthwise_conv = ConvBNLayer( + in_c=ch_1, + out_c=ch_2, + filter_size=3, + stride=stride, + padding='SAME', + num_groups=int(ch_1), + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra2_dw") + self.normal_conv = ConvBNLayer( + in_c=ch_2, + out_c=ch_2, + filter_size=1, + stride=1, + padding='SAME', + act='relu6', + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name=name + "_extra2_sep") + + def forward(self, inputs): + x = self.pointwise_conv(inputs) + x = self.depthwise_conv(x) + x = self.normal_conv(x) + return x + + +@register +@serializable +class MobileNetV3(nn.Layer): + __shared__ = ['norm_type'] + + def __init__( + self, + scale=1.0, + model_name="large", + feature_maps=[6, 12, 15], + with_extra_blocks=False, + extra_block_filters=[[256, 512], [128, 256], [128, 256], [64, 128]], + lr_mult_list=[1.0, 1.0, 1.0, 1.0, 1.0], + conv_decay=0.0, + multiplier=1.0, + norm_type='bn', + norm_decay=0.0, + freeze_norm=False): + super(MobileNetV3, self).__init__() + if isinstance(feature_maps, Integral): + feature_maps = [feature_maps] + if norm_type == 'sync_bn' and freeze_norm: + raise ValueError( + "The norm_type should not be sync_bn when freeze_norm is True") + self.feature_maps = feature_maps + self.with_extra_blocks = with_extra_blocks + self.extra_block_filters = extra_block_filters + + inplanes = 16 + if model_name == "large": + self.cfg = [ + # k, exp, c, se, nl, s, + [3, 16, 16, False, "relu", 1], + [3, 64, 24, False, "relu", 2], + [3, 72, 24, False, "relu", 1], + [5, 72, 40, True, "relu", 2], # RCNN output + [5, 120, 40, True, "relu", 1], + [5, 120, 40, True, "relu", 1], # YOLOv3 output + [3, 240, 80, False, "hard_swish", 2], # RCNN output + [3, 200, 80, False, "hard_swish", 1], + [3, 184, 80, False, "hard_swish", 1], + [3, 184, 80, False, "hard_swish", 1], + [3, 480, 112, True, "hard_swish", 1], + [3, 672, 112, True, "hard_swish", 1], # YOLOv3 output + [5, 672, 160, True, "hard_swish", 2], # SSD/SSDLite/RCNN output + [5, 960, 160, True, "hard_swish", 1], + [5, 960, 160, True, "hard_swish", 1], # YOLOv3 output + ] + elif model_name == "small": + self.cfg = [ + # k, exp, c, se, nl, s, + [3, 16, 16, True, "relu", 2], + [3, 72, 24, False, "relu", 2], # RCNN output + [3, 88, 24, False, "relu", 1], # YOLOv3 output + [5, 96, 40, True, "hard_swish", 2], # RCNN output + [5, 240, 40, True, "hard_swish", 1], + [5, 240, 40, True, "hard_swish", 1], + [5, 120, 48, True, "hard_swish", 1], + [5, 144, 48, True, "hard_swish", 1], # YOLOv3 output + [5, 288, 96, True, "hard_swish", 2], # SSD/SSDLite/RCNN output + [5, 576, 96, True, "hard_swish", 1], + [5, 576, 96, True, "hard_swish", 1], # YOLOv3 output + ] + else: + raise NotImplementedError( + "mode[{}_model] is not implemented!".format(model_name)) + + if multiplier != 1.0: + self.cfg[-3][2] = int(self.cfg[-3][2] * multiplier) + self.cfg[-2][1] = int(self.cfg[-2][1] * multiplier) + self.cfg[-2][2] = int(self.cfg[-2][2] * multiplier) + self.cfg[-1][1] = int(self.cfg[-1][1] * multiplier) + self.cfg[-1][2] = int(self.cfg[-1][2] * multiplier) + + self.conv1 = ConvBNLayer( + in_c=3, + out_c=make_divisible(inplanes * scale), + filter_size=3, + stride=2, + padding=1, + num_groups=1, + act="hard_swish", + lr_mult=lr_mult_list[0], + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="conv1") + + self._out_channels = [] + self.block_list = [] + i = 0 + inplanes = make_divisible(inplanes * scale) + for (k, exp, c, se, nl, s) in self.cfg: + lr_idx = min(i // 3, len(lr_mult_list) - 1) + lr_mult = lr_mult_list[lr_idx] + + # for SSD/SSDLite, first head input is after ResidualUnit expand_conv + return_list = self.with_extra_blocks and i + 2 in self.feature_maps + + block = self.add_sublayer( + "conv" + str(i + 2), + sublayer=ResidualUnit( + in_c=inplanes, + mid_c=make_divisible(scale * exp), + out_c=make_divisible(scale * c), + filter_size=k, + stride=s, + use_se=se, + act=nl, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + return_list=return_list, + name="conv" + str(i + 2))) + self.block_list.append(block) + inplanes = make_divisible(scale * c) + i += 1 + self._update_out_channels( + make_divisible(scale * exp) + if return_list else inplanes, i + 1, feature_maps) + + if self.with_extra_blocks: + self.extra_block_list = [] + extra_out_c = make_divisible(scale * self.cfg[-1][1]) + lr_idx = min(i // 3, len(lr_mult_list) - 1) + lr_mult = lr_mult_list[lr_idx] + + conv_extra = self.add_sublayer( + "conv" + str(i + 2), + sublayer=ConvBNLayer( + in_c=inplanes, + out_c=extra_out_c, + filter_size=1, + stride=1, + padding=0, + num_groups=1, + act="hard_swish", + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name="conv" + str(i + 2))) + self.extra_block_list.append(conv_extra) + i += 1 + self._update_out_channels(extra_out_c, i + 1, feature_maps) + + for j, block_filter in enumerate(self.extra_block_filters): + in_c = extra_out_c if j == 0 else self.extra_block_filters[j - + 1][1] + conv_extra = self.add_sublayer( + "conv" + str(i + 2), + sublayer=ExtraBlockDW( + in_c, + block_filter[0], + block_filter[1], + stride=2, + lr_mult=lr_mult, + conv_decay=conv_decay, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + name='conv' + str(i + 2))) + self.extra_block_list.append(conv_extra) + i += 1 + self._update_out_channels(block_filter[1], i + 1, feature_maps) + + def _update_out_channels(self, channel, feature_idx, feature_maps): + if feature_idx in feature_maps: + self._out_channels.append(channel) + + def forward(self, inputs): + x = self.conv1(inputs['image']) + outs = [] + for idx, block in enumerate(self.block_list): + x = block(x) + if idx + 2 in self.feature_maps: + if isinstance(x, list): + outs.append(x[0]) + x = x[1] + else: + outs.append(x) + + if not self.with_extra_blocks: + return outs + + for i, block in enumerate(self.extra_block_list): + idx = i + len(self.block_list) + x = block(x) + if idx + 2 in self.feature_maps: + outs.append(x) + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/name_adapter.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/name_adapter.py new file mode 100644 index 000000000..4afbb9b18 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/name_adapter.py @@ -0,0 +1,69 @@ +class NameAdapter(object): + """Fix the backbones variable names for pretrained weight""" + + def __init__(self, model): + super(NameAdapter, self).__init__() + self.model = model + + @property + def model_type(self): + return getattr(self.model, '_model_type', '') + + @property + def variant(self): + return getattr(self.model, 'variant', '') + + def fix_conv_norm_name(self, name): + if name == "conv1": + bn_name = "bn_" + name + else: + bn_name = "bn" + name[3:] + # the naming rule is same as pretrained weight + if self.model_type == 'SEResNeXt': + bn_name = name + "_bn" + return bn_name + + def fix_shortcut_name(self, name): + if self.model_type == 'SEResNeXt': + name = 'conv' + name + '_prj' + return name + + def fix_bottleneck_name(self, name): + if self.model_type == 'SEResNeXt': + conv_name1 = 'conv' + name + '_x1' + conv_name2 = 'conv' + name + '_x2' + conv_name3 = 'conv' + name + '_x3' + shortcut_name = name + else: + conv_name1 = name + "_branch2a" + conv_name2 = name + "_branch2b" + conv_name3 = name + "_branch2c" + shortcut_name = name + "_branch1" + return conv_name1, conv_name2, conv_name3, shortcut_name + + def fix_basicblock_name(self, name): + if self.model_type == 'SEResNeXt': + conv_name1 = 'conv' + name + '_x1' + conv_name2 = 'conv' + name + '_x2' + shortcut_name = name + else: + conv_name1 = name + "_branch2a" + conv_name2 = name + "_branch2b" + shortcut_name = name + "_branch1" + return conv_name1, conv_name2, shortcut_name + + def fix_layer_warp_name(self, stage_num, count, i): + name = 'res' + str(stage_num) + if count > 10 and stage_num == 4: + if i == 0: + conv_name = name + "a" + else: + conv_name = name + "b" + str(i) + else: + conv_name = name + chr(ord("a") + i) + if self.model_type == 'SEResNeXt': + conv_name = str(stage_num + 2) + '_' + str(i + 1) + return conv_name + + def fix_c1_stage_name(self): + return "res_conv1" if self.model_type == 'ResNeXt' else "conv1" diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/res2net.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/res2net.py new file mode 100644 index 000000000..9e7677247 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/res2net.py @@ -0,0 +1,357 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from numbers import Integral + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ..shape_spec import ShapeSpec +from .resnet import ConvNormLayer + +__all__ = ['Res2Net', 'Res2NetC5'] + +Res2Net_cfg = { + 50: [3, 4, 6, 3], + 101: [3, 4, 23, 3], + 152: [3, 8, 36, 3], + 200: [3, 12, 48, 3] +} + + +class BottleNeck(nn.Layer): + def __init__(self, + ch_in, + ch_out, + stride, + shortcut, + width, + scales=4, + variant='b', + groups=1, + lr=1.0, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + dcn_v2=False): + super(BottleNeck, self).__init__() + + self.shortcut = shortcut + self.scales = scales + self.stride = stride + if not shortcut: + if variant == 'd' and stride == 2: + self.branch1 = nn.Sequential() + self.branch1.add_sublayer( + 'pool', + nn.AvgPool2D( + kernel_size=2, stride=2, padding=0, ceil_mode=True)) + self.branch1.add_sublayer( + 'conv', + ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr)) + else: + self.branch1 = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=stride, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2a = ConvNormLayer( + ch_in=ch_in, + ch_out=width * scales, + filter_size=1, + stride=stride if variant == 'a' else 1, + groups=1, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2b = nn.LayerList([ + ConvNormLayer( + ch_in=width, + ch_out=width, + filter_size=3, + stride=1 if variant == 'a' else stride, + groups=groups, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr, + dcn_v2=dcn_v2) for _ in range(self.scales - 1) + ]) + + self.branch2c = ConvNormLayer( + ch_in=width * scales, + ch_out=ch_out, + filter_size=1, + stride=1, + groups=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + def forward(self, inputs): + + out = self.branch2a(inputs) + feature_split = paddle.split(out, self.scales, 1) + out_split = [] + for i in range(self.scales - 1): + if i == 0 or self.stride == 2: + out_split.append(self.branch2b[i](feature_split[i])) + else: + out_split.append(self.branch2b[i](paddle.add(feature_split[i], + out_split[-1]))) + if self.stride == 1: + out_split.append(feature_split[-1]) + else: + out_split.append(F.avg_pool2d(feature_split[-1], 3, self.stride, 1)) + out = self.branch2c(paddle.concat(out_split, 1)) + + if self.shortcut: + short = inputs + else: + short = self.branch1(inputs) + + out = paddle.add(out, short) + out = F.relu(out) + + return out + + +class Blocks(nn.Layer): + def __init__(self, + ch_in, + ch_out, + count, + stage_num, + width, + scales=4, + variant='b', + groups=1, + lr=1.0, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + dcn_v2=False): + super(Blocks, self).__init__() + + self.blocks = nn.Sequential() + for i in range(count): + self.blocks.add_sublayer( + str(i), + BottleNeck( + ch_in=ch_in if i == 0 else ch_out, + ch_out=ch_out, + stride=2 if i == 0 and stage_num != 2 else 1, + shortcut=False if i == 0 else True, + width=width * (2**(stage_num - 2)), + scales=scales, + variant=variant, + groups=groups, + lr=lr, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + dcn_v2=dcn_v2)) + + def forward(self, inputs): + return self.blocks(inputs) + + +@register +@serializable +class Res2Net(nn.Layer): + """ + Res2Net, see https://arxiv.org/abs/1904.01169 + Args: + depth (int): Res2Net depth, should be 50, 101, 152, 200. + width (int): Res2Net width + scales (int): Res2Net scale + variant (str): Res2Net variant, supports 'a', 'b', 'c', 'd' currently + lr_mult_list (list): learning rate ratio of different resnet stages(2,3,4,5), + lower learning rate ratio is need for pretrained model + got using distillation(default as [1.0, 1.0, 1.0, 1.0]). + groups (int): The groups number of the Conv Layer. + norm_type (str): normalization type, 'bn' or 'sync_bn' + norm_decay (float): weight decay for normalization layer weights + freeze_norm (bool): freeze normalization layers + freeze_at (int): freeze the backbone at which stage + return_idx (list): index of stages whose feature maps are returned, + index 0 stands for res2 + dcn_v2_stages (list): index of stages who select deformable conv v2 + num_stages (int): number of stages created + + """ + __shared__ = ['norm_type'] + + def __init__(self, + depth=50, + width=26, + scales=4, + variant='b', + lr_mult_list=[1.0, 1.0, 1.0, 1.0], + groups=1, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + freeze_at=0, + return_idx=[0, 1, 2, 3], + dcn_v2_stages=[-1], + num_stages=4): + super(Res2Net, self).__init__() + + self._model_type = 'Res2Net' if groups == 1 else 'Res2NeXt' + + assert depth in [50, 101, 152, 200], \ + "depth {} not in [50, 101, 152, 200]" + assert variant in ['a', 'b', 'c', 'd'], "invalid Res2Net variant" + assert num_stages >= 1 and num_stages <= 4 + + self.depth = depth + self.variant = variant + self.norm_type = norm_type + self.norm_decay = norm_decay + self.freeze_norm = freeze_norm + self.freeze_at = freeze_at + if isinstance(return_idx, Integral): + return_idx = [return_idx] + assert max(return_idx) < num_stages, \ + 'the maximum return index must smaller than num_stages, ' \ + 'but received maximum return index is {} and num_stages ' \ + 'is {}'.format(max(return_idx), num_stages) + self.return_idx = return_idx + self.num_stages = num_stages + assert len(lr_mult_list) == 4, \ + "lr_mult_list length must be 4 but got {}".format(len(lr_mult_list)) + if isinstance(dcn_v2_stages, Integral): + dcn_v2_stages = [dcn_v2_stages] + assert max(dcn_v2_stages) < num_stages + self.dcn_v2_stages = dcn_v2_stages + + block_nums = Res2Net_cfg[depth] + + # C1 stage + if self.variant in ['c', 'd']: + conv_def = [ + [3, 32, 3, 2, "conv1_1"], + [32, 32, 3, 1, "conv1_2"], + [32, 64, 3, 1, "conv1_3"], + ] + else: + conv_def = [[3, 64, 7, 2, "conv1"]] + self.res1 = nn.Sequential() + for (c_in, c_out, k, s, _name) in conv_def: + self.res1.add_sublayer( + _name, + ConvNormLayer( + ch_in=c_in, + ch_out=c_out, + filter_size=k, + stride=s, + groups=1, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=1.0)) + + self._in_channels = [64, 256, 512, 1024] + self._out_channels = [256, 512, 1024, 2048] + self._out_strides = [4, 8, 16, 32] + + # C2-C5 stages + self.res_layers = [] + for i in range(num_stages): + lr_mult = lr_mult_list[i] + stage_num = i + 2 + self.res_layers.append( + self.add_sublayer( + "res{}".format(stage_num), + Blocks( + self._in_channels[i], + self._out_channels[i], + count=block_nums[i], + stage_num=stage_num, + width=width, + scales=scales, + groups=groups, + lr=lr_mult, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + dcn_v2=(i in self.dcn_v2_stages)))) + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self._out_channels[i], stride=self._out_strides[i]) + for i in self.return_idx + ] + + def forward(self, inputs): + x = inputs['image'] + res1 = self.res1(x) + x = F.max_pool2d(res1, kernel_size=3, stride=2, padding=1) + outs = [] + for idx, stage in enumerate(self.res_layers): + x = stage(x) + if idx == self.freeze_at: + x.stop_gradient = True + if idx in self.return_idx: + outs.append(x) + return outs + + +@register +class Res2NetC5(nn.Layer): + def __init__(self, depth=50, width=26, scales=4, variant='b'): + super(Res2NetC5, self).__init__() + feat_in, feat_out = [1024, 2048] + self.res5 = Blocks( + feat_in, + feat_out, + count=3, + stage_num=5, + width=width, + scales=scales, + variant=variant) + self.feat_out = feat_out + + @property + def out_shape(self): + return [ShapeSpec( + channels=self.feat_out, + stride=32, )] + + def forward(self, roi_feat, stage=0): + y = self.res5(roi_feat) + return y diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/resnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/resnet.py new file mode 100644 index 000000000..d4bc878ea --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/resnet.py @@ -0,0 +1,613 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import math +from numbers import Integral + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from paddle.regularizer import L2Decay +from paddle.nn.initializer import Uniform +from paddle import ParamAttr +from paddle.nn.initializer import Constant +from paddle.vision.ops import DeformConv2D +from .name_adapter import NameAdapter +from ..shape_spec import ShapeSpec + +__all__ = ['ResNet', 'Res5Head', 'Blocks', 'BasicBlock', 'BottleNeck'] + +ResNet_cfg = { + 18: [2, 2, 2, 2], + 34: [3, 4, 6, 3], + 50: [3, 4, 6, 3], + 101: [3, 4, 23, 3], + 152: [3, 8, 36, 3], +} + + +class ConvNormLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + stride, + groups=1, + act=None, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + lr=1.0, + dcn_v2=False): + super(ConvNormLayer, self).__init__() + assert norm_type in ['bn', 'sync_bn'] + self.norm_type = norm_type + self.act = act + self.dcn_v2 = dcn_v2 + + if not self.dcn_v2: + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=groups, + weight_attr=ParamAttr(learning_rate=lr), + bias_attr=False) + else: + self.offset_channel = 2 * filter_size**2 + self.mask_channel = filter_size**2 + + self.conv_offset = nn.Conv2D( + in_channels=ch_in, + out_channels=3 * filter_size**2, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + weight_attr=ParamAttr(initializer=Constant(0.)), + bias_attr=ParamAttr(initializer=Constant(0.))) + self.conv = DeformConv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + dilation=1, + groups=groups, + weight_attr=ParamAttr(learning_rate=lr), + bias_attr=False) + + norm_lr = 0. if freeze_norm else lr + param_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + bias_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + + global_stats = True if freeze_norm else False + if norm_type == 'sync_bn': + self.norm = nn.SyncBatchNorm( + ch_out, weight_attr=param_attr, bias_attr=bias_attr) + else: + self.norm = nn.BatchNorm( + ch_out, + act=None, + param_attr=param_attr, + bias_attr=bias_attr, + use_global_stats=global_stats) + norm_params = self.norm.parameters() + + if freeze_norm: + for param in norm_params: + param.stop_gradient = True + + def forward(self, inputs): + if not self.dcn_v2: + out = self.conv(inputs) + else: + offset_mask = self.conv_offset(inputs) + offset, mask = paddle.split( + offset_mask, + num_or_sections=[self.offset_channel, self.mask_channel], + axis=1) + mask = F.sigmoid(mask) + out = self.conv(inputs, offset, mask=mask) + + if self.norm_type in ['bn', 'sync_bn']: + out = self.norm(out) + if self.act: + out = getattr(F, self.act)(out) + return out + + +class SELayer(nn.Layer): + def __init__(self, ch, reduction_ratio=16): + super(SELayer, self).__init__() + self.pool = nn.AdaptiveAvgPool2D(1) + stdv = 1.0 / math.sqrt(ch) + c_ = ch // reduction_ratio + self.squeeze = nn.Linear( + ch, + c_, + weight_attr=paddle.ParamAttr(initializer=Uniform(-stdv, stdv)), + bias_attr=True) + + stdv = 1.0 / math.sqrt(c_) + self.extract = nn.Linear( + c_, + ch, + weight_attr=paddle.ParamAttr(initializer=Uniform(-stdv, stdv)), + bias_attr=True) + + def forward(self, inputs): + out = self.pool(inputs) + out = paddle.squeeze(out, axis=[2, 3]) + out = self.squeeze(out) + out = F.relu(out) + out = self.extract(out) + out = F.sigmoid(out) + out = paddle.unsqueeze(out, axis=[2, 3]) + scale = out * inputs + return scale + + +class BasicBlock(nn.Layer): + + expansion = 1 + + def __init__(self, + ch_in, + ch_out, + stride, + shortcut, + variant='b', + groups=1, + base_width=64, + lr=1.0, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + dcn_v2=False, + std_senet=False): + super(BasicBlock, self).__init__() + assert groups == 1 and base_width == 64, 'BasicBlock only supports groups=1 and base_width=64' + + self.shortcut = shortcut + if not shortcut: + if variant == 'd' and stride == 2: + self.short = nn.Sequential() + self.short.add_sublayer( + 'pool', + nn.AvgPool2D( + kernel_size=2, stride=2, padding=0, ceil_mode=True)) + self.short.add_sublayer( + 'conv', + ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr)) + else: + self.short = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=1, + stride=stride, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2a = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=3, + stride=stride, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2b = ConvNormLayer( + ch_in=ch_out, + ch_out=ch_out, + filter_size=3, + stride=1, + act=None, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr, + dcn_v2=dcn_v2) + + self.std_senet = std_senet + if self.std_senet: + self.se = SELayer(ch_out) + + def forward(self, inputs): + out = self.branch2a(inputs) + out = self.branch2b(out) + if self.std_senet: + out = self.se(out) + + if self.shortcut: + short = inputs + else: + short = self.short(inputs) + + out = paddle.add(x=out, y=short) + out = F.relu(out) + + return out + + +class BottleNeck(nn.Layer): + + expansion = 4 + + def __init__(self, + ch_in, + ch_out, + stride, + shortcut, + variant='b', + groups=1, + base_width=4, + lr=1.0, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + dcn_v2=False, + std_senet=False): + super(BottleNeck, self).__init__() + if variant == 'a': + stride1, stride2 = stride, 1 + else: + stride1, stride2 = 1, stride + + # ResNeXt + width = int(ch_out * (base_width / 64.)) * groups + + self.shortcut = shortcut + if not shortcut: + if variant == 'd' and stride == 2: + self.short = nn.Sequential() + self.short.add_sublayer( + 'pool', + nn.AvgPool2D( + kernel_size=2, stride=2, padding=0, ceil_mode=True)) + self.short.add_sublayer( + 'conv', + ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out * self.expansion, + filter_size=1, + stride=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr)) + else: + self.short = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out * self.expansion, + filter_size=1, + stride=stride, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2a = ConvNormLayer( + ch_in=ch_in, + ch_out=width, + filter_size=1, + stride=stride1, + groups=1, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.branch2b = ConvNormLayer( + ch_in=width, + ch_out=width, + filter_size=3, + stride=stride2, + groups=groups, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr, + dcn_v2=dcn_v2) + + self.branch2c = ConvNormLayer( + ch_in=width, + ch_out=ch_out * self.expansion, + filter_size=1, + stride=1, + groups=1, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=lr) + + self.std_senet = std_senet + if self.std_senet: + self.se = SELayer(ch_out * self.expansion) + + def forward(self, inputs): + + out = self.branch2a(inputs) + out = self.branch2b(out) + out = self.branch2c(out) + + if self.std_senet: + out = self.se(out) + + if self.shortcut: + short = inputs + else: + short = self.short(inputs) + + out = paddle.add(x=out, y=short) + out = F.relu(out) + + return out + + +class Blocks(nn.Layer): + def __init__(self, + block, + ch_in, + ch_out, + count, + name_adapter, + stage_num, + variant='b', + groups=1, + base_width=64, + lr=1.0, + norm_type='bn', + norm_decay=0., + freeze_norm=True, + dcn_v2=False, + std_senet=False): + super(Blocks, self).__init__() + + self.blocks = [] + for i in range(count): + conv_name = name_adapter.fix_layer_warp_name(stage_num, count, i) + layer = self.add_sublayer( + conv_name, + block( + ch_in=ch_in, + ch_out=ch_out, + stride=2 if i == 0 and stage_num != 2 else 1, + shortcut=False if i == 0 else True, + variant=variant, + groups=groups, + base_width=base_width, + lr=lr, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + dcn_v2=dcn_v2, + std_senet=std_senet)) + self.blocks.append(layer) + if i == 0: + ch_in = ch_out * block.expansion + + def forward(self, inputs): + block_out = inputs + for block in self.blocks: + block_out = block(block_out) + return block_out + + +@register +@serializable +class ResNet(nn.Layer): + __shared__ = ['norm_type'] + + def __init__(self, + depth=50, + ch_in=64, + variant='b', + lr_mult_list=[1.0, 1.0, 1.0, 1.0], + groups=1, + base_width=64, + norm_type='bn', + norm_decay=0, + freeze_norm=True, + freeze_at=0, + return_idx=[0, 1, 2, 3], + dcn_v2_stages=[-1], + num_stages=4, + std_senet=False): + """ + Residual Network, see https://arxiv.org/abs/1512.03385 + + Args: + depth (int): ResNet depth, should be 18, 34, 50, 101, 152. + ch_in (int): output channel of first stage, default 64 + variant (str): ResNet variant, supports 'a', 'b', 'c', 'd' currently + lr_mult_list (list): learning rate ratio of different resnet stages(2,3,4,5), + lower learning rate ratio is need for pretrained model + got using distillation(default as [1.0, 1.0, 1.0, 1.0]). + groups (int): group convolution cardinality + base_width (int): base width of each group convolution + norm_type (str): normalization type, 'bn', 'sync_bn' or 'affine_channel' + norm_decay (float): weight decay for normalization layer weights + freeze_norm (bool): freeze normalization layers + freeze_at (int): freeze the backbone at which stage + return_idx (list): index of the stages whose feature maps are returned + dcn_v2_stages (list): index of stages who select deformable conv v2 + num_stages (int): total num of stages + std_senet (bool): whether use senet, default True + """ + super(ResNet, self).__init__() + self._model_type = 'ResNet' if groups == 1 else 'ResNeXt' + assert num_stages >= 1 and num_stages <= 4 + self.depth = depth + self.variant = variant + self.groups = groups + self.base_width = base_width + self.norm_type = norm_type + self.norm_decay = norm_decay + self.freeze_norm = freeze_norm + self.freeze_at = freeze_at + if isinstance(return_idx, Integral): + return_idx = [return_idx] + assert max(return_idx) < num_stages, \ + 'the maximum return index must smaller than num_stages, ' \ + 'but received maximum return index is {} and num_stages ' \ + 'is {}'.format(max(return_idx), num_stages) + self.return_idx = return_idx + self.num_stages = num_stages + assert len(lr_mult_list) == 4, \ + "lr_mult_list length must be 4 but got {}".format(len(lr_mult_list)) + if isinstance(dcn_v2_stages, Integral): + dcn_v2_stages = [dcn_v2_stages] + assert max(dcn_v2_stages) < num_stages + + if isinstance(dcn_v2_stages, Integral): + dcn_v2_stages = [dcn_v2_stages] + assert max(dcn_v2_stages) < num_stages + self.dcn_v2_stages = dcn_v2_stages + + block_nums = ResNet_cfg[depth] + na = NameAdapter(self) + + conv1_name = na.fix_c1_stage_name() + if variant in ['c', 'd']: + conv_def = [ + [3, ch_in // 2, 3, 2, "conv1_1"], + [ch_in // 2, ch_in // 2, 3, 1, "conv1_2"], + [ch_in // 2, ch_in, 3, 1, "conv1_3"], + ] + else: + conv_def = [[3, ch_in, 7, 2, conv1_name]] + self.conv1 = nn.Sequential() + for (c_in, c_out, k, s, _name) in conv_def: + self.conv1.add_sublayer( + _name, + ConvNormLayer( + ch_in=c_in, + ch_out=c_out, + filter_size=k, + stride=s, + groups=1, + act='relu', + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + lr=1.0)) + + self.ch_in = ch_in + ch_out_list = [64, 128, 256, 512] + block = BottleNeck if depth >= 50 else BasicBlock + + self._out_channels = [block.expansion * v for v in ch_out_list] + self._out_strides = [4, 8, 16, 32] + + self.res_layers = [] + for i in range(num_stages): + lr_mult = lr_mult_list[i] + stage_num = i + 2 + res_name = "res{}".format(stage_num) + res_layer = self.add_sublayer( + res_name, + Blocks( + block, + self.ch_in, + ch_out_list[i], + count=block_nums[i], + name_adapter=na, + stage_num=stage_num, + variant=variant, + groups=groups, + base_width=base_width, + lr=lr_mult, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + dcn_v2=(i in self.dcn_v2_stages), + std_senet=std_senet)) + self.res_layers.append(res_layer) + self.ch_in = self._out_channels[i] + + if freeze_at >= 0: + self._freeze_parameters(self.conv1) + for i in range(min(freeze_at + 1, num_stages)): + self._freeze_parameters(self.res_layers[i]) + + def _freeze_parameters(self, m): + for p in m.parameters(): + p.stop_gradient = True + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self._out_channels[i], stride=self._out_strides[i]) + for i in self.return_idx + ] + + def forward(self, inputs): + x = inputs['image'] + conv1 = self.conv1(x) + x = F.max_pool2d(conv1, kernel_size=3, stride=2, padding=1) + outs = [] + for idx, stage in enumerate(self.res_layers): + x = stage(x) + if idx in self.return_idx: + outs.append(x) + return outs + + +@register +class Res5Head(nn.Layer): + def __init__(self, depth=50): + super(Res5Head, self).__init__() + feat_in, feat_out = [1024, 512] + if depth < 50: + feat_in = 256 + na = NameAdapter(self) + block = BottleNeck if depth >= 50 else BasicBlock + self.res5 = Blocks( + block, feat_in, feat_out, count=3, name_adapter=na, stage_num=5) + self.feat_out = feat_out if depth < 50 else feat_out * 4 + + @property + def out_shape(self): + return [ShapeSpec( + channels=self.feat_out, + stride=16, )] + + def forward(self, roi_feat, stage=0): + y = self.res5(roi_feat) + return y diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/senet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/senet.py new file mode 100644 index 000000000..eb0bad33f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/senet.py @@ -0,0 +1,139 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle.nn as nn + +from ppdet.core.workspace import register, serializable +from .resnet import ResNet, Blocks, BasicBlock, BottleNeck + +__all__ = ['SENet', 'SERes5Head'] + + +@register +@serializable +class SENet(ResNet): + __shared__ = ['norm_type'] + + def __init__(self, + depth=50, + variant='b', + lr_mult_list=[1.0, 1.0, 1.0, 1.0], + groups=1, + base_width=64, + norm_type='bn', + norm_decay=0, + freeze_norm=True, + freeze_at=0, + return_idx=[0, 1, 2, 3], + dcn_v2_stages=[-1], + std_senet=True, + num_stages=4): + """ + Squeeze-and-Excitation Networks, see https://arxiv.org/abs/1709.01507 + + Args: + depth (int): SENet depth, should be 50, 101, 152 + variant (str): ResNet variant, supports 'a', 'b', 'c', 'd' currently + lr_mult_list (list): learning rate ratio of different resnet stages(2,3,4,5), + lower learning rate ratio is need for pretrained model + got using distillation(default as [1.0, 1.0, 1.0, 1.0]). + groups (int): group convolution cardinality + base_width (int): base width of each group convolution + norm_type (str): normalization type, 'bn', 'sync_bn' or 'affine_channel' + norm_decay (float): weight decay for normalization layer weights + freeze_norm (bool): freeze normalization layers + freeze_at (int): freeze the backbone at which stage + return_idx (list): index of the stages whose feature maps are returned + dcn_v2_stages (list): index of stages who select deformable conv v2 + std_senet (bool): whether use senet, default True + num_stages (int): total num of stages + """ + + super(SENet, self).__init__( + depth=depth, + variant=variant, + lr_mult_list=lr_mult_list, + ch_in=128, + groups=groups, + base_width=base_width, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + freeze_at=freeze_at, + return_idx=return_idx, + dcn_v2_stages=dcn_v2_stages, + std_senet=std_senet, + num_stages=num_stages) + + +@register +class SERes5Head(nn.Layer): + def __init__(self, + depth=50, + variant='b', + lr_mult=1.0, + groups=1, + base_width=64, + norm_type='bn', + norm_decay=0, + dcn_v2=False, + freeze_norm=False, + std_senet=True): + """ + SERes5Head layer + + Args: + depth (int): SENet depth, should be 50, 101, 152 + variant (str): ResNet variant, supports 'a', 'b', 'c', 'd' currently + lr_mult (list): learning rate ratio of SERes5Head, default as 1.0. + groups (int): group convolution cardinality + base_width (int): base width of each group convolution + norm_type (str): normalization type, 'bn', 'sync_bn' or 'affine_channel' + norm_decay (float): weight decay for normalization layer weights + dcn_v2_stages (list): index of stages who select deformable conv v2 + std_senet (bool): whether use senet, default True + + """ + super(SERes5Head, self).__init__() + ch_out = 512 + ch_in = 256 if depth < 50 else 1024 + na = NameAdapter(self) + block = BottleNeck if depth >= 50 else BasicBlock + self.res5 = Blocks( + block, + ch_in, + ch_out, + count=3, + name_adapter=na, + stage_num=5, + variant=variant, + groups=groups, + base_width=base_width, + lr=lr_mult, + norm_type=norm_type, + norm_decay=norm_decay, + freeze_norm=freeze_norm, + dcn_v2=dcn_v2, + std_senet=std_senet) + self.ch_out = ch_out * block.expansion + + @property + def out_shape(self): + return [ShapeSpec( + channels=self.ch_out, + stride=16, )] + + def forward(self, roi_feat): + y = self.res5(roi_feat) + return y diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/shufflenet_v2.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/shufflenet_v2.py new file mode 100644 index 000000000..59b0502a1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/shufflenet_v2.py @@ -0,0 +1,246 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +from paddle import ParamAttr +from paddle.nn import Conv2D, MaxPool2D, AdaptiveAvgPool2D, BatchNorm +from paddle.nn.initializer import KaimingNormal +from paddle.regularizer import L2Decay + +from ppdet.core.workspace import register, serializable +from numbers import Integral +from ..shape_spec import ShapeSpec +from ppdet.modeling.ops import channel_shuffle + +__all__ = ['ShuffleNetV2'] + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size, + stride, + padding, + groups=1, + act=None): + super(ConvBNLayer, self).__init__() + self._conv = Conv2D( + in_channels=in_channels, + out_channels=out_channels, + kernel_size=kernel_size, + stride=stride, + padding=padding, + groups=groups, + weight_attr=ParamAttr(initializer=KaimingNormal()), + bias_attr=False) + + self._batch_norm = BatchNorm( + out_channels, + param_attr=ParamAttr(regularizer=L2Decay(0.0)), + bias_attr=ParamAttr(regularizer=L2Decay(0.0)), + act=act) + + def forward(self, inputs): + y = self._conv(inputs) + y = self._batch_norm(y) + return y + + +class InvertedResidual(nn.Layer): + def __init__(self, in_channels, out_channels, stride, act="relu"): + super(InvertedResidual, self).__init__() + self._conv_pw = ConvBNLayer( + in_channels=in_channels // 2, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + self._conv_dw = ConvBNLayer( + in_channels=out_channels // 2, + out_channels=out_channels // 2, + kernel_size=3, + stride=stride, + padding=1, + groups=out_channels // 2, + act=None) + self._conv_linear = ConvBNLayer( + in_channels=out_channels // 2, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + + def forward(self, inputs): + x1, x2 = paddle.split( + inputs, + num_or_sections=[inputs.shape[1] // 2, inputs.shape[1] // 2], + axis=1) + x2 = self._conv_pw(x2) + x2 = self._conv_dw(x2) + x2 = self._conv_linear(x2) + out = paddle.concat([x1, x2], axis=1) + return channel_shuffle(out, 2) + + +class InvertedResidualDS(nn.Layer): + def __init__(self, in_channels, out_channels, stride, act="relu"): + super(InvertedResidualDS, self).__init__() + + # branch1 + self._conv_dw_1 = ConvBNLayer( + in_channels=in_channels, + out_channels=in_channels, + kernel_size=3, + stride=stride, + padding=1, + groups=in_channels, + act=None) + self._conv_linear_1 = ConvBNLayer( + in_channels=in_channels, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + # branch2 + self._conv_pw_2 = ConvBNLayer( + in_channels=in_channels, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + self._conv_dw_2 = ConvBNLayer( + in_channels=out_channels // 2, + out_channels=out_channels // 2, + kernel_size=3, + stride=stride, + padding=1, + groups=out_channels // 2, + act=None) + self._conv_linear_2 = ConvBNLayer( + in_channels=out_channels // 2, + out_channels=out_channels // 2, + kernel_size=1, + stride=1, + padding=0, + groups=1, + act=act) + + def forward(self, inputs): + x1 = self._conv_dw_1(inputs) + x1 = self._conv_linear_1(x1) + x2 = self._conv_pw_2(inputs) + x2 = self._conv_dw_2(x2) + x2 = self._conv_linear_2(x2) + out = paddle.concat([x1, x2], axis=1) + + return channel_shuffle(out, 2) + + +@register +@serializable +class ShuffleNetV2(nn.Layer): + def __init__(self, scale=1.0, act="relu", feature_maps=[5, 13, 17]): + super(ShuffleNetV2, self).__init__() + self.scale = scale + if isinstance(feature_maps, Integral): + feature_maps = [feature_maps] + self.feature_maps = feature_maps + stage_repeats = [4, 8, 4] + + if scale == 0.25: + stage_out_channels = [-1, 24, 24, 48, 96, 512] + elif scale == 0.33: + stage_out_channels = [-1, 24, 32, 64, 128, 512] + elif scale == 0.5: + stage_out_channels = [-1, 24, 48, 96, 192, 1024] + elif scale == 1.0: + stage_out_channels = [-1, 24, 116, 232, 464, 1024] + elif scale == 1.5: + stage_out_channels = [-1, 24, 176, 352, 704, 1024] + elif scale == 2.0: + stage_out_channels = [-1, 24, 224, 488, 976, 2048] + else: + raise NotImplementedError("This scale size:[" + str(scale) + + "] is not implemented!") + + self._out_channels = [] + self._feature_idx = 0 + # 1. conv1 + self._conv1 = ConvBNLayer( + in_channels=3, + out_channels=stage_out_channels[1], + kernel_size=3, + stride=2, + padding=1, + act=act) + self._max_pool = MaxPool2D(kernel_size=3, stride=2, padding=1) + self._feature_idx += 1 + + # 2. bottleneck sequences + self._block_list = [] + for stage_id, num_repeat in enumerate(stage_repeats): + for i in range(num_repeat): + if i == 0: + block = self.add_sublayer( + name=str(stage_id + 2) + '_' + str(i + 1), + sublayer=InvertedResidualDS( + in_channels=stage_out_channels[stage_id + 1], + out_channels=stage_out_channels[stage_id + 2], + stride=2, + act=act)) + else: + block = self.add_sublayer( + name=str(stage_id + 2) + '_' + str(i + 1), + sublayer=InvertedResidual( + in_channels=stage_out_channels[stage_id + 2], + out_channels=stage_out_channels[stage_id + 2], + stride=1, + act=act)) + self._block_list.append(block) + self._feature_idx += 1 + self._update_out_channels(stage_out_channels[stage_id + 2], + self._feature_idx, self.feature_maps) + + def _update_out_channels(self, channel, feature_idx, feature_maps): + if feature_idx in feature_maps: + self._out_channels.append(channel) + + def forward(self, inputs): + y = self._conv1(inputs['image']) + y = self._max_pool(y) + outs = [] + for i, inv in enumerate(self._block_list): + y = inv(y) + if i + 2 in self.feature_maps: + outs.append(y) + + return outs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/swin_transformer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/swin_transformer.py new file mode 100644 index 000000000..027e4f67a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/swin_transformer.py @@ -0,0 +1,740 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/microsoft/Swin-Transformer/blob/main/models/swin_transformer.py +Ths copyright of microsoft/Swin-Transformer is as follows: +MIT License [see LICENSE for details] +""" + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import TruncatedNormal, Constant, Assign +from ppdet.modeling.shape_spec import ShapeSpec +from ppdet.core.workspace import register, serializable +import numpy as np + +# Common initializations +ones_ = Constant(value=1.) +zeros_ = Constant(value=0.) +trunc_normal_ = TruncatedNormal(std=.02) + + +# Common Functions +def to_2tuple(x): + return tuple([x] * 2) + + +def add_parameter(layer, datas, name=None): + parameter = layer.create_parameter( + shape=(datas.shape), default_initializer=Assign(datas)) + if name: + layer.add_parameter(name, parameter) + return parameter + + +# Common Layers +def drop_path(x, drop_prob=0., training=False): + """ + Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks). + the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper... + See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... + """ + if drop_prob == 0. or not training: + return x + keep_prob = paddle.to_tensor(1 - drop_prob) + shape = (paddle.shape(x)[0], ) + (1, ) * (x.ndim - 1) + random_tensor = keep_prob + paddle.rand(shape, dtype=x.dtype) + random_tensor = paddle.floor(random_tensor) # binarize + output = x.divide(keep_prob) * random_tensor + return output + + +class DropPath(nn.Layer): + def __init__(self, drop_prob=None): + super(DropPath, self).__init__() + self.drop_prob = drop_prob + + def forward(self, x): + return drop_path(x, self.drop_prob, self.training) + + +class Identity(nn.Layer): + def __init__(self): + super(Identity, self).__init__() + + def forward(self, input): + return input + + +class Mlp(nn.Layer): + def __init__(self, + in_features, + hidden_features=None, + out_features=None, + act_layer=nn.GELU, + drop=0.): + super().__init__() + out_features = out_features or in_features + hidden_features = hidden_features or in_features + self.fc1 = nn.Linear(in_features, hidden_features) + self.act = act_layer() + self.fc2 = nn.Linear(hidden_features, out_features) + self.drop = nn.Dropout(drop) + + def forward(self, x): + x = self.fc1(x) + x = self.act(x) + x = self.drop(x) + x = self.fc2(x) + x = self.drop(x) + return x + + +def window_partition(x, window_size): + """ + Args: + x: (B, H, W, C) + window_size (int): window size + Returns: + windows: (num_windows*B, window_size, window_size, C) + """ + B, H, W, C = x.shape + x = x.reshape( + [B, H // window_size, window_size, W // window_size, window_size, C]) + windows = x.transpose([0, 1, 3, 2, 4, 5]).reshape( + [-1, window_size, window_size, C]) + return windows + + +def window_reverse(windows, window_size, H, W): + """ + Args: + windows: (num_windows*B, window_size, window_size, C) + window_size (int): Window size + H (int): Height of image + W (int): Width of image + Returns: + x: (B, H, W, C) + """ + B = int(windows.shape[0] / (H * W / window_size / window_size)) + x = windows.reshape( + [B, H // window_size, W // window_size, window_size, window_size, -1]) + x = x.transpose([0, 1, 3, 2, 4, 5]).reshape([B, H, W, -1]) + return x + + +class WindowAttention(nn.Layer): + """ Window based multi-head self attention (W-MSA) module with relative position bias. + It supports both of shifted and non-shifted window. + + Args: + dim (int): Number of input channels. + window_size (tuple[int]): The height and width of the window. + num_heads (int): Number of attention heads. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set + attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0 + proj_drop (float, optional): Dropout ratio of output. Default: 0.0 + """ + + def __init__(self, + dim, + window_size, + num_heads, + qkv_bias=True, + qk_scale=None, + attn_drop=0., + proj_drop=0.): + + super().__init__() + self.dim = dim + self.window_size = window_size # Wh, Ww + self.num_heads = num_heads + head_dim = dim // num_heads + self.scale = qk_scale or head_dim**-0.5 + + # define a parameter table of relative position bias + self.relative_position_bias_table = add_parameter( + self, + paddle.zeros(((2 * window_size[0] - 1) * (2 * window_size[1] - 1), + num_heads))) # 2*Wh-1 * 2*Ww-1, nH + + # get pair-wise relative position index for each token inside the window + coords_h = paddle.arange(self.window_size[0]) + coords_w = paddle.arange(self.window_size[1]) + coords = paddle.stack(paddle.meshgrid( + [coords_h, coords_w])) # 2, Wh, Ww + coords_flatten = paddle.flatten(coords, 1) # 2, Wh*Ww + coords_flatten_1 = coords_flatten.unsqueeze(axis=2) + coords_flatten_2 = coords_flatten.unsqueeze(axis=1) + relative_coords = coords_flatten_1 - coords_flatten_2 + relative_coords = relative_coords.transpose( + [1, 2, 0]) # Wh*Ww, Wh*Ww, 2 + relative_coords[:, :, 0] += self.window_size[ + 0] - 1 # shift to start from 0 + relative_coords[:, :, 1] += self.window_size[1] - 1 + relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1 + self.relative_position_index = relative_coords.sum(-1) # Wh*Ww, Wh*Ww + self.register_buffer("relative_position_index", + self.relative_position_index) + + self.qkv = nn.Linear(dim, dim * 3, bias_attr=qkv_bias) + self.attn_drop = nn.Dropout(attn_drop) + self.proj = nn.Linear(dim, dim) + self.proj_drop = nn.Dropout(proj_drop) + + trunc_normal_(self.relative_position_bias_table) + self.softmax = nn.Softmax(axis=-1) + + def forward(self, x, mask=None): + """ Forward function. + Args: + x: input features with shape of (num_windows*B, N, C) + mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None + """ + B_, N, C = x.shape + qkv = self.qkv(x).reshape( + [B_, N, 3, self.num_heads, C // self.num_heads]).transpose( + [2, 0, 3, 1, 4]) + q, k, v = qkv[0], qkv[1], qkv[2] + + q = q * self.scale + attn = paddle.mm(q, k.transpose([0, 1, 3, 2])) + + index = self.relative_position_index.reshape([-1]) + + relative_position_bias = paddle.index_select( + self.relative_position_bias_table, index) + relative_position_bias = relative_position_bias.reshape([ + self.window_size[0] * self.window_size[1], + self.window_size[0] * self.window_size[1], -1 + ]) # Wh*Ww,Wh*Ww,nH + relative_position_bias = relative_position_bias.transpose( + [2, 0, 1]) # nH, Wh*Ww, Wh*Ww + attn = attn + relative_position_bias.unsqueeze(0) + + if mask is not None: + nW = mask.shape[0] + attn = attn.reshape([B_ // nW, nW, self.num_heads, N, N + ]) + mask.unsqueeze(1).unsqueeze(0) + attn = attn.reshape([-1, self.num_heads, N, N]) + attn = self.softmax(attn) + else: + attn = self.softmax(attn) + + attn = self.attn_drop(attn) + + # x = (attn @ v).transpose(1, 2).reshape([B_, N, C]) + x = paddle.mm(attn, v).transpose([0, 2, 1, 3]).reshape([B_, N, C]) + x = self.proj(x) + x = self.proj_drop(x) + return x + + +class SwinTransformerBlock(nn.Layer): + """ Swin Transformer Block. + Args: + dim (int): Number of input channels. + num_heads (int): Number of attention heads. + window_size (int): Window size. + shift_size (int): Shift size for SW-MSA. + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. + drop (float, optional): Dropout rate. Default: 0.0 + attn_drop (float, optional): Attention dropout rate. Default: 0.0 + drop_path (float, optional): Stochastic depth rate. Default: 0.0 + act_layer (nn.Layer, optional): Activation layer. Default: nn.GELU + norm_layer (nn.Layer, optional): Normalization layer. Default: nn.LayerNorm + """ + + def __init__(self, + dim, + num_heads, + window_size=7, + shift_size=0, + mlp_ratio=4., + qkv_bias=True, + qk_scale=None, + drop=0., + attn_drop=0., + drop_path=0., + act_layer=nn.GELU, + norm_layer=nn.LayerNorm): + super().__init__() + self.dim = dim + self.num_heads = num_heads + self.window_size = window_size + self.shift_size = shift_size + self.mlp_ratio = mlp_ratio + assert 0 <= self.shift_size < self.window_size, "shift_size must in 0-window_size" + + self.norm1 = norm_layer(dim) + self.attn = WindowAttention( + dim, + window_size=to_2tuple(self.window_size), + num_heads=num_heads, + qkv_bias=qkv_bias, + qk_scale=qk_scale, + attn_drop=attn_drop, + proj_drop=drop) + + self.drop_path = DropPath(drop_path) if drop_path > 0. else Identity() + self.norm2 = norm_layer(dim) + mlp_hidden_dim = int(dim * mlp_ratio) + self.mlp = Mlp(in_features=dim, + hidden_features=mlp_hidden_dim, + act_layer=act_layer, + drop=drop) + + self.H = None + self.W = None + + def forward(self, x, mask_matrix): + """ Forward function. + Args: + x: Input feature, tensor size (B, H*W, C). + H, W: Spatial resolution of the input feature. + mask_matrix: Attention mask for cyclic shift. + """ + B, L, C = x.shape + H, W = self.H, self.W + assert L == H * W, "input feature has wrong size" + + shortcut = x + x = self.norm1(x) + x = x.reshape([B, H, W, C]) + + # pad feature maps to multiples of window size + pad_l = pad_t = 0 + pad_r = (self.window_size - W % self.window_size) % self.window_size + pad_b = (self.window_size - H % self.window_size) % self.window_size + x = F.pad(x, [0, pad_l, 0, pad_b, 0, pad_r, 0, pad_t]) + _, Hp, Wp, _ = x.shape + + # cyclic shift + if self.shift_size > 0: + shifted_x = paddle.roll( + x, shifts=(-self.shift_size, -self.shift_size), axis=(1, 2)) + attn_mask = mask_matrix + else: + shifted_x = x + attn_mask = None + + # partition windows + x_windows = window_partition( + shifted_x, self.window_size) # nW*B, window_size, window_size, C + x_windows = x_windows.reshape( + [-1, self.window_size * self.window_size, + C]) # nW*B, window_size*window_size, C + + # W-MSA/SW-MSA + attn_windows = self.attn( + x_windows, mask=attn_mask) # nW*B, window_size*window_size, C + + # merge windows + attn_windows = attn_windows.reshape( + [-1, self.window_size, self.window_size, C]) + shifted_x = window_reverse(attn_windows, self.window_size, Hp, + Wp) # B H' W' C + + # reverse cyclic shift + if self.shift_size > 0: + x = paddle.roll( + shifted_x, + shifts=(self.shift_size, self.shift_size), + axis=(1, 2)) + else: + x = shifted_x + + if pad_r > 0 or pad_b > 0: + x = x[:, :H, :W, :] + + x = x.reshape([B, H * W, C]) + + # FFN + x = shortcut + self.drop_path(x) + x = x + self.drop_path(self.mlp(self.norm2(x))) + + return x + + +class PatchMerging(nn.Layer): + r""" Patch Merging Layer. + Args: + dim (int): Number of input channels. + norm_layer (nn.Layer, optional): Normalization layer. Default: nn.LayerNorm + """ + + def __init__(self, dim, norm_layer=nn.LayerNorm): + super().__init__() + self.dim = dim + self.reduction = nn.Linear(4 * dim, 2 * dim, bias_attr=False) + self.norm = norm_layer(4 * dim) + + def forward(self, x, H, W): + """ Forward function. + Args: + x: Input feature, tensor size (B, H*W, C). + H, W: Spatial resolution of the input feature. + """ + B, L, C = x.shape + assert L == H * W, "input feature has wrong size" + + x = x.reshape([B, H, W, C]) + + # padding + pad_input = (H % 2 == 1) or (W % 2 == 1) + if pad_input: + x = F.pad(x, [0, 0, 0, W % 2, 0, H % 2]) + + x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C + x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C + x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C + x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C + x = paddle.concat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C + x = x.reshape([B, H * W // 4, 4 * C]) # B H/2*W/2 4*C + + x = self.norm(x) + x = self.reduction(x) + + return x + + +class BasicLayer(nn.Layer): + """ A basic Swin Transformer layer for one stage. + Args: + dim (int): Number of input channels. + input_resolution (tuple[int]): Input resolution. + depth (int): Number of blocks. + num_heads (int): Number of attention heads. + window_size (int): Local window size. + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. + qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set. + drop (float, optional): Dropout rate. Default: 0.0 + attn_drop (float, optional): Attention dropout rate. Default: 0.0 + drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0 + norm_layer (nn.Layer, optional): Normalization layer. Default: nn.LayerNorm + downsample (nn.Layer | None, optional): Downsample layer at the end of the layer. Default: None + """ + + def __init__(self, + dim, + depth, + num_heads, + window_size=7, + mlp_ratio=4., + qkv_bias=True, + qk_scale=None, + drop=0., + attn_drop=0., + drop_path=0., + norm_layer=nn.LayerNorm, + downsample=None): + super().__init__() + self.window_size = window_size + self.shift_size = window_size // 2 + self.depth = depth + + # build blocks + self.blocks = nn.LayerList([ + SwinTransformerBlock( + dim=dim, + num_heads=num_heads, + window_size=window_size, + shift_size=0 if (i % 2 == 0) else window_size // 2, + mlp_ratio=mlp_ratio, + qkv_bias=qkv_bias, + qk_scale=qk_scale, + drop=drop, + attn_drop=attn_drop, + drop_path=drop_path[i] + if isinstance(drop_path, np.ndarray) else drop_path, + norm_layer=norm_layer) for i in range(depth) + ]) + + # patch merging layer + if downsample is not None: + self.downsample = downsample(dim=dim, norm_layer=norm_layer) + else: + self.downsample = None + + def forward(self, x, H, W): + """ Forward function. + Args: + x: Input feature, tensor size (B, H*W, C). + H, W: Spatial resolution of the input feature. + """ + + # calculate attention mask for SW-MSA + Hp = int(np.ceil(H / self.window_size)) * self.window_size + Wp = int(np.ceil(W / self.window_size)) * self.window_size + img_mask = paddle.fluid.layers.zeros( + [1, Hp, Wp, 1], dtype='float32') # 1 Hp Wp 1 + h_slices = (slice(0, -self.window_size), + slice(-self.window_size, -self.shift_size), + slice(-self.shift_size, None)) + w_slices = (slice(0, -self.window_size), + slice(-self.window_size, -self.shift_size), + slice(-self.shift_size, None)) + cnt = 0 + for h in h_slices: + for w in w_slices: + img_mask[:, h, w, :] = cnt + cnt += 1 + mask_windows = window_partition( + img_mask, self.window_size) # nW, window_size, window_size, 1 + mask_windows = mask_windows.reshape( + [-1, self.window_size * self.window_size]) + attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2) + huns = -100.0 * paddle.ones_like(attn_mask) + attn_mask = huns * (attn_mask != 0).astype("float32") + + for blk in self.blocks: + blk.H, blk.W = H, W + x = blk(x, attn_mask) + if self.downsample is not None: + x_down = self.downsample(x, H, W) + Wh, Ww = (H + 1) // 2, (W + 1) // 2 + return x, H, W, x_down, Wh, Ww + else: + return x, H, W, x, H, W + + +class PatchEmbed(nn.Layer): + """ Image to Patch Embedding + Args: + patch_size (int): Patch token size. Default: 4. + in_chans (int): Number of input image channels. Default: 3. + embed_dim (int): Number of linear projection output channels. Default: 96. + norm_layer (nn.Layer, optional): Normalization layer. Default: None + """ + + def __init__(self, patch_size=4, in_chans=3, embed_dim=96, norm_layer=None): + super().__init__() + patch_size = to_2tuple(patch_size) + self.patch_size = patch_size + + self.in_chans = in_chans + self.embed_dim = embed_dim + + self.proj = nn.Conv2D( + in_chans, embed_dim, kernel_size=patch_size, stride=patch_size) + if norm_layer is not None: + self.norm = norm_layer(embed_dim) + else: + self.norm = None + + def forward(self, x): + B, C, H, W = x.shape + # assert [H, W] == self.img_size[:2], "Input image size ({H}*{W}) doesn't match model ({}*{}).".format(H, W, self.img_size[0], self.img_size[1]) + if W % self.patch_size[1] != 0: + x = F.pad(x, [0, self.patch_size[1] - W % self.patch_size[1], 0, 0]) + if H % self.patch_size[0] != 0: + x = F.pad(x, [0, 0, 0, self.patch_size[0] - H % self.patch_size[0]]) + + x = self.proj(x) + if self.norm is not None: + _, _, Wh, Ww = x.shape + x = x.flatten(2).transpose([0, 2, 1]) + x = self.norm(x) + x = x.transpose([0, 2, 1]).reshape([-1, self.embed_dim, Wh, Ww]) + + return x + + +@register +@serializable +class SwinTransformer(nn.Layer): + """ Swin Transformer + A PaddlePaddle impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - + https://arxiv.org/pdf/2103.14030 + + Args: + img_size (int | tuple(int)): Input image size. Default 224 + patch_size (int | tuple(int)): Patch size. Default: 4 + in_chans (int): Number of input image channels. Default: 3 + num_classes (int): Number of classes for classification head. Default: 1000 + embed_dim (int): Patch embedding dimension. Default: 96 + depths (tuple(int)): Depth of each Swin Transformer layer. + num_heads (tuple(int)): Number of attention heads in different layers. + window_size (int): Window size. Default: 7 + mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4 + qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True + qk_scale (float): Override default qk scale of head_dim ** -0.5 if set. Default: None + drop_rate (float): Dropout rate. Default: 0 + attn_drop_rate (float): Attention dropout rate. Default: 0 + drop_path_rate (float): Stochastic depth rate. Default: 0.1 + norm_layer (nn.Layer): Normalization layer. Default: nn.LayerNorm. + ape (bool): If True, add absolute position embedding to the patch embedding. Default: False + patch_norm (bool): If True, add normalization after patch embedding. Default: True + """ + + def __init__(self, + pretrain_img_size=224, + patch_size=4, + in_chans=3, + embed_dim=96, + depths=[2, 2, 6, 2], + num_heads=[3, 6, 12, 24], + window_size=7, + mlp_ratio=4., + qkv_bias=True, + qk_scale=None, + drop_rate=0., + attn_drop_rate=0., + drop_path_rate=0.2, + norm_layer=nn.LayerNorm, + ape=False, + patch_norm=True, + out_indices=(0, 1, 2, 3), + frozen_stages=-1, + pretrained=None): + super(SwinTransformer, self).__init__() + + self.pretrain_img_size = pretrain_img_size + self.num_layers = len(depths) + self.embed_dim = embed_dim + self.ape = ape + self.patch_norm = patch_norm + self.out_indices = out_indices + self.frozen_stages = frozen_stages + + # split image into non-overlapping patches + self.patch_embed = PatchEmbed( + patch_size=patch_size, + in_chans=in_chans, + embed_dim=embed_dim, + norm_layer=norm_layer if self.patch_norm else None) + + # absolute position embedding + if self.ape: + pretrain_img_size = to_2tuple(pretrain_img_size) + patch_size = to_2tuple(patch_size) + patches_resolution = [ + pretrain_img_size[0] // patch_size[0], + pretrain_img_size[1] // patch_size[1] + ] + + self.absolute_pos_embed = add_parameter( + self, + paddle.zeros((1, embed_dim, patches_resolution[0], + patches_resolution[1]))) + trunc_normal_(self.absolute_pos_embed) + + self.pos_drop = nn.Dropout(p=drop_rate) + + # stochastic depth + dpr = np.linspace(0, drop_path_rate, + sum(depths)) # stochastic depth decay rule + + # build layers + self.layers = nn.LayerList() + for i_layer in range(self.num_layers): + layer = BasicLayer( + dim=int(embed_dim * 2**i_layer), + depth=depths[i_layer], + num_heads=num_heads[i_layer], + window_size=window_size, + mlp_ratio=mlp_ratio, + qkv_bias=qkv_bias, + qk_scale=qk_scale, + drop=drop_rate, + attn_drop=attn_drop_rate, + drop_path=dpr[sum(depths[:i_layer]):sum(depths[:i_layer + 1])], + norm_layer=norm_layer, + downsample=PatchMerging + if (i_layer < self.num_layers - 1) else None) + self.layers.append(layer) + + num_features = [int(embed_dim * 2**i) for i in range(self.num_layers)] + self.num_features = num_features + + # add a norm layer for each output + for i_layer in out_indices: + layer = norm_layer(num_features[i_layer]) + layer_name = f'norm{i_layer}' + self.add_sublayer(layer_name, layer) + + self.apply(self._init_weights) + self._freeze_stages() + if pretrained: + if 'http' in pretrained: #URL + path = paddle.utils.download.get_weights_path_from_url( + pretrained) + else: #model in local path + path = pretrained + self.set_state_dict(paddle.load(path)) + + def _freeze_stages(self): + if self.frozen_stages >= 0: + self.patch_embed.eval() + for param in self.patch_embed.parameters(): + param.requires_grad = False + + if self.frozen_stages >= 1 and self.ape: + self.absolute_pos_embed.requires_grad = False + + if self.frozen_stages >= 2: + self.pos_drop.eval() + for i in range(0, self.frozen_stages - 1): + m = self.layers[i] + m.eval() + for param in m.parameters(): + param.requires_grad = False + + def _init_weights(self, m): + if isinstance(m, nn.Linear): + trunc_normal_(m.weight) + if isinstance(m, nn.Linear) and m.bias is not None: + zeros_(m.bias) + elif isinstance(m, nn.LayerNorm): + zeros_(m.bias) + ones_(m.weight) + + def forward(self, x): + """Forward function.""" + x = self.patch_embed(x['image']) + _, _, Wh, Ww = x.shape + if self.ape: + # interpolate the position embedding to the corresponding size + absolute_pos_embed = F.interpolate( + self.absolute_pos_embed, size=(Wh, Ww), mode='bicubic') + x = (x + absolute_pos_embed).flatten(2).transpose([0, 2, 1]) + else: + x = x.flatten(2).transpose([0, 2, 1]) + x = self.pos_drop(x) + outs = [] + for i in range(self.num_layers): + layer = self.layers[i] + x_out, H, W, x, Wh, Ww = layer(x, Wh, Ww) + if i in self.out_indices: + norm_layer = getattr(self, f'norm{i}') + x_out = norm_layer(x_out) + out = x_out.reshape((-1, H, W, self.num_features[i])).transpose( + (0, 3, 1, 2)) + outs.append(out) + + return tuple(outs) + + @property + def out_shape(self): + out_strides = [4, 8, 16, 32] + return [ + ShapeSpec( + channels=self.num_features[i], stride=out_strides[i]) + for i in self.out_indices + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/vgg.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/vgg.py new file mode 100644 index 000000000..e05753209 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/backbones/vgg.py @@ -0,0 +1,210 @@ +from __future__ import division + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn import Conv2D, MaxPool2D +from ppdet.core.workspace import register, serializable +from ..shape_spec import ShapeSpec + +__all__ = ['VGG'] + +VGG_cfg = {16: [2, 2, 3, 3, 3], 19: [2, 2, 4, 4, 4]} + + +class ConvBlock(nn.Layer): + def __init__(self, + in_channels, + out_channels, + groups, + pool_size=2, + pool_stride=2, + pool_padding=0, + name=None): + super(ConvBlock, self).__init__() + + self.groups = groups + self.conv0 = nn.Conv2D( + in_channels=in_channels, + out_channels=out_channels, + kernel_size=3, + stride=1, + padding=1) + self.conv_out_list = [] + for i in range(1, groups): + conv_out = self.add_sublayer( + 'conv{}'.format(i), + Conv2D( + in_channels=out_channels, + out_channels=out_channels, + kernel_size=3, + stride=1, + padding=1)) + self.conv_out_list.append(conv_out) + + self.pool = MaxPool2D( + kernel_size=pool_size, + stride=pool_stride, + padding=pool_padding, + ceil_mode=True) + + def forward(self, inputs): + out = self.conv0(inputs) + out = F.relu(out) + for conv_i in self.conv_out_list: + out = conv_i(out) + out = F.relu(out) + pool = self.pool(out) + return out, pool + + +class ExtraBlock(nn.Layer): + def __init__(self, + in_channels, + mid_channels, + out_channels, + padding, + stride, + kernel_size, + name=None): + super(ExtraBlock, self).__init__() + + self.conv0 = Conv2D( + in_channels=in_channels, + out_channels=mid_channels, + kernel_size=1, + stride=1, + padding=0) + self.conv1 = Conv2D( + in_channels=mid_channels, + out_channels=out_channels, + kernel_size=kernel_size, + stride=stride, + padding=padding) + + def forward(self, inputs): + out = self.conv0(inputs) + out = F.relu(out) + out = self.conv1(out) + out = F.relu(out) + return out + + +class L2NormScale(nn.Layer): + def __init__(self, num_channels, scale=1.0): + super(L2NormScale, self).__init__() + self.scale = self.create_parameter( + attr=ParamAttr(initializer=paddle.nn.initializer.Constant(scale)), + shape=[num_channels]) + + def forward(self, inputs): + out = F.normalize(inputs, axis=1, epsilon=1e-10) + # out = self.scale.unsqueeze(0).unsqueeze(2).unsqueeze(3).expand_as( + # out) * out + out = self.scale.unsqueeze(0).unsqueeze(2).unsqueeze(3) * out + return out + + +@register +@serializable +class VGG(nn.Layer): + def __init__(self, + depth=16, + normalizations=[20., -1, -1, -1, -1, -1], + extra_block_filters=[[256, 512, 1, 2, 3], [128, 256, 1, 2, 3], + [128, 256, 0, 1, 3], + [128, 256, 0, 1, 3]]): + super(VGG, self).__init__() + + assert depth in [16, 19], \ + "depth as 16/19 supported currently, but got {}".format(depth) + self.depth = depth + self.groups = VGG_cfg[depth] + self.normalizations = normalizations + self.extra_block_filters = extra_block_filters + + self._out_channels = [] + + self.conv_block_0 = ConvBlock( + 3, 64, self.groups[0], 2, 2, 0, name="conv1_") + self.conv_block_1 = ConvBlock( + 64, 128, self.groups[1], 2, 2, 0, name="conv2_") + self.conv_block_2 = ConvBlock( + 128, 256, self.groups[2], 2, 2, 0, name="conv3_") + self.conv_block_3 = ConvBlock( + 256, 512, self.groups[3], 2, 2, 0, name="conv4_") + self.conv_block_4 = ConvBlock( + 512, 512, self.groups[4], 3, 1, 1, name="conv5_") + self._out_channels.append(512) + + self.fc6 = Conv2D( + in_channels=512, + out_channels=1024, + kernel_size=3, + stride=1, + padding=6, + dilation=6) + self.fc7 = Conv2D( + in_channels=1024, + out_channels=1024, + kernel_size=1, + stride=1, + padding=0) + self._out_channels.append(1024) + + # extra block + self.extra_convs = [] + last_channels = 1024 + for i, v in enumerate(self.extra_block_filters): + assert len(v) == 5, "extra_block_filters size not fix" + extra_conv = self.add_sublayer("conv{}".format(6 + i), + ExtraBlock(last_channels, v[0], v[1], + v[2], v[3], v[4])) + last_channels = v[1] + self.extra_convs.append(extra_conv) + self._out_channels.append(last_channels) + + self.norms = [] + for i, n in enumerate(self.normalizations): + if n != -1: + norm = self.add_sublayer("norm{}".format(i), + L2NormScale( + self.extra_block_filters[i][1], n)) + else: + norm = None + self.norms.append(norm) + + def forward(self, inputs): + outputs = [] + + conv, pool = self.conv_block_0(inputs['image']) + conv, pool = self.conv_block_1(pool) + conv, pool = self.conv_block_2(pool) + conv, pool = self.conv_block_3(pool) + outputs.append(conv) + + conv, pool = self.conv_block_4(pool) + out = self.fc6(pool) + out = F.relu(out) + out = self.fc7(out) + out = F.relu(out) + outputs.append(out) + + if not self.extra_block_filters: + return outputs + + # extra block + for extra_conv in self.extra_convs: + out = extra_conv(out) + outputs.append(out) + + for i, n in enumerate(self.normalizations): + if n != -1: + outputs[i] = self.norms[i](outputs[i]) + + return outputs + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/bbox_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/bbox_utils.py new file mode 100644 index 000000000..e040ba69b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/bbox_utils.py @@ -0,0 +1,753 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import math +import paddle +import numpy as np + + +def bbox2delta(src_boxes, tgt_boxes, weights): + src_w = src_boxes[:, 2] - src_boxes[:, 0] + src_h = src_boxes[:, 3] - src_boxes[:, 1] + src_ctr_x = src_boxes[:, 0] + 0.5 * src_w + src_ctr_y = src_boxes[:, 1] + 0.5 * src_h + + tgt_w = tgt_boxes[:, 2] - tgt_boxes[:, 0] + tgt_h = tgt_boxes[:, 3] - tgt_boxes[:, 1] + tgt_ctr_x = tgt_boxes[:, 0] + 0.5 * tgt_w + tgt_ctr_y = tgt_boxes[:, 1] + 0.5 * tgt_h + + wx, wy, ww, wh = weights + dx = wx * (tgt_ctr_x - src_ctr_x) / src_w + dy = wy * (tgt_ctr_y - src_ctr_y) / src_h + dw = ww * paddle.log(tgt_w / src_w) + dh = wh * paddle.log(tgt_h / src_h) + + deltas = paddle.stack((dx, dy, dw, dh), axis=1) + return deltas + + +def delta2bbox(deltas, boxes, weights): + clip_scale = math.log(1000.0 / 16) + + widths = boxes[:, 2] - boxes[:, 0] + heights = boxes[:, 3] - boxes[:, 1] + ctr_x = boxes[:, 0] + 0.5 * widths + ctr_y = boxes[:, 1] + 0.5 * heights + + wx, wy, ww, wh = weights + dx = deltas[:, 0::4] / wx + dy = deltas[:, 1::4] / wy + dw = deltas[:, 2::4] / ww + dh = deltas[:, 3::4] / wh + # Prevent sending too large values into paddle.exp() + dw = paddle.clip(dw, max=clip_scale) + dh = paddle.clip(dh, max=clip_scale) + + pred_ctr_x = dx * widths.unsqueeze(1) + ctr_x.unsqueeze(1) + pred_ctr_y = dy * heights.unsqueeze(1) + ctr_y.unsqueeze(1) + pred_w = paddle.exp(dw) * widths.unsqueeze(1) + pred_h = paddle.exp(dh) * heights.unsqueeze(1) + + pred_boxes = [] + pred_boxes.append(pred_ctr_x - 0.5 * pred_w) + pred_boxes.append(pred_ctr_y - 0.5 * pred_h) + pred_boxes.append(pred_ctr_x + 0.5 * pred_w) + pred_boxes.append(pred_ctr_y + 0.5 * pred_h) + pred_boxes = paddle.stack(pred_boxes, axis=-1) + + return pred_boxes + + +def expand_bbox(bboxes, scale): + w_half = (bboxes[:, 2] - bboxes[:, 0]) * .5 + h_half = (bboxes[:, 3] - bboxes[:, 1]) * .5 + x_c = (bboxes[:, 2] + bboxes[:, 0]) * .5 + y_c = (bboxes[:, 3] + bboxes[:, 1]) * .5 + + w_half *= scale + h_half *= scale + + bboxes_exp = np.zeros(bboxes.shape, dtype=np.float32) + bboxes_exp[:, 0] = x_c - w_half + bboxes_exp[:, 2] = x_c + w_half + bboxes_exp[:, 1] = y_c - h_half + bboxes_exp[:, 3] = y_c + h_half + + return bboxes_exp + + +def clip_bbox(boxes, im_shape): + h, w = im_shape[0], im_shape[1] + x1 = boxes[:, 0].clip(0, w) + y1 = boxes[:, 1].clip(0, h) + x2 = boxes[:, 2].clip(0, w) + y2 = boxes[:, 3].clip(0, h) + return paddle.stack([x1, y1, x2, y2], axis=1) + + +def nonempty_bbox(boxes, min_size=0, return_mask=False): + w = boxes[:, 2] - boxes[:, 0] + h = boxes[:, 3] - boxes[:, 1] + mask = paddle.logical_and(h > min_size, w > min_size) + if return_mask: + return mask + keep = paddle.nonzero(mask).flatten() + return keep + + +def bbox_area(boxes): + return (boxes[:, 2] - boxes[:, 0]) * (boxes[:, 3] - boxes[:, 1]) + + +def bbox_overlaps(boxes1, boxes2): + """ + Calculate overlaps between boxes1 and boxes2 + + Args: + boxes1 (Tensor): boxes with shape [M, 4] + boxes2 (Tensor): boxes with shape [N, 4] + + Return: + overlaps (Tensor): overlaps between boxes1 and boxes2 with shape [M, N] + """ + M = boxes1.shape[0] + N = boxes2.shape[0] + if M * N == 0: + return paddle.zeros([M, N], dtype='float32') + area1 = bbox_area(boxes1) + area2 = bbox_area(boxes2) + + xy_max = paddle.minimum( + paddle.unsqueeze(boxes1, 1)[:, :, 2:], boxes2[:, 2:]) + xy_min = paddle.maximum( + paddle.unsqueeze(boxes1, 1)[:, :, :2], boxes2[:, :2]) + width_height = xy_max - xy_min + width_height = width_height.clip(min=0) + inter = width_height.prod(axis=2) + + overlaps = paddle.where(inter > 0, inter / + (paddle.unsqueeze(area1, 1) + area2 - inter), + paddle.zeros_like(inter)) + return overlaps + + +def batch_bbox_overlaps(bboxes1, + bboxes2, + mode='iou', + is_aligned=False, + eps=1e-6): + """Calculate overlap between two set of bboxes. + If ``is_aligned `` is ``False``, then calculate the overlaps between each + bbox of bboxes1 and bboxes2, otherwise the overlaps between each aligned + pair of bboxes1 and bboxes2. + Args: + bboxes1 (Tensor): shape (B, m, 4) in format or empty. + bboxes2 (Tensor): shape (B, n, 4) in format or empty. + B indicates the batch dim, in shape (B1, B2, ..., Bn). + If ``is_aligned `` is ``True``, then m and n must be equal. + mode (str): "iou" (intersection over union) or "iof" (intersection over + foreground). + is_aligned (bool, optional): If True, then m and n must be equal. + Default False. + eps (float, optional): A value added to the denominator for numerical + stability. Default 1e-6. + Returns: + Tensor: shape (m, n) if ``is_aligned `` is False else shape (m,) + """ + assert mode in ['iou', 'iof', 'giou'], 'Unsupported mode {}'.format(mode) + # Either the boxes are empty or the length of boxes's last dimenstion is 4 + assert (bboxes1.shape[-1] == 4 or bboxes1.shape[0] == 0) + assert (bboxes2.shape[-1] == 4 or bboxes2.shape[0] == 0) + + # Batch dim must be the same + # Batch dim: (B1, B2, ... Bn) + assert bboxes1.shape[:-2] == bboxes2.shape[:-2] + batch_shape = bboxes1.shape[:-2] + + rows = bboxes1.shape[-2] if bboxes1.shape[0] > 0 else 0 + cols = bboxes2.shape[-2] if bboxes2.shape[0] > 0 else 0 + if is_aligned: + assert rows == cols + + if rows * cols == 0: + if is_aligned: + return paddle.full(batch_shape + (rows, ), 1) + else: + return paddle.full(batch_shape + (rows, cols), 1) + + area1 = (bboxes1[:, 2] - bboxes1[:, 0]) * (bboxes1[:, 3] - bboxes1[:, 1]) + area2 = (bboxes2[:, 2] - bboxes2[:, 0]) * (bboxes2[:, 3] - bboxes2[:, 1]) + + if is_aligned: + lt = paddle.maximum(bboxes1[:, :2], bboxes2[:, :2]) # [B, rows, 2] + rb = paddle.minimum(bboxes1[:, 2:], bboxes2[:, 2:]) # [B, rows, 2] + + wh = (rb - lt).clip(min=0) # [B, rows, 2] + overlap = wh[:, 0] * wh[:, 1] + + if mode in ['iou', 'giou']: + union = area1 + area2 - overlap + else: + union = area1 + if mode == 'giou': + enclosed_lt = paddle.minimum(bboxes1[:, :2], bboxes2[:, :2]) + enclosed_rb = paddle.maximum(bboxes1[:, 2:], bboxes2[:, 2:]) + else: + lt = paddle.maximum(bboxes1[:, :2].reshape([rows, 1, 2]), + bboxes2[:, :2]) # [B, rows, cols, 2] + rb = paddle.minimum(bboxes1[:, 2:].reshape([rows, 1, 2]), + bboxes2[:, 2:]) # [B, rows, cols, 2] + + wh = (rb - lt).clip(min=0) # [B, rows, cols, 2] + overlap = wh[:, :, 0] * wh[:, :, 1] + + if mode in ['iou', 'giou']: + union = area1.reshape([rows,1]) \ + + area2.reshape([1,cols]) - overlap + else: + union = area1[:, None] + if mode == 'giou': + enclosed_lt = paddle.minimum(bboxes1[:, :2].reshape([rows, 1, 2]), + bboxes2[:, :2]) + enclosed_rb = paddle.maximum(bboxes1[:, 2:].reshape([rows, 1, 2]), + bboxes2[:, 2:]) + + eps = paddle.to_tensor([eps]) + union = paddle.maximum(union, eps) + ious = overlap / union + if mode in ['iou', 'iof']: + return ious + # calculate gious + enclose_wh = (enclosed_rb - enclosed_lt).clip(min=0) + enclose_area = enclose_wh[:, :, 0] * enclose_wh[:, :, 1] + enclose_area = paddle.maximum(enclose_area, eps) + gious = ious - (enclose_area - union) / enclose_area + return 1 - gious + + +def xywh2xyxy(box): + x, y, w, h = box + x1 = x - w * 0.5 + y1 = y - h * 0.5 + x2 = x + w * 0.5 + y2 = y + h * 0.5 + return [x1, y1, x2, y2] + + +def make_grid(h, w, dtype): + yv, xv = paddle.meshgrid([paddle.arange(h), paddle.arange(w)]) + return paddle.stack((xv, yv), 2).cast(dtype=dtype) + + +def decode_yolo(box, anchor, downsample_ratio): + """decode yolo box + + Args: + box (list): [x, y, w, h], all have the shape [b, na, h, w, 1] + anchor (list): anchor with the shape [na, 2] + downsample_ratio (int): downsample ratio, default 32 + scale (float): scale, default 1. + + Return: + box (list): decoded box, [x, y, w, h], all have the shape [b, na, h, w, 1] + """ + x, y, w, h = box + na, grid_h, grid_w = x.shape[1:4] + grid = make_grid(grid_h, grid_w, x.dtype).reshape((1, 1, grid_h, grid_w, 2)) + x1 = (x + grid[:, :, :, :, 0:1]) / grid_w + y1 = (y + grid[:, :, :, :, 1:2]) / grid_h + + anchor = paddle.to_tensor(anchor) + anchor = paddle.cast(anchor, x.dtype) + anchor = anchor.reshape((1, na, 1, 1, 2)) + w1 = paddle.exp(w) * anchor[:, :, :, :, 0:1] / (downsample_ratio * grid_w) + h1 = paddle.exp(h) * anchor[:, :, :, :, 1:2] / (downsample_ratio * grid_h) + + return [x1, y1, w1, h1] + + +def iou_similarity(box1, box2, eps=1e-9): + """Calculate iou of box1 and box2 + + Args: + box1 (Tensor): box with the shape [N, M1, 4] + box2 (Tensor): box with the shape [N, M2, 4] + + Return: + iou (Tensor): iou between box1 and box2 with the shape [N, M1, M2] + """ + box1 = box1.unsqueeze(2) # [N, M1, 4] -> [N, M1, 1, 4] + box2 = box2.unsqueeze(1) # [N, M2, 4] -> [N, 1, M2, 4] + px1y1, px2y2 = box1[:, :, :, 0:2], box1[:, :, :, 2:4] + gx1y1, gx2y2 = box2[:, :, :, 0:2], box2[:, :, :, 2:4] + x1y1 = paddle.maximum(px1y1, gx1y1) + x2y2 = paddle.minimum(px2y2, gx2y2) + overlap = (x2y2 - x1y1).clip(0).prod(-1) + area1 = (px2y2 - px1y1).clip(0).prod(-1) + area2 = (gx2y2 - gx1y1).clip(0).prod(-1) + union = area1 + area2 - overlap + eps + return overlap / union + + +def bbox_iou(box1, box2, giou=False, diou=False, ciou=False, eps=1e-9): + """calculate the iou of box1 and box2 + + Args: + box1 (list): [x, y, w, h], all have the shape [b, na, h, w, 1] + box2 (list): [x, y, w, h], all have the shape [b, na, h, w, 1] + giou (bool): whether use giou or not, default False + diou (bool): whether use diou or not, default False + ciou (bool): whether use ciou or not, default False + eps (float): epsilon to avoid divide by zero + + Return: + iou (Tensor): iou of box1 and box1, with the shape [b, na, h, w, 1] + """ + px1, py1, px2, py2 = box1 + gx1, gy1, gx2, gy2 = box2 + x1 = paddle.maximum(px1, gx1) + y1 = paddle.maximum(py1, gy1) + x2 = paddle.minimum(px2, gx2) + y2 = paddle.minimum(py2, gy2) + + overlap = ((x2 - x1).clip(0)) * ((y2 - y1).clip(0)) + + area1 = (px2 - px1) * (py2 - py1) + area1 = area1.clip(0) + + area2 = (gx2 - gx1) * (gy2 - gy1) + area2 = area2.clip(0) + + union = area1 + area2 - overlap + eps + iou = overlap / union + + if giou or ciou or diou: + # convex w, h + cw = paddle.maximum(px2, gx2) - paddle.minimum(px1, gx1) + ch = paddle.maximum(py2, gy2) - paddle.minimum(py1, gy1) + if giou: + c_area = cw * ch + eps + return iou - (c_area - union) / c_area + else: + # convex diagonal squared + c2 = cw**2 + ch**2 + eps + # center distance + rho2 = ((px1 + px2 - gx1 - gx2)**2 + (py1 + py2 - gy1 - gy2)**2) / 4 + if diou: + return iou - rho2 / c2 + else: + w1, h1 = px2 - px1, py2 - py1 + eps + w2, h2 = gx2 - gx1, gy2 - gy1 + eps + delta = paddle.atan(w1 / h1) - paddle.atan(w2 / h2) + v = (4 / math.pi**2) * paddle.pow(delta, 2) + alpha = v / (1 + eps - iou + v) + alpha.stop_gradient = True + return iou - (rho2 / c2 + v * alpha) + else: + return iou + + +def rect2rbox(bboxes): + """ + :param bboxes: shape (n, 4) (xmin, ymin, xmax, ymax) + :return: dbboxes: shape (n, 5) (x_ctr, y_ctr, w, h, angle) + """ + bboxes = bboxes.reshape(-1, 4) + num_boxes = bboxes.shape[0] + + x_ctr = (bboxes[:, 2] + bboxes[:, 0]) / 2.0 + y_ctr = (bboxes[:, 3] + bboxes[:, 1]) / 2.0 + edges1 = np.abs(bboxes[:, 2] - bboxes[:, 0]) + edges2 = np.abs(bboxes[:, 3] - bboxes[:, 1]) + angles = np.zeros([num_boxes], dtype=bboxes.dtype) + + inds = edges1 < edges2 + + rboxes = np.stack((x_ctr, y_ctr, edges1, edges2, angles), axis=1) + rboxes[inds, 2] = edges2[inds] + rboxes[inds, 3] = edges1[inds] + rboxes[inds, 4] = np.pi / 2.0 + return rboxes + + +def delta2rbox(rrois, + deltas, + means=[0, 0, 0, 0, 0], + stds=[1, 1, 1, 1, 1], + wh_ratio_clip=1e-6): + """ + :param rrois: (cx, cy, w, h, theta) + :param deltas: (dx, dy, dw, dh, dtheta) + :param means: + :param stds: + :param wh_ratio_clip: + :return: + """ + means = paddle.to_tensor(means) + stds = paddle.to_tensor(stds) + deltas = paddle.reshape(deltas, [-1, deltas.shape[-1]]) + denorm_deltas = deltas * stds + means + + dx = denorm_deltas[:, 0] + dy = denorm_deltas[:, 1] + dw = denorm_deltas[:, 2] + dh = denorm_deltas[:, 3] + dangle = denorm_deltas[:, 4] + + max_ratio = np.abs(np.log(wh_ratio_clip)) + dw = paddle.clip(dw, min=-max_ratio, max=max_ratio) + dh = paddle.clip(dh, min=-max_ratio, max=max_ratio) + + rroi_x = rrois[:, 0] + rroi_y = rrois[:, 1] + rroi_w = rrois[:, 2] + rroi_h = rrois[:, 3] + rroi_angle = rrois[:, 4] + + gx = dx * rroi_w * paddle.cos(rroi_angle) - dy * rroi_h * paddle.sin( + rroi_angle) + rroi_x + gy = dx * rroi_w * paddle.sin(rroi_angle) + dy * rroi_h * paddle.cos( + rroi_angle) + rroi_y + gw = rroi_w * dw.exp() + gh = rroi_h * dh.exp() + ga = np.pi * dangle + rroi_angle + ga = (ga + np.pi / 4) % np.pi - np.pi / 4 + ga = paddle.to_tensor(ga) + + gw = paddle.to_tensor(gw, dtype='float32') + gh = paddle.to_tensor(gh, dtype='float32') + bboxes = paddle.stack([gx, gy, gw, gh, ga], axis=-1) + return bboxes + + +def rbox2delta(proposals, gt, means=[0, 0, 0, 0, 0], stds=[1, 1, 1, 1, 1]): + """ + + Args: + proposals: + gt: + means: 1x5 + stds: 1x5 + + Returns: + + """ + proposals = proposals.astype(np.float64) + + PI = np.pi + + gt_widths = gt[..., 2] + gt_heights = gt[..., 3] + gt_angle = gt[..., 4] + + proposals_widths = proposals[..., 2] + proposals_heights = proposals[..., 3] + proposals_angle = proposals[..., 4] + + coord = gt[..., 0:2] - proposals[..., 0:2] + dx = (np.cos(proposals[..., 4]) * coord[..., 0] + np.sin(proposals[..., 4]) + * coord[..., 1]) / proposals_widths + dy = (-np.sin(proposals[..., 4]) * coord[..., 0] + np.cos(proposals[..., 4]) + * coord[..., 1]) / proposals_heights + dw = np.log(gt_widths / proposals_widths) + dh = np.log(gt_heights / proposals_heights) + da = (gt_angle - proposals_angle) + + da = (da + PI / 4) % PI - PI / 4 + da /= PI + + deltas = np.stack([dx, dy, dw, dh, da], axis=-1) + means = np.array(means, dtype=deltas.dtype) + stds = np.array(stds, dtype=deltas.dtype) + deltas = (deltas - means) / stds + deltas = deltas.astype(np.float32) + return deltas + + +def bbox_decode(bbox_preds, + anchors, + means=[0, 0, 0, 0, 0], + stds=[1, 1, 1, 1, 1]): + """decode bbox from deltas + Args: + bbox_preds: [N,H,W,5] + anchors: [H*W,5] + return: + bboxes: [N,H,W,5] + """ + means = paddle.to_tensor(means) + stds = paddle.to_tensor(stds) + num_imgs, H, W, _ = bbox_preds.shape + bboxes_list = [] + for img_id in range(num_imgs): + bbox_pred = bbox_preds[img_id] + # bbox_pred.shape=[5,H,W] + bbox_delta = bbox_pred + anchors = paddle.to_tensor(anchors) + bboxes = delta2rbox( + anchors, bbox_delta, means, stds, wh_ratio_clip=1e-6) + bboxes = paddle.reshape(bboxes, [H, W, 5]) + bboxes_list.append(bboxes) + return paddle.stack(bboxes_list, axis=0) + + +def poly2rbox(polys): + """ + poly:[x0,y0,x1,y1,x2,y2,x3,y3] + to + rotated_boxes:[x_ctr,y_ctr,w,h,angle] + """ + rotated_boxes = [] + for poly in polys: + poly = np.array(poly[:8], dtype=np.float32) + + pt1 = (poly[0], poly[1]) + pt2 = (poly[2], poly[3]) + pt3 = (poly[4], poly[5]) + pt4 = (poly[6], poly[7]) + + edge1 = np.sqrt((pt1[0] - pt2[0]) * (pt1[0] - pt2[0]) + (pt1[1] - pt2[ + 1]) * (pt1[1] - pt2[1])) + edge2 = np.sqrt((pt2[0] - pt3[0]) * (pt2[0] - pt3[0]) + (pt2[1] - pt3[ + 1]) * (pt2[1] - pt3[1])) + + width = max(edge1, edge2) + height = min(edge1, edge2) + + rbox_angle = 0 + if edge1 > edge2: + rbox_angle = np.arctan2( + float(pt2[1] - pt1[1]), float(pt2[0] - pt1[0])) + elif edge2 >= edge1: + rbox_angle = np.arctan2( + float(pt4[1] - pt1[1]), float(pt4[0] - pt1[0])) + + def norm_angle(angle, range=[-np.pi / 4, np.pi]): + return (angle - range[0]) % range[1] + range[0] + + rbox_angle = norm_angle(rbox_angle) + + x_ctr = float(pt1[0] + pt3[0]) / 2 + y_ctr = float(pt1[1] + pt3[1]) / 2 + rotated_box = np.array([x_ctr, y_ctr, width, height, rbox_angle]) + rotated_boxes.append(rotated_box) + ret_rotated_boxes = np.array(rotated_boxes) + assert ret_rotated_boxes.shape[1] == 5 + return ret_rotated_boxes + + +def cal_line_length(point1, point2): + import math + return math.sqrt( + math.pow(point1[0] - point2[0], 2) + math.pow(point1[1] - point2[1], 2)) + + +def get_best_begin_point_single(coordinate): + x1, y1, x2, y2, x3, y3, x4, y4 = coordinate + xmin = min(x1, x2, x3, x4) + ymin = min(y1, y2, y3, y4) + xmax = max(x1, x2, x3, x4) + ymax = max(y1, y2, y3, y4) + combinate = [[[x1, y1], [x2, y2], [x3, y3], [x4, y4]], + [[x4, y4], [x1, y1], [x2, y2], [x3, y3]], + [[x3, y3], [x4, y4], [x1, y1], [x2, y2]], + [[x2, y2], [x3, y3], [x4, y4], [x1, y1]]] + dst_coordinate = [[xmin, ymin], [xmax, ymin], [xmax, ymax], [xmin, ymax]] + force = 100000000.0 + force_flag = 0 + for i in range(4): + temp_force = cal_line_length(combinate[i][0], dst_coordinate[0]) \ + + cal_line_length(combinate[i][1], dst_coordinate[1]) \ + + cal_line_length(combinate[i][2], dst_coordinate[2]) \ + + cal_line_length(combinate[i][3], dst_coordinate[3]) + if temp_force < force: + force = temp_force + force_flag = i + if force_flag != 0: + pass + return np.array(combinate[force_flag]).reshape(8) + + +def rbox2poly_np(rrects): + """ + rrect:[x_ctr,y_ctr,w,h,angle] + to + poly:[x0,y0,x1,y1,x2,y2,x3,y3] + """ + polys = [] + for i in range(rrects.shape[0]): + rrect = rrects[i] + # x_ctr, y_ctr, width, height, angle = rrect[:5] + x_ctr = rrect[0] + y_ctr = rrect[1] + width = rrect[2] + height = rrect[3] + angle = rrect[4] + tl_x, tl_y, br_x, br_y = -width / 2, -height / 2, width / 2, height / 2 + rect = np.array([[tl_x, br_x, br_x, tl_x], [tl_y, tl_y, br_y, br_y]]) + R = np.array([[np.cos(angle), -np.sin(angle)], + [np.sin(angle), np.cos(angle)]]) + poly = R.dot(rect) + x0, x1, x2, x3 = poly[0, :4] + x_ctr + y0, y1, y2, y3 = poly[1, :4] + y_ctr + poly = np.array([x0, y0, x1, y1, x2, y2, x3, y3], dtype=np.float32) + poly = get_best_begin_point_single(poly) + polys.append(poly) + polys = np.array(polys) + return polys + + +def rbox2poly(rrects): + """ + rrect:[x_ctr,y_ctr,w,h,angle] + to + poly:[x0,y0,x1,y1,x2,y2,x3,y3] + """ + N = paddle.shape(rrects)[0] + + x_ctr = rrects[:, 0] + y_ctr = rrects[:, 1] + width = rrects[:, 2] + height = rrects[:, 3] + angle = rrects[:, 4] + + tl_x, tl_y, br_x, br_y = -width * 0.5, -height * 0.5, width * 0.5, height * 0.5 + + normal_rects = paddle.stack( + [tl_x, br_x, br_x, tl_x, tl_y, tl_y, br_y, br_y], axis=0) + normal_rects = paddle.reshape(normal_rects, [2, 4, N]) + normal_rects = paddle.transpose(normal_rects, [2, 0, 1]) + + sin, cos = paddle.sin(angle), paddle.cos(angle) + # M.shape=[N,2,2] + M = paddle.stack([cos, -sin, sin, cos], axis=0) + M = paddle.reshape(M, [2, 2, N]) + M = paddle.transpose(M, [2, 0, 1]) + + # polys:[N,8] + polys = paddle.matmul(M, normal_rects) + polys = paddle.transpose(polys, [2, 1, 0]) + polys = paddle.reshape(polys, [-1, N]) + polys = paddle.transpose(polys, [1, 0]) + + tmp = paddle.stack( + [x_ctr, y_ctr, x_ctr, y_ctr, x_ctr, y_ctr, x_ctr, y_ctr], axis=1) + polys = polys + tmp + return polys + + +def bbox_iou_np_expand(box1, box2, x1y1x2y2=True, eps=1e-16): + """ + Calculate the iou of box1 and box2 with numpy. + + Args: + box1 (ndarray): [N, 4] + box2 (ndarray): [M, 4], usually N != M + x1y1x2y2 (bool): whether in x1y1x2y2 stype, default True + eps (float): epsilon to avoid divide by zero + Return: + iou (ndarray): iou of box1 and box2, [N, M] + """ + N, M = len(box1), len(box2) # usually N != M + if x1y1x2y2: + b1_x1, b1_y1 = box1[:, 0], box1[:, 1] + b1_x2, b1_y2 = box1[:, 2], box1[:, 3] + b2_x1, b2_y1 = box2[:, 0], box2[:, 1] + b2_x2, b2_y2 = box2[:, 2], box2[:, 3] + else: + # cxcywh style + # Transform from center and width to exact coordinates + b1_x1, b1_x2 = box1[:, 0] - box1[:, 2] / 2, box1[:, 0] + box1[:, 2] / 2 + b1_y1, b1_y2 = box1[:, 1] - box1[:, 3] / 2, box1[:, 1] + box1[:, 3] / 2 + b2_x1, b2_x2 = box2[:, 0] - box2[:, 2] / 2, box2[:, 0] + box2[:, 2] / 2 + b2_y1, b2_y2 = box2[:, 1] - box2[:, 3] / 2, box2[:, 1] + box2[:, 3] / 2 + + # get the coordinates of the intersection rectangle + inter_rect_x1 = np.zeros((N, M), dtype=np.float32) + inter_rect_y1 = np.zeros((N, M), dtype=np.float32) + inter_rect_x2 = np.zeros((N, M), dtype=np.float32) + inter_rect_y2 = np.zeros((N, M), dtype=np.float32) + for i in range(len(box2)): + inter_rect_x1[:, i] = np.maximum(b1_x1, b2_x1[i]) + inter_rect_y1[:, i] = np.maximum(b1_y1, b2_y1[i]) + inter_rect_x2[:, i] = np.minimum(b1_x2, b2_x2[i]) + inter_rect_y2[:, i] = np.minimum(b1_y2, b2_y2[i]) + # Intersection area + inter_area = np.maximum(inter_rect_x2 - inter_rect_x1, 0) * np.maximum( + inter_rect_y2 - inter_rect_y1, 0) + # Union Area + b1_area = np.repeat( + ((b1_x2 - b1_x1) * (b1_y2 - b1_y1)).reshape(-1, 1), M, axis=-1) + b2_area = np.repeat( + ((b2_x2 - b2_x1) * (b2_y2 - b2_y1)).reshape(1, -1), N, axis=0) + + ious = inter_area / (b1_area + b2_area - inter_area + eps) + return ious + + +def bbox2distance(points, bbox, max_dis=None, eps=0.1): + """Decode bounding box based on distances. + Args: + points (Tensor): Shape (n, 2), [x, y]. + bbox (Tensor): Shape (n, 4), "xyxy" format + max_dis (float): Upper bound of the distance. + eps (float): a small value to ensure target < max_dis, instead <= + Returns: + Tensor: Decoded distances. + """ + left = points[:, 0] - bbox[:, 0] + top = points[:, 1] - bbox[:, 1] + right = bbox[:, 2] - points[:, 0] + bottom = bbox[:, 3] - points[:, 1] + if max_dis is not None: + left = left.clip(min=0, max=max_dis - eps) + top = top.clip(min=0, max=max_dis - eps) + right = right.clip(min=0, max=max_dis - eps) + bottom = bottom.clip(min=0, max=max_dis - eps) + return paddle.stack([left, top, right, bottom], -1) + + +def distance2bbox(points, distance, max_shape=None): + """Decode distance prediction to bounding box. + Args: + points (Tensor): Shape (n, 2), [x, y]. + distance (Tensor): Distance from the given point to 4 + boundaries (left, top, right, bottom). + max_shape (tuple): Shape of the image. + Returns: + Tensor: Decoded bboxes. + """ + x1 = points[:, 0] - distance[:, 0] + y1 = points[:, 1] - distance[:, 1] + x2 = points[:, 0] + distance[:, 2] + y2 = points[:, 1] + distance[:, 3] + if max_shape is not None: + x1 = x1.clip(min=0, max=max_shape[1]) + y1 = y1.clip(min=0, max=max_shape[0]) + x2 = x2.clip(min=0, max=max_shape[1]) + y2 = y2.clip(min=0, max=max_shape[0]) + return paddle.stack([x1, y1, x2, y2], -1) + + +def bbox_center(boxes): + """Get bbox centers from boxes. + Args: + boxes (Tensor): boxes with shape (N, 4), "xmin, ymin, xmax, ymax" format. + Returns: + Tensor: boxes centers with shape (N, 2), "cx, cy" format. + """ + boxes_cx = (boxes[:, 0] + boxes[:, 2]) / 2 + boxes_cy = (boxes[:, 1] + boxes[:, 3]) / 2 + return paddle.stack([boxes_cx, boxes_cy], axis=-1) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__init__.py new file mode 100644 index 000000000..b6b928608 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__init__.py @@ -0,0 +1,53 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import bbox_head +from . import mask_head +from . import yolo_head +from . import roi_extractor +from . import ssd_head +from . import fcos_head +from . import solov2_head +from . import ttf_head +from . import cascade_head +from . import face_head +from . import s2anet_head +from . import keypoint_hrhrnet_head +from . import centernet_head +from . import gfl_head +from . import simota_head +from . import pico_head +from . import detr_head +from . import sparsercnn_head +from . import tood_head + +from .bbox_head import * +from .mask_head import * +from .yolo_head import * +from .roi_extractor import * +from .ssd_head import * +from .fcos_head import * +from .solov2_head import * +from .ttf_head import * +from .cascade_head import * +from .face_head import * +from .s2anet_head import * +from .keypoint_hrhrnet_head import * +from .centernet_head import * +from .gfl_head import * +from .simota_head import * +from .pico_head import * +from .detr_head import * +from .sparsercnn_head import * +from .tood_head import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..9813bf8c0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/bbox_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/bbox_head.cpython-37.pyc new file mode 100644 index 000000000..6e35d8b73 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/bbox_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/cascade_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/cascade_head.cpython-37.pyc new file mode 100644 index 000000000..fc1fc24f8 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/cascade_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/centernet_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/centernet_head.cpython-37.pyc new file mode 100644 index 000000000..78dfbeb45 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/centernet_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/detr_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/detr_head.cpython-37.pyc new file mode 100644 index 000000000..d72d46bc6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/detr_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/face_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/face_head.cpython-37.pyc new file mode 100644 index 000000000..c77948da9 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/face_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/fcos_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/fcos_head.cpython-37.pyc new file mode 100644 index 000000000..b2efd5f47 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/fcos_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/gfl_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/gfl_head.cpython-37.pyc new file mode 100644 index 000000000..2478604a1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/gfl_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/keypoint_hrhrnet_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/keypoint_hrhrnet_head.cpython-37.pyc new file mode 100644 index 000000000..fff6a345f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/keypoint_hrhrnet_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/mask_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/mask_head.cpython-37.pyc new file mode 100644 index 000000000..0dce48ccb Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/mask_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/pico_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/pico_head.cpython-37.pyc new file mode 100644 index 000000000..ef672f171 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/pico_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/roi_extractor.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/roi_extractor.cpython-37.pyc new file mode 100644 index 000000000..f432d300c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/roi_extractor.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/s2anet_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/s2anet_head.cpython-37.pyc new file mode 100644 index 000000000..3cf65d2bd Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/s2anet_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/simota_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/simota_head.cpython-37.pyc new file mode 100644 index 000000000..302f42806 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/simota_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/solov2_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/solov2_head.cpython-37.pyc new file mode 100644 index 000000000..ed7b79f62 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/solov2_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/sparsercnn_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/sparsercnn_head.cpython-37.pyc new file mode 100644 index 000000000..ab0357c75 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/sparsercnn_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ssd_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ssd_head.cpython-37.pyc new file mode 100644 index 000000000..8d47ca572 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ssd_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/tood_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/tood_head.cpython-37.pyc new file mode 100644 index 000000000..4fdc6a1d0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/tood_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ttf_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ttf_head.cpython-37.pyc new file mode 100644 index 000000000..bc43a5621 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/ttf_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/yolo_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/yolo_head.cpython-37.pyc new file mode 100644 index 000000000..6e73b0ef0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/__pycache__/yolo_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/bbox_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/bbox_head.py new file mode 100644 index 000000000..e4d7d6878 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/bbox_head.py @@ -0,0 +1,376 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, XavierUniform, KaimingNormal +from paddle.regularizer import L2Decay + +from ppdet.core.workspace import register, create +from .roi_extractor import RoIAlign +from ..shape_spec import ShapeSpec +from ..bbox_utils import bbox2delta +from ppdet.modeling.layers import ConvNormLayer + +__all__ = ['TwoFCHead', 'XConvNormHead', 'BBoxHead'] + + +@register +class TwoFCHead(nn.Layer): + """ + RCNN bbox head with Two fc layers to extract feature + + Args: + in_channel (int): Input channel which can be derived by from_config + out_channel (int): Output channel + resolution (int): Resolution of input feature map, default 7 + """ + + def __init__(self, in_channel=256, out_channel=1024, resolution=7): + super(TwoFCHead, self).__init__() + self.in_channel = in_channel + self.out_channel = out_channel + fan = in_channel * resolution * resolution + self.fc6 = nn.Linear( + in_channel * resolution * resolution, + out_channel, + weight_attr=paddle.ParamAttr( + initializer=XavierUniform(fan_out=fan))) + self.fc6.skip_quant = True + + self.fc7 = nn.Linear( + out_channel, + out_channel, + weight_attr=paddle.ParamAttr(initializer=XavierUniform())) + self.fc7.skip_quant = True + + @classmethod + def from_config(cls, cfg, input_shape): + s = input_shape + s = s[0] if isinstance(s, (list, tuple)) else s + return {'in_channel': s.channels} + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, )] + + def forward(self, rois_feat): + rois_feat = paddle.flatten(rois_feat, start_axis=1, stop_axis=-1) + fc6 = self.fc6(rois_feat) + fc6 = F.relu(fc6) + fc7 = self.fc7(fc6) + fc7 = F.relu(fc7) + return fc7 + + +@register +class XConvNormHead(nn.Layer): + __shared__ = ['norm_type', 'freeze_norm'] + """ + RCNN bbox head with serveral convolution layers + + Args: + in_channel (int): Input channels which can be derived by from_config + num_convs (int): The number of conv layers + conv_dim (int): The number of channels for the conv layers + out_channel (int): Output channels + resolution (int): Resolution of input feature map + norm_type (string): Norm type, bn, gn, sync_bn are available, + default `gn` + freeze_norm (bool): Whether to freeze the norm + stage_name (string): Prefix name for conv layer, '' by default + """ + + def __init__(self, + in_channel=256, + num_convs=4, + conv_dim=256, + out_channel=1024, + resolution=7, + norm_type='gn', + freeze_norm=False, + stage_name=''): + super(XConvNormHead, self).__init__() + self.in_channel = in_channel + self.num_convs = num_convs + self.conv_dim = conv_dim + self.out_channel = out_channel + self.norm_type = norm_type + self.freeze_norm = freeze_norm + + self.bbox_head_convs = [] + fan = conv_dim * 3 * 3 + initializer = KaimingNormal(fan_in=fan) + for i in range(self.num_convs): + in_c = in_channel if i == 0 else conv_dim + head_conv_name = stage_name + 'bbox_head_conv{}'.format(i) + head_conv = self.add_sublayer( + head_conv_name, + ConvNormLayer( + ch_in=in_c, + ch_out=conv_dim, + filter_size=3, + stride=1, + norm_type=self.norm_type, + freeze_norm=self.freeze_norm, + initializer=initializer)) + self.bbox_head_convs.append(head_conv) + + fan = conv_dim * resolution * resolution + self.fc6 = nn.Linear( + conv_dim * resolution * resolution, + out_channel, + weight_attr=paddle.ParamAttr( + initializer=XavierUniform(fan_out=fan)), + bias_attr=paddle.ParamAttr( + learning_rate=2., regularizer=L2Decay(0.))) + + @classmethod + def from_config(cls, cfg, input_shape): + s = input_shape + s = s[0] if isinstance(s, (list, tuple)) else s + return {'in_channel': s.channels} + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, )] + + def forward(self, rois_feat): + for i in range(self.num_convs): + rois_feat = F.relu(self.bbox_head_convs[i](rois_feat)) + rois_feat = paddle.flatten(rois_feat, start_axis=1, stop_axis=-1) + fc6 = F.relu(self.fc6(rois_feat)) + return fc6 + + +@register +class BBoxHead(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['bbox_assigner', 'bbox_loss'] + """ + RCNN bbox head + + Args: + head (nn.Layer): Extract feature in bbox head + in_channel (int): Input channel after RoI extractor + roi_extractor (object): The module of RoI Extractor + bbox_assigner (object): The module of Box Assigner, label and sample the + box. + with_pool (bool): Whether to use pooling for the RoI feature. + num_classes (int): The number of classes + bbox_weight (List[float]): The weight to get the decode box + """ + + def __init__(self, + head, + in_channel, + roi_extractor=RoIAlign().__dict__, + bbox_assigner='BboxAssigner', + with_pool=False, + num_classes=80, + bbox_weight=[10., 10., 5., 5.], + bbox_loss=None): + super(BBoxHead, self).__init__() + self.head = head + self.roi_extractor = roi_extractor + if isinstance(roi_extractor, dict): + self.roi_extractor = RoIAlign(**roi_extractor) + self.bbox_assigner = bbox_assigner + + self.with_pool = with_pool + self.num_classes = num_classes + self.bbox_weight = bbox_weight + self.bbox_loss = bbox_loss + + self.bbox_score = nn.Linear( + in_channel, + self.num_classes + 1, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0.0, std=0.01))) + self.bbox_score.skip_quant = True + + self.bbox_delta = nn.Linear( + in_channel, + 4 * self.num_classes, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0.0, std=0.001))) + self.bbox_delta.skip_quant = True + self.assigned_label = None + self.assigned_rois = None + + @classmethod + def from_config(cls, cfg, input_shape): + roi_pooler = cfg['roi_extractor'] + assert isinstance(roi_pooler, dict) + kwargs = RoIAlign.from_config(cfg, input_shape) + roi_pooler.update(kwargs) + kwargs = {'input_shape': input_shape} + head = create(cfg['head'], **kwargs) + return { + 'roi_extractor': roi_pooler, + 'head': head, + 'in_channel': head.out_shape[0].channels + } + + def forward(self, body_feats=None, rois=None, rois_num=None, inputs=None): + """ + body_feats (list[Tensor]): Feature maps from backbone + rois (list[Tensor]): RoIs generated from RPN module + rois_num (Tensor): The number of RoIs in each image + inputs (dict{Tensor}): The ground-truth of image + """ + if self.training: + rois, rois_num, targets = self.bbox_assigner(rois, rois_num, inputs) + self.assigned_rois = (rois, rois_num) + self.assigned_targets = targets + + rois_feat = self.roi_extractor(body_feats, rois, rois_num) + bbox_feat = self.head(rois_feat) + if self.with_pool: + feat = F.adaptive_avg_pool2d(bbox_feat, output_size=1) + feat = paddle.squeeze(feat, axis=[2, 3]) + else: + feat = bbox_feat + scores = self.bbox_score(feat) + deltas = self.bbox_delta(feat) + + if self.training: + loss = self.get_loss(scores, deltas, targets, rois, + self.bbox_weight) + return loss, bbox_feat + else: + pred = self.get_prediction(scores, deltas) + return pred, self.head + + def get_loss(self, scores, deltas, targets, rois, bbox_weight): + """ + scores (Tensor): scores from bbox head outputs + deltas (Tensor): deltas from bbox head outputs + targets (list[List[Tensor]]): bbox targets containing tgt_labels, tgt_bboxes and tgt_gt_inds + rois (List[Tensor]): RoIs generated in each batch + """ + cls_name = 'loss_bbox_cls' + reg_name = 'loss_bbox_reg' + loss_bbox = {} + + # TODO: better pass args + tgt_labels, tgt_bboxes, tgt_gt_inds = targets + + # bbox cls + tgt_labels = paddle.concat(tgt_labels) if len( + tgt_labels) > 1 else tgt_labels[0] + valid_inds = paddle.nonzero(tgt_labels >= 0).flatten() + if valid_inds.shape[0] == 0: + loss_bbox[cls_name] = paddle.zeros([1], dtype='float32') + else: + tgt_labels = tgt_labels.cast('int64') + tgt_labels.stop_gradient = True + loss_bbox_cls = F.cross_entropy( + input=scores, label=tgt_labels, reduction='mean') + loss_bbox[cls_name] = loss_bbox_cls + + # bbox reg + + cls_agnostic_bbox_reg = deltas.shape[1] == 4 + + fg_inds = paddle.nonzero( + paddle.logical_and(tgt_labels >= 0, tgt_labels < + self.num_classes)).flatten() + + if fg_inds.numel() == 0: + loss_bbox[reg_name] = paddle.zeros([1], dtype='float32') + return loss_bbox + + if cls_agnostic_bbox_reg: + reg_delta = paddle.gather(deltas, fg_inds) + else: + fg_gt_classes = paddle.gather(tgt_labels, fg_inds) + + reg_row_inds = paddle.arange(fg_gt_classes.shape[0]).unsqueeze(1) + reg_row_inds = paddle.tile(reg_row_inds, [1, 4]).reshape([-1, 1]) + + reg_col_inds = 4 * fg_gt_classes.unsqueeze(1) + paddle.arange(4) + + reg_col_inds = reg_col_inds.reshape([-1, 1]) + reg_inds = paddle.concat([reg_row_inds, reg_col_inds], axis=1) + + reg_delta = paddle.gather(deltas, fg_inds) + reg_delta = paddle.gather_nd(reg_delta, reg_inds).reshape([-1, 4]) + rois = paddle.concat(rois) if len(rois) > 1 else rois[0] + tgt_bboxes = paddle.concat(tgt_bboxes) if len( + tgt_bboxes) > 1 else tgt_bboxes[0] + + reg_target = bbox2delta(rois, tgt_bboxes, bbox_weight) + reg_target = paddle.gather(reg_target, fg_inds) + reg_target.stop_gradient = True + + if self.bbox_loss is not None: + reg_delta = self.bbox_transform(reg_delta) + reg_target = self.bbox_transform(reg_target) + loss_bbox_reg = self.bbox_loss( + reg_delta, reg_target).sum() / tgt_labels.shape[0] + loss_bbox_reg *= self.num_classes + else: + loss_bbox_reg = paddle.abs(reg_delta - reg_target).sum( + ) / tgt_labels.shape[0] + + loss_bbox[reg_name] = loss_bbox_reg + + return loss_bbox + + def bbox_transform(self, deltas, weights=[0.1, 0.1, 0.2, 0.2]): + wx, wy, ww, wh = weights + + deltas = paddle.reshape(deltas, shape=(0, -1, 4)) + + dx = paddle.slice(deltas, axes=[2], starts=[0], ends=[1]) * wx + dy = paddle.slice(deltas, axes=[2], starts=[1], ends=[2]) * wy + dw = paddle.slice(deltas, axes=[2], starts=[2], ends=[3]) * ww + dh = paddle.slice(deltas, axes=[2], starts=[3], ends=[4]) * wh + + dw = paddle.clip(dw, -1.e10, np.log(1000. / 16)) + dh = paddle.clip(dh, -1.e10, np.log(1000. / 16)) + + pred_ctr_x = dx + pred_ctr_y = dy + pred_w = paddle.exp(dw) + pred_h = paddle.exp(dh) + + x1 = pred_ctr_x - 0.5 * pred_w + y1 = pred_ctr_y - 0.5 * pred_h + x2 = pred_ctr_x + 0.5 * pred_w + y2 = pred_ctr_y + 0.5 * pred_h + + x1 = paddle.reshape(x1, shape=(-1, )) + y1 = paddle.reshape(y1, shape=(-1, )) + x2 = paddle.reshape(x2, shape=(-1, )) + y2 = paddle.reshape(y2, shape=(-1, )) + + return paddle.concat([x1, y1, x2, y2]) + + def get_prediction(self, score, delta): + bbox_prob = F.softmax(score) + return delta, bbox_prob + + def get_head(self, ): + return self.head + + def get_assigned_targets(self, ): + return self.assigned_targets + + def get_assigned_rois(self, ): + return self.assigned_rois diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/cascade_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/cascade_head.py new file mode 100644 index 000000000..935642bd6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/cascade_head.py @@ -0,0 +1,283 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal + +from ppdet.core.workspace import register +from .bbox_head import BBoxHead, TwoFCHead, XConvNormHead +from .roi_extractor import RoIAlign +from ..shape_spec import ShapeSpec +from ..bbox_utils import delta2bbox, clip_bbox, nonempty_bbox + +__all__ = ['CascadeTwoFCHead', 'CascadeXConvNormHead', 'CascadeHead'] + + +@register +class CascadeTwoFCHead(nn.Layer): + __shared__ = ['num_cascade_stage'] + """ + Cascade RCNN bbox head with Two fc layers to extract feature + + Args: + in_channel (int): Input channel which can be derived by from_config + out_channel (int): Output channel + resolution (int): Resolution of input feature map, default 7 + num_cascade_stage (int): The number of cascade stage, default 3 + """ + + def __init__(self, + in_channel=256, + out_channel=1024, + resolution=7, + num_cascade_stage=3): + super(CascadeTwoFCHead, self).__init__() + + self.in_channel = in_channel + self.out_channel = out_channel + + self.head_list = [] + for stage in range(num_cascade_stage): + head_per_stage = self.add_sublayer( + str(stage), TwoFCHead(in_channel, out_channel, resolution)) + self.head_list.append(head_per_stage) + + @classmethod + def from_config(cls, cfg, input_shape): + s = input_shape + s = s[0] if isinstance(s, (list, tuple)) else s + return {'in_channel': s.channels} + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, )] + + def forward(self, rois_feat, stage=0): + out = self.head_list[stage](rois_feat) + return out + + +@register +class CascadeXConvNormHead(nn.Layer): + __shared__ = ['norm_type', 'freeze_norm', 'num_cascade_stage'] + """ + Cascade RCNN bbox head with serveral convolution layers + + Args: + in_channel (int): Input channels which can be derived by from_config + num_convs (int): The number of conv layers + conv_dim (int): The number of channels for the conv layers + out_channel (int): Output channels + resolution (int): Resolution of input feature map + norm_type (string): Norm type, bn, gn, sync_bn are available, + default `gn` + freeze_norm (bool): Whether to freeze the norm + num_cascade_stage (int): The number of cascade stage, default 3 + """ + + def __init__(self, + in_channel=256, + num_convs=4, + conv_dim=256, + out_channel=1024, + resolution=7, + norm_type='gn', + freeze_norm=False, + num_cascade_stage=3): + super(CascadeXConvNormHead, self).__init__() + self.in_channel = in_channel + self.out_channel = out_channel + + self.head_list = [] + for stage in range(num_cascade_stage): + head_per_stage = self.add_sublayer( + str(stage), + XConvNormHead( + in_channel, + num_convs, + conv_dim, + out_channel, + resolution, + norm_type, + freeze_norm, + stage_name='stage{}_'.format(stage))) + self.head_list.append(head_per_stage) + + @classmethod + def from_config(cls, cfg, input_shape): + s = input_shape + s = s[0] if isinstance(s, (list, tuple)) else s + return {'in_channel': s.channels} + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, )] + + def forward(self, rois_feat, stage=0): + out = self.head_list[stage](rois_feat) + return out + + +@register +class CascadeHead(BBoxHead): + __shared__ = ['num_classes', 'num_cascade_stages'] + __inject__ = ['bbox_assigner', 'bbox_loss'] + """ + Cascade RCNN bbox head + + Args: + head (nn.Layer): Extract feature in bbox head + in_channel (int): Input channel after RoI extractor + roi_extractor (object): The module of RoI Extractor + bbox_assigner (object): The module of Box Assigner, label and sample the + box. + num_classes (int): The number of classes + bbox_weight (List[List[float]]): The weight to get the decode box and the + length of weight is the number of cascade stage + num_cascade_stages (int): THe number of stage to refine the box + """ + + def __init__(self, + head, + in_channel, + roi_extractor=RoIAlign().__dict__, + bbox_assigner='BboxAssigner', + num_classes=80, + bbox_weight=[[10., 10., 5., 5.], [20.0, 20.0, 10.0, 10.0], + [30.0, 30.0, 15.0, 15.0]], + num_cascade_stages=3, + bbox_loss=None): + nn.Layer.__init__(self, ) + self.head = head + self.roi_extractor = roi_extractor + if isinstance(roi_extractor, dict): + self.roi_extractor = RoIAlign(**roi_extractor) + self.bbox_assigner = bbox_assigner + + self.num_classes = num_classes + self.bbox_weight = bbox_weight + self.num_cascade_stages = num_cascade_stages + self.bbox_loss = bbox_loss + + self.bbox_score_list = [] + self.bbox_delta_list = [] + for i in range(num_cascade_stages): + score_name = 'bbox_score_stage{}'.format(i) + delta_name = 'bbox_delta_stage{}'.format(i) + bbox_score = self.add_sublayer( + score_name, + nn.Linear( + in_channel, + self.num_classes + 1, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0.0, std=0.01)))) + + bbox_delta = self.add_sublayer( + delta_name, + nn.Linear( + in_channel, + 4, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0.0, std=0.001)))) + self.bbox_score_list.append(bbox_score) + self.bbox_delta_list.append(bbox_delta) + self.assigned_label = None + self.assigned_rois = None + + def forward(self, body_feats=None, rois=None, rois_num=None, inputs=None): + """ + body_feats (list[Tensor]): Feature maps from backbone + rois (Tensor): RoIs generated from RPN module + rois_num (Tensor): The number of RoIs in each image + inputs (dict{Tensor}): The ground-truth of image + """ + targets = [] + if self.training: + rois, rois_num, targets = self.bbox_assigner(rois, rois_num, inputs) + targets_list = [targets] + self.assigned_rois = (rois, rois_num) + self.assigned_targets = targets + + pred_bbox = None + head_out_list = [] + for i in range(self.num_cascade_stages): + if i > 0: + rois, rois_num = self._get_rois_from_boxes(pred_bbox, + inputs['im_shape']) + if self.training: + rois, rois_num, targets = self.bbox_assigner( + rois, rois_num, inputs, i, is_cascade=True) + targets_list.append(targets) + + rois_feat = self.roi_extractor(body_feats, rois, rois_num) + bbox_feat = self.head(rois_feat, i) + scores = self.bbox_score_list[i](bbox_feat) + deltas = self.bbox_delta_list[i](bbox_feat) + head_out_list.append([scores, deltas, rois]) + pred_bbox = self._get_pred_bbox(deltas, rois, self.bbox_weight[i]) + + if self.training: + loss = {} + for stage, value in enumerate(zip(head_out_list, targets_list)): + (scores, deltas, rois), targets = value + loss_stage = self.get_loss(scores, deltas, targets, rois, + self.bbox_weight[stage]) + for k, v in loss_stage.items(): + loss[k + "_stage{}".format( + stage)] = v / self.num_cascade_stages + + return loss, bbox_feat + else: + scores, deltas, self.refined_rois = self.get_prediction( + head_out_list) + return (deltas, scores), self.head + + def _get_rois_from_boxes(self, boxes, im_shape): + rois = [] + for i, boxes_per_image in enumerate(boxes): + clip_box = clip_bbox(boxes_per_image, im_shape[i]) + if self.training: + keep = nonempty_bbox(clip_box) + if keep.shape[0] == 0: + keep = paddle.zeros([1], dtype='int32') + clip_box = paddle.gather(clip_box, keep) + rois.append(clip_box) + rois_num = paddle.concat([paddle.shape(r)[0] for r in rois]) + return rois, rois_num + + def _get_pred_bbox(self, deltas, proposals, weights): + pred_proposals = paddle.concat(proposals) if len( + proposals) > 1 else proposals[0] + pred_bbox = delta2bbox(deltas, pred_proposals, weights) + pred_bbox = paddle.reshape(pred_bbox, [-1, deltas.shape[-1]]) + num_prop = [] + for p in proposals: + num_prop.append(p.shape[0]) + return pred_bbox.split(num_prop) + + def get_prediction(self, head_out_list): + """ + head_out_list(List[Tensor]): scores, deltas, rois + """ + pred_list = [] + scores_list = [F.softmax(head[0]) for head in head_out_list] + scores = paddle.add_n(scores_list) / self.num_cascade_stages + # Get deltas and rois from the last stage + _, deltas, rois = head_out_list[-1] + return scores, deltas, rois + + def get_refined_rois(self, ): + return self.refined_rois diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/centernet_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/centernet_head.py new file mode 100644 index 000000000..ce8b5c15d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/centernet_head.py @@ -0,0 +1,291 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import math +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Constant, Uniform +from ppdet.core.workspace import register +from ppdet.modeling.losses import CTFocalLoss, GIoULoss + + +class ConvLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + kernel_size, + stride=1, + padding=0, + dilation=1, + groups=1, + bias=False): + super(ConvLayer, self).__init__() + bias_attr = False + fan_in = ch_in * kernel_size**2 + bound = 1 / math.sqrt(fan_in) + param_attr = paddle.ParamAttr(initializer=Uniform(-bound, bound)) + if bias: + bias_attr = paddle.ParamAttr(initializer=Constant(0.)) + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=kernel_size, + stride=stride, + padding=padding, + dilation=dilation, + groups=groups, + weight_attr=param_attr, + bias_attr=bias_attr) + + def forward(self, inputs): + out = self.conv(inputs) + return out + + +@register +class CenterNetHead(nn.Layer): + """ + Args: + in_channels (int): the channel number of input to CenterNetHead. + num_classes (int): the number of classes, 80 (COCO dataset) by default. + head_planes (int): the channel number in all head, 256 by default. + heatmap_weight (float): the weight of heatmap loss, 1 by default. + regress_ltrb (bool): whether to regress left/top/right/bottom or + width/height for a box, true by default + size_weight (float): the weight of box size loss, 0.1 by default. + size_loss (): the type of size regression loss, 'L1 loss' by default. + offset_weight (float): the weight of center offset loss, 1 by default. + iou_weight (float): the weight of iou head loss, 0 by default. + """ + + __shared__ = ['num_classes'] + + def __init__(self, + in_channels, + num_classes=80, + head_planes=256, + heatmap_weight=1, + regress_ltrb=True, + size_weight=0.1, + size_loss='L1', + offset_weight=1, + iou_weight=0): + super(CenterNetHead, self).__init__() + self.regress_ltrb = regress_ltrb + self.weights = { + 'heatmap': heatmap_weight, + 'size': size_weight, + 'offset': offset_weight, + 'iou': iou_weight + } + + # heatmap head + self.heatmap = nn.Sequential( + ConvLayer( + in_channels, head_planes, kernel_size=3, padding=1, bias=True), + nn.ReLU(), + ConvLayer( + head_planes, + num_classes, + kernel_size=1, + stride=1, + padding=0, + bias=True)) + with paddle.no_grad(): + self.heatmap[2].conv.bias[:] = -2.19 + + # size(ltrb or wh) head + self.size = nn.Sequential( + ConvLayer( + in_channels, head_planes, kernel_size=3, padding=1, bias=True), + nn.ReLU(), + ConvLayer( + head_planes, + 4 if regress_ltrb else 2, + kernel_size=1, + stride=1, + padding=0, + bias=True)) + self.size_loss = size_loss + + # offset head + self.offset = nn.Sequential( + ConvLayer( + in_channels, head_planes, kernel_size=3, padding=1, bias=True), + nn.ReLU(), + ConvLayer( + head_planes, 2, kernel_size=1, stride=1, padding=0, bias=True)) + + # iou head (optinal) + if iou_weight > 0: + self.iou = nn.Sequential( + ConvLayer( + in_channels, + head_planes, + kernel_size=3, + padding=1, + bias=True), + nn.ReLU(), + ConvLayer( + head_planes, + 4 if regress_ltrb else 2, + kernel_size=1, + stride=1, + padding=0, + bias=True)) + + @classmethod + def from_config(cls, cfg, input_shape): + if isinstance(input_shape, (list, tuple)): + input_shape = input_shape[0] + return {'in_channels': input_shape.channels} + + def forward(self, feat, inputs): + heatmap = self.heatmap(feat) + size = self.size(feat) + offset = self.offset(feat) + iou = self.iou(feat) if hasattr(self, 'iou_weight') else None + + if self.training: + loss = self.get_loss( + inputs, self.weights, heatmap, size, offset, iou=iou) + return loss + else: + heatmap = F.sigmoid(heatmap) + head_outs = {'heatmap': heatmap, 'size': size, 'offset': offset} + if iou is not None: + head_outs.update({'iou': iou}) + return head_outs + + def get_loss(self, inputs, weights, heatmap, size, offset, iou=None): + # heatmap head loss: CTFocalLoss + heatmap_target = inputs['heatmap'] + heatmap = paddle.clip(F.sigmoid(heatmap), 1e-4, 1 - 1e-4) + ctfocal_loss = CTFocalLoss() + heatmap_loss = ctfocal_loss(heatmap, heatmap_target) + + # size head loss: L1 loss or GIoU loss + index = inputs['index'] + mask = inputs['index_mask'] + size = paddle.transpose(size, perm=[0, 2, 3, 1]) + size_n, size_h, size_w, size_c = size.shape + size = paddle.reshape(size, shape=[size_n, -1, size_c]) + index = paddle.unsqueeze(index, 2) + batch_inds = list() + for i in range(size_n): + batch_ind = paddle.full( + shape=[1, index.shape[1], 1], fill_value=i, dtype='int64') + batch_inds.append(batch_ind) + batch_inds = paddle.concat(batch_inds, axis=0) + index = paddle.concat(x=[batch_inds, index], axis=2) + pos_size = paddle.gather_nd(size, index=index) + mask = paddle.unsqueeze(mask, axis=2) + size_mask = paddle.expand_as(mask, pos_size) + size_mask = paddle.cast(size_mask, dtype=pos_size.dtype) + pos_num = size_mask.sum() + size_mask.stop_gradient = True + if self.size_loss == 'L1': + if self.regress_ltrb: + size_target = inputs['size'] + # shape: [bs, max_per_img, 4] + else: + if inputs['size'].shape[-1] == 2: + # inputs['size'] is wh, and regress as wh + # shape: [bs, max_per_img, 2] + size_target = inputs['size'] + else: + # inputs['size'] is ltrb, but regress as wh + # shape: [bs, max_per_img, 4] + size_target = inputs['size'][:, :, 0:2] + inputs['size'][:, :, 2:] + + size_target.stop_gradient = True + size_loss = F.l1_loss( + pos_size * size_mask, size_target * size_mask, reduction='sum') + size_loss = size_loss / (pos_num + 1e-4) + elif self.size_loss == 'giou': + size_target = inputs['bbox_xys'] + size_target.stop_gradient = True + centers_x = (size_target[:, :, 0:1] + size_target[:, :, 2:3]) / 2.0 + centers_y = (size_target[:, :, 1:2] + size_target[:, :, 3:4]) / 2.0 + x1 = centers_x - pos_size[:, :, 0:1] + y1 = centers_y - pos_size[:, :, 1:2] + x2 = centers_x + pos_size[:, :, 2:3] + y2 = centers_y + pos_size[:, :, 3:4] + pred_boxes = paddle.concat([x1, y1, x2, y2], axis=-1) + giou_loss = GIoULoss(reduction='sum') + size_loss = giou_loss( + pred_boxes * size_mask, + size_target * size_mask, + iou_weight=size_mask, + loc_reweight=None) + size_loss = size_loss / (pos_num + 1e-4) + + # offset head loss: L1 loss + offset_target = inputs['offset'] + offset = paddle.transpose(offset, perm=[0, 2, 3, 1]) + offset_n, offset_h, offset_w, offset_c = offset.shape + offset = paddle.reshape(offset, shape=[offset_n, -1, offset_c]) + pos_offset = paddle.gather_nd(offset, index=index) + offset_mask = paddle.expand_as(mask, pos_offset) + offset_mask = paddle.cast(offset_mask, dtype=pos_offset.dtype) + pos_num = offset_mask.sum() + offset_mask.stop_gradient = True + offset_target.stop_gradient = True + offset_loss = F.l1_loss( + pos_offset * offset_mask, + offset_target * offset_mask, + reduction='sum') + offset_loss = offset_loss / (pos_num + 1e-4) + + # iou head loss: GIoU loss + if iou is not None: + iou = paddle.transpose(iou, perm=[0, 2, 3, 1]) + iou_n, iou_h, iou_w, iou_c = iou.shape + iou = paddle.reshape(iou, shape=[iou_n, -1, iou_c]) + pos_iou = paddle.gather_nd(iou, index=index) + iou_mask = paddle.expand_as(mask, pos_iou) + iou_mask = paddle.cast(iou_mask, dtype=pos_iou.dtype) + pos_num = iou_mask.sum() + iou_mask.stop_gradient = True + gt_bbox_xys = inputs['bbox_xys'] + gt_bbox_xys.stop_gradient = True + centers_x = (gt_bbox_xys[:, :, 0:1] + gt_bbox_xys[:, :, 2:3]) / 2.0 + centers_y = (gt_bbox_xys[:, :, 1:2] + gt_bbox_xys[:, :, 3:4]) / 2.0 + x1 = centers_x - pos_size[:, :, 0:1] + y1 = centers_y - pos_size[:, :, 1:2] + x2 = centers_x + pos_size[:, :, 2:3] + y2 = centers_y + pos_size[:, :, 3:4] + pred_boxes = paddle.concat([x1, y1, x2, y2], axis=-1) + giou_loss = GIoULoss(reduction='sum') + iou_loss = giou_loss( + pred_boxes * iou_mask, + gt_bbox_xys * iou_mask, + iou_weight=iou_mask, + loc_reweight=None) + iou_loss = iou_loss / (pos_num + 1e-4) + + losses = { + 'heatmap_loss': heatmap_loss, + 'size_loss': size_loss, + 'offset_loss': offset_loss, + } + det_loss = weights['heatmap'] * heatmap_loss + weights[ + 'size'] * size_loss + weights['offset'] * offset_loss + + if iou is not None: + losses.update({'iou_loss': iou_loss}) + det_loss = det_loss + weights['iou'] * iou_loss + losses.update({'det_loss': det_loss}) + return losses diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/detr_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/detr_head.py new file mode 100644 index 000000000..6ca3499b9 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/detr_head.py @@ -0,0 +1,364 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +import pycocotools.mask as mask_util +from ..initializer import linear_init_, constant_ +from ..transformers.utils import inverse_sigmoid + +__all__ = ['DETRHead', 'DeformableDETRHead'] + + +class MLP(nn.Layer): + """This code is based on + https://github.com/facebookresearch/detr/blob/main/models/detr.py + """ + + def __init__(self, input_dim, hidden_dim, output_dim, num_layers): + super().__init__() + self.num_layers = num_layers + h = [hidden_dim] * (num_layers - 1) + self.layers = nn.LayerList( + nn.Linear(n, k) for n, k in zip([input_dim] + h, h + [output_dim])) + + self._reset_parameters() + + def _reset_parameters(self): + for l in self.layers: + linear_init_(l) + + def forward(self, x): + for i, layer in enumerate(self.layers): + x = F.relu(layer(x)) if i < self.num_layers - 1 else layer(x) + return x + + +class MultiHeadAttentionMap(nn.Layer): + """This code is based on + https://github.com/facebookresearch/detr/blob/main/models/segmentation.py + + This is a 2D attention module, which only returns the attention softmax (no multiplication by value) + """ + + def __init__(self, query_dim, hidden_dim, num_heads, dropout=0.0, + bias=True): + super().__init__() + self.num_heads = num_heads + self.hidden_dim = hidden_dim + self.dropout = nn.Dropout(dropout) + + weight_attr = paddle.ParamAttr( + initializer=paddle.nn.initializer.XavierUniform()) + bias_attr = paddle.framework.ParamAttr( + initializer=paddle.nn.initializer.Constant()) if bias else False + + self.q_proj = nn.Linear(query_dim, hidden_dim, weight_attr, bias_attr) + self.k_proj = nn.Conv2D( + query_dim, + hidden_dim, + 1, + weight_attr=weight_attr, + bias_attr=bias_attr) + + self.normalize_fact = float(hidden_dim / self.num_heads)**-0.5 + + def forward(self, q, k, mask=None): + q = self.q_proj(q) + k = self.k_proj(k) + bs, num_queries, n, c, h, w = q.shape[0], q.shape[1], self.num_heads,\ + self.hidden_dim // self.num_heads, k.shape[-2], k.shape[-1] + qh = q.reshape([bs, num_queries, n, c]) + kh = k.reshape([bs, n, c, h, w]) + # weights = paddle.einsum("bqnc,bnchw->bqnhw", qh * self.normalize_fact, kh) + qh = qh.transpose([0, 2, 1, 3]).reshape([-1, num_queries, c]) + kh = kh.reshape([-1, c, h * w]) + weights = paddle.bmm(qh * self.normalize_fact, kh).reshape( + [bs, n, num_queries, h, w]).transpose([0, 2, 1, 3, 4]) + + if mask is not None: + weights += mask + # fix a potenial bug: https://github.com/facebookresearch/detr/issues/247 + weights = F.softmax(weights.flatten(3), axis=-1).reshape(weights.shape) + weights = self.dropout(weights) + return weights + + +class MaskHeadFPNConv(nn.Layer): + """This code is based on + https://github.com/facebookresearch/detr/blob/main/models/segmentation.py + + Simple convolutional head, using group norm. + Upsampling is done using a FPN approach + """ + + def __init__(self, input_dim, fpn_dims, context_dim, num_groups=8): + super().__init__() + + inter_dims = [input_dim, + ] + [context_dim // (2**i) for i in range(1, 5)] + weight_attr = paddle.ParamAttr( + initializer=paddle.nn.initializer.KaimingUniform()) + bias_attr = paddle.framework.ParamAttr( + initializer=paddle.nn.initializer.Constant()) + + self.conv0 = self._make_layers(input_dim, input_dim, 3, num_groups, + weight_attr, bias_attr) + self.conv_inter = nn.LayerList() + for in_dims, out_dims in zip(inter_dims[:-1], inter_dims[1:]): + self.conv_inter.append( + self._make_layers(in_dims, out_dims, 3, num_groups, weight_attr, + bias_attr)) + + self.conv_out = nn.Conv2D( + inter_dims[-1], + 1, + 3, + padding=1, + weight_attr=weight_attr, + bias_attr=bias_attr) + + self.adapter = nn.LayerList() + for i in range(len(fpn_dims)): + self.adapter.append( + nn.Conv2D( + fpn_dims[i], + inter_dims[i + 1], + 1, + weight_attr=weight_attr, + bias_attr=bias_attr)) + + def _make_layers(self, + in_dims, + out_dims, + kernel_size, + num_groups, + weight_attr=None, + bias_attr=None): + return nn.Sequential( + nn.Conv2D( + in_dims, + out_dims, + kernel_size, + padding=kernel_size // 2, + weight_attr=weight_attr, + bias_attr=bias_attr), + nn.GroupNorm(num_groups, out_dims), + nn.ReLU()) + + def forward(self, x, bbox_attention_map, fpns): + x = paddle.concat([ + x.tile([bbox_attention_map.shape[1], 1, 1, 1]), + bbox_attention_map.flatten(0, 1) + ], 1) + x = self.conv0(x) + for inter_layer, adapter_layer, feat in zip(self.conv_inter[:-1], + self.adapter, fpns): + feat = adapter_layer(feat).tile( + [bbox_attention_map.shape[1], 1, 1, 1]) + x = inter_layer(x) + x = feat + F.interpolate(x, size=feat.shape[-2:]) + + x = self.conv_inter[-1](x) + x = self.conv_out(x) + return x + + +@register +class DETRHead(nn.Layer): + __shared__ = ['num_classes', 'hidden_dim', 'use_focal_loss'] + __inject__ = ['loss'] + + def __init__(self, + num_classes=80, + hidden_dim=256, + nhead=8, + num_mlp_layers=3, + loss='DETRLoss', + fpn_dims=[1024, 512, 256], + with_mask_head=False, + use_focal_loss=False): + super(DETRHead, self).__init__() + # add background class + self.num_classes = num_classes if use_focal_loss else num_classes + 1 + self.hidden_dim = hidden_dim + self.loss = loss + self.with_mask_head = with_mask_head + self.use_focal_loss = use_focal_loss + + self.score_head = nn.Linear(hidden_dim, self.num_classes) + self.bbox_head = MLP(hidden_dim, + hidden_dim, + output_dim=4, + num_layers=num_mlp_layers) + if self.with_mask_head: + self.bbox_attention = MultiHeadAttentionMap(hidden_dim, hidden_dim, + nhead) + self.mask_head = MaskHeadFPNConv(hidden_dim + nhead, fpn_dims, + hidden_dim) + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.score_head) + + @classmethod + def from_config(cls, cfg, hidden_dim, nhead, input_shape): + + return { + 'hidden_dim': hidden_dim, + 'nhead': nhead, + 'fpn_dims': [i.channels for i in input_shape[::-1]][1:] + } + + @staticmethod + def get_gt_mask_from_polygons(gt_poly, pad_mask): + out_gt_mask = [] + for polygons, padding in zip(gt_poly, pad_mask): + height, width = int(padding[:, 0].sum()), int(padding[0, :].sum()) + masks = [] + for obj_poly in polygons: + rles = mask_util.frPyObjects(obj_poly, height, width) + rle = mask_util.merge(rles) + masks.append( + paddle.to_tensor(mask_util.decode(rle)).astype('float32')) + masks = paddle.stack(masks) + masks_pad = paddle.zeros( + [masks.shape[0], pad_mask.shape[1], pad_mask.shape[2]]) + masks_pad[:, :height, :width] = masks + out_gt_mask.append(masks_pad) + return out_gt_mask + + def forward(self, out_transformer, body_feats, inputs=None): + r""" + Args: + out_transformer (Tuple): (feats: [num_levels, batch_size, + num_queries, hidden_dim], + memory: [batch_size, hidden_dim, h, w], + src_proj: [batch_size, h*w, hidden_dim], + src_mask: [batch_size, 1, 1, h, w]) + body_feats (List(Tensor)): list[[B, C, H, W]] + inputs (dict): dict(inputs) + """ + feats, memory, src_proj, src_mask = out_transformer + outputs_logit = self.score_head(feats) + outputs_bbox = F.sigmoid(self.bbox_head(feats)) + outputs_seg = None + if self.with_mask_head: + bbox_attention_map = self.bbox_attention(feats[-1], memory, + src_mask) + fpn_feats = [a for a in body_feats[::-1]][1:] + outputs_seg = self.mask_head(src_proj, bbox_attention_map, + fpn_feats) + outputs_seg = outputs_seg.reshape([ + feats.shape[1], feats.shape[2], outputs_seg.shape[-2], + outputs_seg.shape[-1] + ]) + + if self.training: + assert inputs is not None + assert 'gt_bbox' in inputs and 'gt_class' in inputs + gt_mask = self.get_gt_mask_from_polygons( + inputs['gt_poly'], + inputs['pad_mask']) if 'gt_poly' in inputs else None + return self.loss( + outputs_bbox, + outputs_logit, + inputs['gt_bbox'], + inputs['gt_class'], + masks=outputs_seg, + gt_mask=gt_mask) + else: + return (outputs_bbox[-1], outputs_logit[-1], outputs_seg) + + +@register +class DeformableDETRHead(nn.Layer): + __shared__ = ['num_classes', 'hidden_dim'] + __inject__ = ['loss'] + + def __init__(self, + num_classes=80, + hidden_dim=512, + nhead=8, + num_mlp_layers=3, + loss='DETRLoss'): + super(DeformableDETRHead, self).__init__() + self.num_classes = num_classes + self.hidden_dim = hidden_dim + self.nhead = nhead + self.loss = loss + + self.score_head = nn.Linear(hidden_dim, self.num_classes) + self.bbox_head = MLP(hidden_dim, + hidden_dim, + output_dim=4, + num_layers=num_mlp_layers) + + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.score_head) + constant_(self.score_head.bias, -4.595) + constant_(self.bbox_head.layers[-1].weight) + + with paddle.no_grad(): + bias = paddle.zeros_like(self.bbox_head.layers[-1].bias) + bias[2:] = -2.0 + self.bbox_head.layers[-1].bias.set_value(bias) + + @classmethod + def from_config(cls, cfg, hidden_dim, nhead, input_shape): + return {'hidden_dim': hidden_dim, 'nhead': nhead} + + def forward(self, out_transformer, body_feats, inputs=None): + r""" + Args: + out_transformer (Tuple): (feats: [num_levels, batch_size, + num_queries, hidden_dim], + memory: [batch_size, + \sum_{l=0}^{L-1} H_l \cdot W_l, hidden_dim], + reference_points: [batch_size, num_queries, 2]) + body_feats (List(Tensor)): list[[B, C, H, W]] + inputs (dict): dict(inputs) + """ + feats, memory, reference_points = out_transformer + reference_points = inverse_sigmoid(reference_points.unsqueeze(0)) + outputs_bbox = self.bbox_head(feats) + + # It's equivalent to "outputs_bbox[:, :, :, :2] += reference_points", + # but the gradient is wrong in paddle. + outputs_bbox = paddle.concat( + [ + outputs_bbox[:, :, :, :2] + reference_points, + outputs_bbox[:, :, :, 2:] + ], + axis=-1) + + outputs_bbox = F.sigmoid(outputs_bbox) + outputs_logit = self.score_head(feats) + + if self.training: + assert inputs is not None + assert 'gt_bbox' in inputs and 'gt_class' in inputs + + return self.loss(outputs_bbox, outputs_logit, inputs['gt_bbox'], + inputs['gt_class']) + else: + return (outputs_bbox[-1], outputs_logit[-1], None) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/face_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/face_head.py new file mode 100644 index 000000000..bb51f2eb9 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/face_head.py @@ -0,0 +1,110 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register +from ..layers import AnchorGeneratorSSD + + +@register +class FaceHead(nn.Layer): + """ + Head block for Face detection network + + Args: + num_classes (int): Number of output classes. + in_channels (int): Number of input channels. + anchor_generator(object): instance of anchor genertor method. + kernel_size (int): kernel size of Conv2D in FaceHead. + padding (int): padding of Conv2D in FaceHead. + conv_decay (float): norm_decay (float): weight decay for conv layer weights. + loss (object): loss of face detection model. + """ + __shared__ = ['num_classes'] + __inject__ = ['anchor_generator', 'loss'] + + def __init__(self, + num_classes=80, + in_channels=[96, 96], + anchor_generator=AnchorGeneratorSSD().__dict__, + kernel_size=3, + padding=1, + conv_decay=0., + loss='SSDLoss'): + super(FaceHead, self).__init__() + # add background class + self.num_classes = num_classes + 1 + self.in_channels = in_channels + self.anchor_generator = anchor_generator + self.loss = loss + + if isinstance(anchor_generator, dict): + self.anchor_generator = AnchorGeneratorSSD(**anchor_generator) + + self.num_priors = self.anchor_generator.num_priors + self.box_convs = [] + self.score_convs = [] + for i, num_prior in enumerate(self.num_priors): + box_conv_name = "boxes{}".format(i) + box_conv = self.add_sublayer( + box_conv_name, + nn.Conv2D( + in_channels=self.in_channels[i], + out_channels=num_prior * 4, + kernel_size=kernel_size, + padding=padding)) + self.box_convs.append(box_conv) + + score_conv_name = "scores{}".format(i) + score_conv = self.add_sublayer( + score_conv_name, + nn.Conv2D( + in_channels=self.in_channels[i], + out_channels=num_prior * self.num_classes, + kernel_size=kernel_size, + padding=padding)) + self.score_convs.append(score_conv) + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + def forward(self, feats, image, gt_bbox=None, gt_class=None): + box_preds = [] + cls_scores = [] + prior_boxes = [] + for feat, box_conv, score_conv in zip(feats, self.box_convs, + self.score_convs): + box_pred = box_conv(feat) + box_pred = paddle.transpose(box_pred, [0, 2, 3, 1]) + box_pred = paddle.reshape(box_pred, [0, -1, 4]) + box_preds.append(box_pred) + + cls_score = score_conv(feat) + cls_score = paddle.transpose(cls_score, [0, 2, 3, 1]) + cls_score = paddle.reshape(cls_score, [0, -1, self.num_classes]) + cls_scores.append(cls_score) + + prior_boxes = self.anchor_generator(feats, image) + + if self.training: + return self.get_loss(box_preds, cls_scores, gt_bbox, gt_class, + prior_boxes) + else: + return (box_preds, cls_scores), prior_boxes + + def get_loss(self, boxes, scores, gt_bbox, gt_class, prior_boxes): + return self.loss(boxes, scores, gt_bbox, gt_class, prior_boxes) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/fcos_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/fcos_head.py new file mode 100644 index 000000000..1d61feed6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/fcos_head.py @@ -0,0 +1,258 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Normal, Constant + +from ppdet.core.workspace import register +from ppdet.modeling.layers import ConvNormLayer + + +class ScaleReg(nn.Layer): + """ + Parameter for scaling the regression outputs. + """ + + def __init__(self): + super(ScaleReg, self).__init__() + self.scale_reg = self.create_parameter( + shape=[1], + attr=ParamAttr(initializer=Constant(value=1.)), + dtype="float32") + + def forward(self, inputs): + out = inputs * self.scale_reg + return out + + +@register +class FCOSFeat(nn.Layer): + """ + FCOSFeat of FCOS + + Args: + feat_in (int): The channel number of input Tensor. + feat_out (int): The channel number of output Tensor. + num_convs (int): The convolution number of the FCOSFeat. + norm_type (str): Normalization type, 'bn'/'sync_bn'/'gn'. + use_dcn (bool): Whether to use dcn in tower or not. + """ + + def __init__(self, + feat_in=256, + feat_out=256, + num_convs=4, + norm_type='bn', + use_dcn=False): + super(FCOSFeat, self).__init__() + self.num_convs = num_convs + self.norm_type = norm_type + self.cls_subnet_convs = [] + self.reg_subnet_convs = [] + for i in range(self.num_convs): + in_c = feat_in if i == 0 else feat_out + + cls_conv_name = 'fcos_head_cls_tower_conv_{}'.format(i) + cls_conv = self.add_sublayer( + cls_conv_name, + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=3, + stride=1, + norm_type=norm_type, + use_dcn=use_dcn, + bias_on=True, + lr_scale=2.)) + self.cls_subnet_convs.append(cls_conv) + + reg_conv_name = 'fcos_head_reg_tower_conv_{}'.format(i) + reg_conv = self.add_sublayer( + reg_conv_name, + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=3, + stride=1, + norm_type=norm_type, + use_dcn=use_dcn, + bias_on=True, + lr_scale=2.)) + self.reg_subnet_convs.append(reg_conv) + + def forward(self, fpn_feat): + cls_feat = fpn_feat + reg_feat = fpn_feat + for i in range(self.num_convs): + cls_feat = F.relu(self.cls_subnet_convs[i](cls_feat)) + reg_feat = F.relu(self.reg_subnet_convs[i](reg_feat)) + return cls_feat, reg_feat + + +@register +class FCOSHead(nn.Layer): + """ + FCOSHead + Args: + fcos_feat (object): Instance of 'FCOSFeat' + num_classes (int): Number of classes + fpn_stride (list): The stride of each FPN Layer + prior_prob (float): Used to set the bias init for the class prediction layer + fcos_loss (object): Instance of 'FCOSLoss' + norm_reg_targets (bool): Normalization the regression target if true + centerness_on_reg (bool): The prediction of centerness on regression or clssification branch + """ + __inject__ = ['fcos_feat', 'fcos_loss'] + __shared__ = ['num_classes'] + + def __init__(self, + fcos_feat, + num_classes=80, + fpn_stride=[8, 16, 32, 64, 128], + prior_prob=0.01, + fcos_loss='FCOSLoss', + norm_reg_targets=True, + centerness_on_reg=True): + super(FCOSHead, self).__init__() + self.fcos_feat = fcos_feat + self.num_classes = num_classes + self.fpn_stride = fpn_stride + self.prior_prob = prior_prob + self.fcos_loss = fcos_loss + self.norm_reg_targets = norm_reg_targets + self.centerness_on_reg = centerness_on_reg + + conv_cls_name = "fcos_head_cls" + bias_init_value = -math.log((1 - self.prior_prob) / self.prior_prob) + self.fcos_head_cls = self.add_sublayer( + conv_cls_name, + nn.Conv2D( + in_channels=256, + out_channels=self.num_classes, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr( + initializer=Constant(value=bias_init_value)))) + + conv_reg_name = "fcos_head_reg" + self.fcos_head_reg = self.add_sublayer( + conv_reg_name, + nn.Conv2D( + in_channels=256, + out_channels=4, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + + conv_centerness_name = "fcos_head_centerness" + self.fcos_head_centerness = self.add_sublayer( + conv_centerness_name, + nn.Conv2D( + in_channels=256, + out_channels=1, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + + self.scales_regs = [] + for i in range(len(self.fpn_stride)): + lvl = int(math.log(int(self.fpn_stride[i]), 2)) + feat_name = 'p{}_feat'.format(lvl) + scale_reg = self.add_sublayer(feat_name, ScaleReg()) + self.scales_regs.append(scale_reg) + + def _compute_locations_by_level(self, fpn_stride, feature): + """ + Compute locations of anchor points of each FPN layer + Args: + fpn_stride (int): The stride of current FPN feature map + feature (Tensor): Tensor of current FPN feature map + Return: + Anchor points locations of current FPN feature map + """ + shape_fm = paddle.shape(feature) + shape_fm.stop_gradient = True + h, w = shape_fm[2], shape_fm[3] + shift_x = paddle.arange(0, w * fpn_stride, fpn_stride) + shift_y = paddle.arange(0, h * fpn_stride, fpn_stride) + shift_x = paddle.unsqueeze(shift_x, axis=0) + shift_y = paddle.unsqueeze(shift_y, axis=1) + shift_x = paddle.expand(shift_x, shape=[h, w]) + shift_y = paddle.expand(shift_y, shape=[h, w]) + shift_x.stop_gradient = True + shift_y.stop_gradient = True + shift_x = paddle.reshape(shift_x, shape=[-1]) + shift_y = paddle.reshape(shift_y, shape=[-1]) + location = paddle.stack( + [shift_x, shift_y], axis=-1) + float(fpn_stride) / 2 + location.stop_gradient = True + return location + + def forward(self, fpn_feats, is_training): + assert len(fpn_feats) == len( + self.fpn_stride + ), "The size of fpn_feats is not equal to size of fpn_stride" + cls_logits_list = [] + bboxes_reg_list = [] + centerness_list = [] + for scale_reg, fpn_stride, fpn_feat in zip(self.scales_regs, + self.fpn_stride, fpn_feats): + fcos_cls_feat, fcos_reg_feat = self.fcos_feat(fpn_feat) + cls_logits = self.fcos_head_cls(fcos_cls_feat) + bbox_reg = scale_reg(self.fcos_head_reg(fcos_reg_feat)) + if self.centerness_on_reg: + centerness = self.fcos_head_centerness(fcos_reg_feat) + else: + centerness = self.fcos_head_centerness(fcos_cls_feat) + if self.norm_reg_targets: + bbox_reg = F.relu(bbox_reg) + if not is_training: + bbox_reg = bbox_reg * fpn_stride + else: + bbox_reg = paddle.exp(bbox_reg) + cls_logits_list.append(cls_logits) + bboxes_reg_list.append(bbox_reg) + centerness_list.append(centerness) + + if not is_training: + locations_list = [] + for fpn_stride, feature in zip(self.fpn_stride, fpn_feats): + location = self._compute_locations_by_level(fpn_stride, feature) + locations_list.append(location) + + return locations_list, cls_logits_list, bboxes_reg_list, centerness_list + else: + return cls_logits_list, bboxes_reg_list, centerness_list + + def get_loss(self, fcos_head_outs, tag_labels, tag_bboxes, tag_centerness): + cls_logits, bboxes_reg, centerness = fcos_head_outs + return self.fcos_loss(cls_logits, bboxes_reg, centerness, tag_labels, + tag_bboxes, tag_centerness) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/gfl_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/gfl_head.py new file mode 100644 index 000000000..17e87a4ef --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/gfl_head.py @@ -0,0 +1,480 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/dense_heads/gfl_head.py + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Normal, Constant + +from ppdet.core.workspace import register +from ppdet.modeling.layers import ConvNormLayer +from ppdet.modeling.bbox_utils import distance2bbox, bbox2distance +from ppdet.data.transform.atss_assigner import bbox_overlaps + + +class ScaleReg(nn.Layer): + """ + Parameter for scaling the regression outputs. + """ + + def __init__(self): + super(ScaleReg, self).__init__() + self.scale_reg = self.create_parameter( + shape=[1], + attr=ParamAttr(initializer=Constant(value=1.)), + dtype="float32") + + def forward(self, inputs): + out = inputs * self.scale_reg + return out + + +class Integral(nn.Layer): + """A fixed layer for calculating integral result from distribution. + This layer calculates the target location by :math: `sum{P(y_i) * y_i}`, + P(y_i) denotes the softmax vector that represents the discrete distribution + y_i denotes the discrete set, usually {0, 1, 2, ..., reg_max} + + Args: + reg_max (int): The maximal value of the discrete set. Default: 16. You + may want to reset it according to your new dataset or related + settings. + """ + + def __init__(self, reg_max=16): + super(Integral, self).__init__() + self.reg_max = reg_max + self.register_buffer('project', + paddle.linspace(0, self.reg_max, self.reg_max + 1)) + + def forward(self, x): + """Forward feature from the regression head to get integral result of + bounding box location. + Args: + x (Tensor): Features of the regression head, shape (N, 4*(n+1)), + n is self.reg_max. + Returns: + x (Tensor): Integral result of box locations, i.e., distance + offsets from the box center in four directions, shape (N, 4). + """ + x = F.softmax(x.reshape([-1, self.reg_max + 1]), axis=1) + x = F.linear(x, self.project).reshape([-1, 4]) + return x + + +@register +class DGQP(nn.Layer): + """Distribution-Guided Quality Predictor of GFocal head + + Args: + reg_topk (int): top-k statistics of distribution to guide LQE + reg_channels (int): hidden layer unit to generate LQE + add_mean (bool): Whether to calculate the mean of top-k statistics + """ + + def __init__(self, reg_topk=4, reg_channels=64, add_mean=True): + super(DGQP, self).__init__() + self.reg_topk = reg_topk + self.reg_channels = reg_channels + self.add_mean = add_mean + self.total_dim = reg_topk + if add_mean: + self.total_dim += 1 + self.reg_conv1 = self.add_sublayer( + 'dgqp_reg_conv1', + nn.Conv2D( + in_channels=4 * self.total_dim, + out_channels=self.reg_channels, + kernel_size=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + self.reg_conv2 = self.add_sublayer( + 'dgqp_reg_conv2', + nn.Conv2D( + in_channels=self.reg_channels, + out_channels=1, + kernel_size=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + + def forward(self, x): + """Forward feature from the regression head to get integral result of + bounding box location. + Args: + x (Tensor): Features of the regression head, shape (N, 4*(n+1)), + n is self.reg_max. + Returns: + x (Tensor): Integral result of box locations, i.e., distance + offsets from the box center in four directions, shape (N, 4). + """ + N, _, H, W = x.shape[:] + prob = F.softmax(x.reshape([N, 4, -1, H, W]), axis=2) + prob_topk, _ = prob.topk(self.reg_topk, axis=2) + if self.add_mean: + stat = paddle.concat( + [prob_topk, prob_topk.mean( + axis=2, keepdim=True)], axis=2) + else: + stat = prob_topk + y = F.relu(self.reg_conv1(stat.reshape([N, -1, H, W]))) + y = F.sigmoid(self.reg_conv2(y)) + return y + + +@register +class GFLHead(nn.Layer): + """ + GFLHead + Args: + conv_feat (object): Instance of 'FCOSFeat' + num_classes (int): Number of classes + fpn_stride (list): The stride of each FPN Layer + prior_prob (float): Used to set the bias init for the class prediction layer + loss_class (object): Instance of QualityFocalLoss. + loss_dfl (object): Instance of DistributionFocalLoss. + loss_bbox (object): Instance of bbox loss. + reg_max: Max value of integral set :math: `{0, ..., reg_max}` + n QFL setting. Default: 16. + """ + __inject__ = [ + 'conv_feat', 'dgqp_module', 'loss_class', 'loss_dfl', 'loss_bbox', 'nms' + ] + __shared__ = ['num_classes'] + + def __init__(self, + conv_feat='FCOSFeat', + dgqp_module=None, + num_classes=80, + fpn_stride=[8, 16, 32, 64, 128], + prior_prob=0.01, + loss_class='QualityFocalLoss', + loss_dfl='DistributionFocalLoss', + loss_bbox='GIoULoss', + reg_max=16, + feat_in_chan=256, + nms=None, + nms_pre=1000, + cell_offset=0): + super(GFLHead, self).__init__() + self.conv_feat = conv_feat + self.dgqp_module = dgqp_module + self.num_classes = num_classes + self.fpn_stride = fpn_stride + self.prior_prob = prior_prob + self.loss_qfl = loss_class + self.loss_dfl = loss_dfl + self.loss_bbox = loss_bbox + self.reg_max = reg_max + self.feat_in_chan = feat_in_chan + self.nms = nms + self.nms_pre = nms_pre + self.cell_offset = cell_offset + self.use_sigmoid = self.loss_qfl.use_sigmoid + if self.use_sigmoid: + self.cls_out_channels = self.num_classes + else: + self.cls_out_channels = self.num_classes + 1 + + conv_cls_name = "gfl_head_cls" + bias_init_value = -math.log((1 - self.prior_prob) / self.prior_prob) + self.gfl_head_cls = self.add_sublayer( + conv_cls_name, + nn.Conv2D( + in_channels=self.feat_in_chan, + out_channels=self.cls_out_channels, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr( + initializer=Constant(value=bias_init_value)))) + + conv_reg_name = "gfl_head_reg" + self.gfl_head_reg = self.add_sublayer( + conv_reg_name, + nn.Conv2D( + in_channels=self.feat_in_chan, + out_channels=4 * (self.reg_max + 1), + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + + self.scales_regs = [] + for i in range(len(self.fpn_stride)): + lvl = int(math.log(int(self.fpn_stride[i]), 2)) + feat_name = 'p{}_feat'.format(lvl) + scale_reg = self.add_sublayer(feat_name, ScaleReg()) + self.scales_regs.append(scale_reg) + + self.distribution_project = Integral(self.reg_max) + + def forward(self, fpn_feats): + assert len(fpn_feats) == len( + self.fpn_stride + ), "The size of fpn_feats is not equal to size of fpn_stride" + cls_logits_list = [] + bboxes_reg_list = [] + for scale_reg, fpn_feat in zip(self.scales_regs, fpn_feats): + conv_cls_feat, conv_reg_feat = self.conv_feat(fpn_feat) + cls_logits = self.gfl_head_cls(conv_cls_feat) + bbox_reg = scale_reg(self.gfl_head_reg(conv_reg_feat)) + if self.dgqp_module: + quality_score = self.dgqp_module(bbox_reg) + cls_logits = F.sigmoid(cls_logits) * quality_score + if not self.training: + cls_logits = F.sigmoid(cls_logits.transpose([0, 2, 3, 1])) + bbox_reg = bbox_reg.transpose([0, 2, 3, 1]) + cls_logits_list.append(cls_logits) + bboxes_reg_list.append(bbox_reg) + + return (cls_logits_list, bboxes_reg_list) + + def _images_to_levels(self, target, num_level_anchors): + """ + Convert targets by image to targets by feature level. + """ + level_targets = [] + start = 0 + for n in num_level_anchors: + end = start + n + level_targets.append(target[:, start:end].squeeze(0)) + start = end + return level_targets + + def _grid_cells_to_center(self, grid_cells): + """ + Get center location of each gird cell + Args: + grid_cells: grid cells of a feature map + Returns: + center points + """ + cells_cx = (grid_cells[:, 2] + grid_cells[:, 0]) / 2 + cells_cy = (grid_cells[:, 3] + grid_cells[:, 1]) / 2 + return paddle.stack([cells_cx, cells_cy], axis=-1) + + def get_loss(self, gfl_head_outs, gt_meta): + cls_logits, bboxes_reg = gfl_head_outs + num_level_anchors = [ + featmap.shape[-2] * featmap.shape[-1] for featmap in cls_logits + ] + grid_cells_list = self._images_to_levels(gt_meta['grid_cells'], + num_level_anchors) + labels_list = self._images_to_levels(gt_meta['labels'], + num_level_anchors) + label_weights_list = self._images_to_levels(gt_meta['label_weights'], + num_level_anchors) + bbox_targets_list = self._images_to_levels(gt_meta['bbox_targets'], + num_level_anchors) + num_total_pos = sum(gt_meta['pos_num']) + try: + num_total_pos = paddle.distributed.all_reduce(num_total_pos.clone( + )) / paddle.distributed.get_world_size() + except: + num_total_pos = max(num_total_pos, 1) + + loss_bbox_list, loss_dfl_list, loss_qfl_list, avg_factor = [], [], [], [] + for cls_score, bbox_pred, grid_cells, labels, label_weights, bbox_targets, stride in zip( + cls_logits, bboxes_reg, grid_cells_list, labels_list, + label_weights_list, bbox_targets_list, self.fpn_stride): + grid_cells = grid_cells.reshape([-1, 4]) + cls_score = cls_score.transpose([0, 2, 3, 1]).reshape( + [-1, self.cls_out_channels]) + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]).reshape( + [-1, 4 * (self.reg_max + 1)]) + bbox_targets = bbox_targets.reshape([-1, 4]) + labels = labels.reshape([-1]) + label_weights = label_weights.reshape([-1]) + + bg_class_ind = self.num_classes + pos_inds = paddle.nonzero( + paddle.logical_and((labels >= 0), (labels < bg_class_ind)), + as_tuple=False).squeeze(1) + score = np.zeros(labels.shape) + if len(pos_inds) > 0: + pos_bbox_targets = paddle.gather(bbox_targets, pos_inds, axis=0) + pos_bbox_pred = paddle.gather(bbox_pred, pos_inds, axis=0) + pos_grid_cells = paddle.gather(grid_cells, pos_inds, axis=0) + pos_grid_cell_centers = self._grid_cells_to_center( + pos_grid_cells) / stride + + weight_targets = F.sigmoid(cls_score.detach()) + weight_targets = paddle.gather( + weight_targets.max(axis=1, keepdim=True), pos_inds, axis=0) + pos_bbox_pred_corners = self.distribution_project(pos_bbox_pred) + pos_decode_bbox_pred = distance2bbox(pos_grid_cell_centers, + pos_bbox_pred_corners) + pos_decode_bbox_targets = pos_bbox_targets / stride + bbox_iou = bbox_overlaps( + pos_decode_bbox_pred.detach().numpy(), + pos_decode_bbox_targets.detach().numpy(), + is_aligned=True) + score[pos_inds.numpy()] = bbox_iou + pred_corners = pos_bbox_pred.reshape([-1, self.reg_max + 1]) + target_corners = bbox2distance(pos_grid_cell_centers, + pos_decode_bbox_targets, + self.reg_max).reshape([-1]) + # regression loss + loss_bbox = paddle.sum( + self.loss_bbox(pos_decode_bbox_pred, + pos_decode_bbox_targets) * weight_targets) + + # dfl loss + loss_dfl = self.loss_dfl( + pred_corners, + target_corners, + weight=weight_targets.expand([-1, 4]).reshape([-1]), + avg_factor=4.0) + else: + loss_bbox = bbox_pred.sum() * 0 + loss_dfl = bbox_pred.sum() * 0 + weight_targets = paddle.to_tensor([0], dtype='float32') + + # qfl loss + score = paddle.to_tensor(score) + loss_qfl = self.loss_qfl( + cls_score, (labels, score), + weight=label_weights, + avg_factor=num_total_pos) + loss_bbox_list.append(loss_bbox) + loss_dfl_list.append(loss_dfl) + loss_qfl_list.append(loss_qfl) + avg_factor.append(weight_targets.sum()) + + avg_factor = sum(avg_factor) + try: + avg_factor = paddle.distributed.all_reduce(avg_factor.clone()) + avg_factor = paddle.clip( + avg_factor / paddle.distributed.get_world_size(), min=1) + except: + avg_factor = max(avg_factor.item(), 1) + if avg_factor <= 0: + loss_qfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + loss_bbox = paddle.to_tensor( + 0, dtype='float32', stop_gradient=False) + loss_dfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + else: + losses_bbox = list(map(lambda x: x / avg_factor, loss_bbox_list)) + losses_dfl = list(map(lambda x: x / avg_factor, loss_dfl_list)) + loss_qfl = sum(loss_qfl_list) + loss_bbox = sum(losses_bbox) + loss_dfl = sum(losses_dfl) + + loss_states = dict( + loss_qfl=loss_qfl, loss_bbox=loss_bbox, loss_dfl=loss_dfl) + + return loss_states + + def get_single_level_center_point(self, featmap_size, stride, + cell_offset=0): + """ + Generate pixel centers of a single stage feature map. + Args: + featmap_size: height and width of the feature map + stride: down sample stride of the feature map + Returns: + y and x of the center points + """ + h, w = featmap_size + x_range = (paddle.arange(w, dtype='float32') + cell_offset) * stride + y_range = (paddle.arange(h, dtype='float32') + cell_offset) * stride + y, x = paddle.meshgrid(y_range, x_range) + y = y.flatten() + x = x.flatten() + return y, x + + def get_bboxes_single(self, + cls_scores, + bbox_preds, + img_shape, + scale_factor, + rescale=True, + cell_offset=0): + assert len(cls_scores) == len(bbox_preds) + mlvl_bboxes = [] + mlvl_scores = [] + for stride, cls_score, bbox_pred in zip(self.fpn_stride, cls_scores, + bbox_preds): + featmap_size = [ + paddle.shape(cls_score)[0], paddle.shape(cls_score)[1] + ] + y, x = self.get_single_level_center_point( + featmap_size, stride, cell_offset=cell_offset) + center_points = paddle.stack([x, y], axis=-1) + scores = cls_score.reshape([-1, self.cls_out_channels]) + bbox_pred = self.distribution_project(bbox_pred) * stride + + if scores.shape[0] > self.nms_pre: + max_scores = scores.max(axis=1) + _, topk_inds = max_scores.topk(self.nms_pre) + center_points = center_points.gather(topk_inds) + bbox_pred = bbox_pred.gather(topk_inds) + scores = scores.gather(topk_inds) + + bboxes = distance2bbox( + center_points, bbox_pred, max_shape=img_shape) + mlvl_bboxes.append(bboxes) + mlvl_scores.append(scores) + mlvl_bboxes = paddle.concat(mlvl_bboxes) + if rescale: + # [h_scale, w_scale] to [w_scale, h_scale, w_scale, h_scale] + im_scale = paddle.concat([scale_factor[::-1], scale_factor[::-1]]) + mlvl_bboxes /= im_scale + mlvl_scores = paddle.concat(mlvl_scores) + mlvl_scores = mlvl_scores.transpose([1, 0]) + return mlvl_bboxes, mlvl_scores + + def decode(self, cls_scores, bbox_preds, im_shape, scale_factor, + cell_offset): + batch_bboxes = [] + batch_scores = [] + for img_id in range(cls_scores[0].shape[0]): + num_levels = len(cls_scores) + cls_score_list = [cls_scores[i][img_id] for i in range(num_levels)] + bbox_pred_list = [bbox_preds[i][img_id] for i in range(num_levels)] + bboxes, scores = self.get_bboxes_single( + cls_score_list, + bbox_pred_list, + im_shape[img_id], + scale_factor[img_id], + cell_offset=cell_offset) + batch_bboxes.append(bboxes) + batch_scores.append(scores) + batch_bboxes = paddle.stack(batch_bboxes, axis=0) + batch_scores = paddle.stack(batch_scores, axis=0) + + return batch_bboxes, batch_scores + + def post_process(self, gfl_head_outs, im_shape, scale_factor): + cls_scores, bboxes_reg = gfl_head_outs + bboxes, score = self.decode(cls_scores, bboxes_reg, im_shape, + scale_factor, self.cell_offset) + bbox_pred, bbox_num, _ = self.nms(bboxes, score) + return bbox_pred, bbox_num diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/keypoint_hrhrnet_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/keypoint_hrhrnet_head.py new file mode 100644 index 000000000..869b1816e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/keypoint_hrhrnet_head.py @@ -0,0 +1,108 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register +from .. import layers as L +from ..backbones.hrnet import BasicBlock + + +@register +class HrHRNetHead(nn.Layer): + __inject__ = ['loss'] + + def __init__(self, num_joints, loss='HrHRNetLoss', swahr=False, width=32): + """ + Head for HigherHRNet network + + Args: + num_joints (int): number of keypoints + hrloss (object): HrHRNetLoss instance + swahr (bool): whether to use swahr + width (int): hrnet channel width + """ + super(HrHRNetHead, self).__init__() + self.loss = loss + + self.num_joints = num_joints + num_featout1 = num_joints * 2 + num_featout2 = num_joints + self.swahr = swahr + self.conv1 = L.Conv2d(width, num_featout1, 1, 1, 0, bias=True) + self.conv2 = L.Conv2d(width, num_featout2, 1, 1, 0, bias=True) + self.deconv = nn.Sequential( + L.ConvTranspose2d( + num_featout1 + width, width, 4, 2, 1, 0, bias=False), + L.BatchNorm2d(width), + L.ReLU()) + self.blocks = nn.Sequential(*(BasicBlock( + num_channels=width, + num_filters=width, + has_se=False, + freeze_norm=False, + name='HrHRNetHead_{}'.format(i)) for i in range(4))) + + self.interpolate = L.Upsample(2, mode='bilinear') + self.concat = L.Concat(dim=1) + if swahr: + self.scalelayer0 = nn.Sequential( + L.Conv2d( + width, num_joints, 1, 1, 0, bias=True), + L.BatchNorm2d(num_joints), + L.ReLU(), + L.Conv2d( + num_joints, + num_joints, + 9, + 1, + 4, + groups=num_joints, + bias=True)) + self.scalelayer1 = nn.Sequential( + L.Conv2d( + width, num_joints, 1, 1, 0, bias=True), + L.BatchNorm2d(num_joints), + L.ReLU(), + L.Conv2d( + num_joints, + num_joints, + 9, + 1, + 4, + groups=num_joints, + bias=True)) + + def forward(self, feats, targets=None): + x1 = feats[0] + xo1 = self.conv1(x1) + x2 = self.blocks(self.deconv(self.concat((x1, xo1)))) + xo2 = self.conv2(x2) + num_joints = self.num_joints + if self.training: + heatmap1, tagmap = paddle.split(xo1, 2, axis=1) + if self.swahr: + so1 = self.scalelayer0(x1) + so2 = self.scalelayer1(x2) + hrhrnet_outputs = ([heatmap1, so1], [xo2, so2], tagmap) + return self.loss(hrhrnet_outputs, targets) + else: + hrhrnet_outputs = (heatmap1, xo2, tagmap) + return self.loss(hrhrnet_outputs, targets) + + # averaged heatmap, upsampled tagmap + upsampled = self.interpolate(xo1) + avg = (upsampled[:, :num_joints] + xo2[:, :num_joints]) / 2 + return avg, upsampled[:, num_joints:] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/mask_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/mask_head.py new file mode 100644 index 000000000..bfce2dc5b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/mask_head.py @@ -0,0 +1,250 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import KaimingNormal + +from ppdet.core.workspace import register, create +from ppdet.modeling.layers import ConvNormLayer +from .roi_extractor import RoIAlign + + +@register +class MaskFeat(nn.Layer): + """ + Feature extraction in Mask head + + Args: + in_channel (int): Input channels + out_channel (int): Output channels + num_convs (int): The number of conv layers, default 4 + norm_type (string | None): Norm type, bn, gn, sync_bn are available, + default None + """ + + def __init__(self, + in_channel=256, + out_channel=256, + num_convs=4, + norm_type=None): + super(MaskFeat, self).__init__() + self.num_convs = num_convs + self.in_channel = in_channel + self.out_channel = out_channel + self.norm_type = norm_type + fan_conv = out_channel * 3 * 3 + fan_deconv = out_channel * 2 * 2 + + mask_conv = nn.Sequential() + if norm_type == 'gn': + for i in range(self.num_convs): + conv_name = 'mask_inter_feat_{}'.format(i + 1) + mask_conv.add_sublayer( + conv_name, + ConvNormLayer( + ch_in=in_channel if i == 0 else out_channel, + ch_out=out_channel, + filter_size=3, + stride=1, + norm_type=self.norm_type, + initializer=KaimingNormal(fan_in=fan_conv), + skip_quant=True)) + mask_conv.add_sublayer(conv_name + 'act', nn.ReLU()) + else: + for i in range(self.num_convs): + conv_name = 'mask_inter_feat_{}'.format(i + 1) + conv = nn.Conv2D( + in_channels=in_channel if i == 0 else out_channel, + out_channels=out_channel, + kernel_size=3, + padding=1, + weight_attr=paddle.ParamAttr( + initializer=KaimingNormal(fan_in=fan_conv))) + conv.skip_quant = True + mask_conv.add_sublayer(conv_name, conv) + mask_conv.add_sublayer(conv_name + 'act', nn.ReLU()) + mask_conv.add_sublayer( + 'conv5_mask', + nn.Conv2DTranspose( + in_channels=self.in_channel, + out_channels=self.out_channel, + kernel_size=2, + stride=2, + weight_attr=paddle.ParamAttr( + initializer=KaimingNormal(fan_in=fan_deconv)))) + mask_conv.add_sublayer('conv5_mask' + 'act', nn.ReLU()) + self.upsample = mask_conv + + @classmethod + def from_config(cls, cfg, input_shape): + if isinstance(input_shape, (list, tuple)): + input_shape = input_shape[0] + return {'in_channel': input_shape.channels, } + + def out_channels(self): + return self.out_channel + + def forward(self, feats): + return self.upsample(feats) + + +@register +class MaskHead(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['mask_assigner'] + """ + RCNN mask head + + Args: + head (nn.Layer): Extract feature in mask head + roi_extractor (object): The module of RoI Extractor + mask_assigner (object): The module of Mask Assigner, + label and sample the mask + num_classes (int): The number of classes + share_bbox_feat (bool): Whether to share the feature from bbox head, + default false + """ + + def __init__(self, + head, + roi_extractor=RoIAlign().__dict__, + mask_assigner='MaskAssigner', + num_classes=80, + share_bbox_feat=False): + super(MaskHead, self).__init__() + self.num_classes = num_classes + + self.roi_extractor = roi_extractor + if isinstance(roi_extractor, dict): + self.roi_extractor = RoIAlign(**roi_extractor) + self.head = head + self.in_channels = head.out_channels() + self.mask_assigner = mask_assigner + self.share_bbox_feat = share_bbox_feat + self.bbox_head = None + + self.mask_fcn_logits = nn.Conv2D( + in_channels=self.in_channels, + out_channels=self.num_classes, + kernel_size=1, + weight_attr=paddle.ParamAttr(initializer=KaimingNormal( + fan_in=self.num_classes))) + self.mask_fcn_logits.skip_quant = True + + @classmethod + def from_config(cls, cfg, input_shape): + roi_pooler = cfg['roi_extractor'] + assert isinstance(roi_pooler, dict) + kwargs = RoIAlign.from_config(cfg, input_shape) + roi_pooler.update(kwargs) + kwargs = {'input_shape': input_shape} + head = create(cfg['head'], **kwargs) + return { + 'roi_extractor': roi_pooler, + 'head': head, + } + + def get_loss(self, mask_logits, mask_label, mask_target, mask_weight): + mask_label = F.one_hot(mask_label, self.num_classes).unsqueeze([2, 3]) + mask_label = paddle.expand_as(mask_label, mask_logits) + mask_label.stop_gradient = True + mask_pred = paddle.gather_nd(mask_logits, paddle.nonzero(mask_label)) + shape = mask_logits.shape + mask_pred = paddle.reshape(mask_pred, [shape[0], shape[2], shape[3]]) + + mask_target = mask_target.cast('float32') + mask_weight = mask_weight.unsqueeze([1, 2]) + loss_mask = F.binary_cross_entropy_with_logits( + mask_pred, mask_target, weight=mask_weight, reduction="mean") + return loss_mask + + def forward_train(self, body_feats, rois, rois_num, inputs, targets, + bbox_feat): + """ + body_feats (list[Tensor]): Multi-level backbone features + rois (list[Tensor]): Proposals for each batch with shape [N, 4] + rois_num (Tensor): The number of proposals for each batch + inputs (dict): ground truth info + """ + tgt_labels, _, tgt_gt_inds = targets + rois, rois_num, tgt_classes, tgt_masks, mask_index, tgt_weights = self.mask_assigner( + rois, tgt_labels, tgt_gt_inds, inputs) + + if self.share_bbox_feat: + rois_feat = paddle.gather(bbox_feat, mask_index) + else: + rois_feat = self.roi_extractor(body_feats, rois, rois_num) + mask_feat = self.head(rois_feat) + mask_logits = self.mask_fcn_logits(mask_feat) + + loss_mask = self.get_loss(mask_logits, tgt_classes, tgt_masks, + tgt_weights) + return {'loss_mask': loss_mask} + + def forward_test(self, + body_feats, + rois, + rois_num, + scale_factor, + feat_func=None): + """ + body_feats (list[Tensor]): Multi-level backbone features + rois (Tensor): Prediction from bbox head with shape [N, 6] + rois_num (Tensor): The number of prediction for each batch + scale_factor (Tensor): The scale factor from origin size to input size + """ + if rois.shape[0] == 0: + mask_out = paddle.full([1, 1, 1, 1], -1) + else: + bbox = [rois[:, 2:]] + labels = rois[:, 0].cast('int32') + rois_feat = self.roi_extractor(body_feats, bbox, rois_num) + if self.share_bbox_feat: + assert feat_func is not None + rois_feat = feat_func(rois_feat) + + mask_feat = self.head(rois_feat) + mask_logit = self.mask_fcn_logits(mask_feat) + mask_num_class = mask_logit.shape[1] + if mask_num_class == 1: + mask_out = F.sigmoid(mask_logit) + else: + num_masks = mask_logit.shape[0] + mask_out = [] + # TODO: need to optimize gather + for i in range(mask_logit.shape[0]): + pred_masks = paddle.unsqueeze( + mask_logit[i, :, :, :], axis=0) + mask = paddle.gather(pred_masks, labels[i], axis=1) + mask_out.append(mask) + mask_out = F.sigmoid(paddle.concat(mask_out)) + return mask_out + + def forward(self, + body_feats, + rois, + rois_num, + inputs, + targets=None, + bbox_feat=None, + feat_func=None): + if self.training: + return self.forward_train(body_feats, rois, rois_num, inputs, + targets, bbox_feat) + else: + im_scale = inputs['scale_factor'] + return self.forward_test(body_feats, rois, rois_num, im_scale, + feat_func) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/pico_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/pico_head.py new file mode 100644 index 000000000..7cfd24c3c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/pico_head.py @@ -0,0 +1,277 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Normal, Constant + +from ppdet.core.workspace import register +from ppdet.modeling.layers import ConvNormLayer +from .simota_head import OTAVFLHead + + +@register +class PicoFeat(nn.Layer): + """ + PicoFeat of PicoDet + + Args: + feat_in (int): The channel number of input Tensor. + feat_out (int): The channel number of output Tensor. + num_convs (int): The convolution number of the LiteGFLFeat. + norm_type (str): Normalization type, 'bn'/'sync_bn'/'gn'. + """ + + def __init__(self, + feat_in=256, + feat_out=96, + num_fpn_stride=3, + num_convs=2, + norm_type='bn', + share_cls_reg=False, + act='hard_swish'): + super(PicoFeat, self).__init__() + self.num_convs = num_convs + self.norm_type = norm_type + self.share_cls_reg = share_cls_reg + self.act = act + self.cls_convs = [] + self.reg_convs = [] + for stage_idx in range(num_fpn_stride): + cls_subnet_convs = [] + reg_subnet_convs = [] + for i in range(self.num_convs): + in_c = feat_in if i == 0 else feat_out + cls_conv_dw = self.add_sublayer( + 'cls_conv_dw{}.{}'.format(stage_idx, i), + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=5, + stride=1, + groups=feat_out, + norm_type=norm_type, + bias_on=False, + lr_scale=2.)) + cls_subnet_convs.append(cls_conv_dw) + cls_conv_pw = self.add_sublayer( + 'cls_conv_pw{}.{}'.format(stage_idx, i), + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=1, + stride=1, + norm_type=norm_type, + bias_on=False, + lr_scale=2.)) + cls_subnet_convs.append(cls_conv_pw) + + if not self.share_cls_reg: + reg_conv_dw = self.add_sublayer( + 'reg_conv_dw{}.{}'.format(stage_idx, i), + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=5, + stride=1, + groups=feat_out, + norm_type=norm_type, + bias_on=False, + lr_scale=2.)) + reg_subnet_convs.append(reg_conv_dw) + reg_conv_pw = self.add_sublayer( + 'reg_conv_pw{}.{}'.format(stage_idx, i), + ConvNormLayer( + ch_in=in_c, + ch_out=feat_out, + filter_size=1, + stride=1, + norm_type=norm_type, + bias_on=False, + lr_scale=2.)) + reg_subnet_convs.append(reg_conv_pw) + self.cls_convs.append(cls_subnet_convs) + self.reg_convs.append(reg_subnet_convs) + + def act_func(self, x): + if self.act == "leaky_relu": + x = F.leaky_relu(x) + elif self.act == "hard_swish": + x = F.hardswish(x) + return x + + def forward(self, fpn_feat, stage_idx): + assert stage_idx < len(self.cls_convs) + cls_feat = fpn_feat + reg_feat = fpn_feat + for i in range(len(self.cls_convs[stage_idx])): + cls_feat = self.act_func(self.cls_convs[stage_idx][i](cls_feat)) + if not self.share_cls_reg: + reg_feat = self.act_func(self.reg_convs[stage_idx][i](reg_feat)) + return cls_feat, reg_feat + + +@register +class PicoHead(OTAVFLHead): + """ + PicoHead + Args: + conv_feat (object): Instance of 'PicoFeat' + num_classes (int): Number of classes + fpn_stride (list): The stride of each FPN Layer + prior_prob (float): Used to set the bias init for the class prediction layer + loss_class (object): Instance of VariFocalLoss. + loss_dfl (object): Instance of DistributionFocalLoss. + loss_bbox (object): Instance of bbox loss. + assigner (object): Instance of label assigner. + reg_max: Max value of integral set :math: `{0, ..., reg_max}` + n QFL setting. Default: 7. + """ + __inject__ = [ + 'conv_feat', 'dgqp_module', 'loss_class', 'loss_dfl', 'loss_bbox', + 'assigner', 'nms' + ] + __shared__ = ['num_classes'] + + def __init__(self, + conv_feat='PicoFeat', + dgqp_module=None, + num_classes=80, + fpn_stride=[8, 16, 32], + prior_prob=0.01, + loss_class='VariFocalLoss', + loss_dfl='DistributionFocalLoss', + loss_bbox='GIoULoss', + assigner='SimOTAAssigner', + reg_max=16, + feat_in_chan=96, + nms=None, + nms_pre=1000, + cell_offset=0): + super(PicoHead, self).__init__( + conv_feat=conv_feat, + dgqp_module=dgqp_module, + num_classes=num_classes, + fpn_stride=fpn_stride, + prior_prob=prior_prob, + loss_class=loss_class, + loss_dfl=loss_dfl, + loss_bbox=loss_bbox, + assigner=assigner, + reg_max=reg_max, + feat_in_chan=feat_in_chan, + nms=nms, + nms_pre=nms_pre, + cell_offset=cell_offset) + self.conv_feat = conv_feat + self.num_classes = num_classes + self.fpn_stride = fpn_stride + self.prior_prob = prior_prob + self.loss_vfl = loss_class + self.loss_dfl = loss_dfl + self.loss_bbox = loss_bbox + self.assigner = assigner + self.reg_max = reg_max + self.feat_in_chan = feat_in_chan + self.nms = nms + self.nms_pre = nms_pre + self.cell_offset = cell_offset + + self.use_sigmoid = self.loss_vfl.use_sigmoid + if self.use_sigmoid: + self.cls_out_channels = self.num_classes + else: + self.cls_out_channels = self.num_classes + 1 + bias_init_value = -math.log((1 - self.prior_prob) / self.prior_prob) + # Clear the super class initialization + self.gfl_head_cls = None + self.gfl_head_reg = None + self.scales_regs = None + + self.head_cls_list = [] + self.head_reg_list = [] + for i in range(len(fpn_stride)): + head_cls = self.add_sublayer( + "head_cls" + str(i), + nn.Conv2D( + in_channels=self.feat_in_chan, + out_channels=self.cls_out_channels + 4 * (self.reg_max + 1) + if self.conv_feat.share_cls_reg else self.cls_out_channels, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr( + initializer=Constant(value=bias_init_value)))) + self.head_cls_list.append(head_cls) + if not self.conv_feat.share_cls_reg: + head_reg = self.add_sublayer( + "head_reg" + str(i), + nn.Conv2D( + in_channels=self.feat_in_chan, + out_channels=4 * (self.reg_max + 1), + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant(value=0)))) + self.head_reg_list.append(head_reg) + + def forward(self, fpn_feats, deploy=False): + assert len(fpn_feats) == len( + self.fpn_stride + ), "The size of fpn_feats is not equal to size of fpn_stride" + cls_logits_list = [] + bboxes_reg_list = [] + for i, fpn_feat in enumerate(fpn_feats): + conv_cls_feat, conv_reg_feat = self.conv_feat(fpn_feat, i) + if self.conv_feat.share_cls_reg: + cls_logits = self.head_cls_list[i](conv_cls_feat) + cls_score, bbox_pred = paddle.split( + cls_logits, + [self.cls_out_channels, 4 * (self.reg_max + 1)], + axis=1) + else: + cls_score = self.head_cls_list[i](conv_cls_feat) + bbox_pred = self.head_reg_list[i](conv_reg_feat) + + if self.dgqp_module: + quality_score = self.dgqp_module(bbox_pred) + cls_score = F.sigmoid(cls_score) * quality_score + + if deploy: + # Now only supports batch size = 1 in deploy + # TODO(ygh): support batch size > 1 + cls_score = F.sigmoid(cls_score).reshape( + [1, self.cls_out_channels, -1]).transpose([0, 2, 1]) + bbox_pred = bbox_pred.reshape([1, (self.reg_max + 1) * 4, + -1]).transpose([0, 2, 1]) + elif not self.training: + cls_score = F.sigmoid(cls_score.transpose([0, 2, 3, 1])) + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]) + + cls_logits_list.append(cls_score) + bboxes_reg_list.append(bbox_pred) + + return (cls_logits_list, bboxes_reg_list) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/roi_extractor.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/roi_extractor.py new file mode 100644 index 000000000..35c3924e3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/roi_extractor.py @@ -0,0 +1,111 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +from ppdet.core.workspace import register +from ppdet.modeling import ops + + +def _to_list(v): + if not isinstance(v, (list, tuple)): + return [v] + return v + + +@register +class RoIAlign(object): + """ + RoI Align module + + For more details, please refer to the document of roi_align in + in ppdet/modeing/ops.py + + Args: + resolution (int): The output size, default 14 + spatial_scale (float): Multiplicative spatial scale factor to translate + ROI coords from their input scale to the scale used when pooling. + default 0.0625 + sampling_ratio (int): The number of sampling points in the interpolation + grid, default 0 + canconical_level (int): The referring level of FPN layer with + specified level. default 4 + canonical_size (int): The referring scale of FPN layer with + specified scale. default 224 + start_level (int): The start level of FPN layer to extract RoI feature, + default 0 + end_level (int): The end level of FPN layer to extract RoI feature, + default 3 + aligned (bool): Whether to add offset to rois' coord in roi_align. + default false + """ + + def __init__(self, + resolution=14, + spatial_scale=0.0625, + sampling_ratio=0, + canconical_level=4, + canonical_size=224, + start_level=0, + end_level=3, + aligned=False): + super(RoIAlign, self).__init__() + self.resolution = resolution + self.spatial_scale = _to_list(spatial_scale) + self.sampling_ratio = sampling_ratio + self.canconical_level = canconical_level + self.canonical_size = canonical_size + self.start_level = start_level + self.end_level = end_level + self.aligned = aligned + + @classmethod + def from_config(cls, cfg, input_shape): + return {'spatial_scale': [1. / i.stride for i in input_shape]} + + def __call__(self, feats, roi, rois_num): + roi = paddle.concat(roi) if len(roi) > 1 else roi[0] + if len(feats) == 1: + rois_feat = ops.roi_align( + feats[self.start_level], + roi, + self.resolution, + self.spatial_scale[0], + rois_num=rois_num, + aligned=self.aligned) + else: + offset = 2 + k_min = self.start_level + offset + k_max = self.end_level + offset + rois_dist, restore_index, rois_num_dist = ops.distribute_fpn_proposals( + roi, + k_min, + k_max, + self.canconical_level, + self.canonical_size, + rois_num=rois_num) + rois_feat_list = [] + for lvl in range(self.start_level, self.end_level + 1): + roi_feat = ops.roi_align( + feats[lvl], + rois_dist[lvl], + self.resolution, + self.spatial_scale[lvl], + sampling_ratio=self.sampling_ratio, + rois_num=rois_num_dist[lvl], + aligned=self.aligned) + rois_feat_list.append(roi_feat) + rois_feat_shuffle = paddle.concat(rois_feat_list) + rois_feat = paddle.gather(rois_feat_shuffle, restore_index) + + return rois_feat diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/s2anet_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/s2anet_head.py new file mode 100644 index 000000000..7910379c4 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/s2anet_head.py @@ -0,0 +1,1048 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# The code is based on https://github.com/csuhan/s2anet/blob/master/mmdet/models/anchor_heads_rotated/s2anet_head.py + +import paddle +from paddle import ParamAttr +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, Constant +from ppdet.core.workspace import register +from ppdet.modeling import ops +from ppdet.modeling import bbox_utils +from ppdet.modeling.proposal_generator.target_layer import RBoxAssigner +import numpy as np + + +class S2ANetAnchorGenerator(nn.Layer): + """ + AnchorGenerator by paddle + """ + + def __init__(self, base_size, scales, ratios, scale_major=True, ctr=None): + super(S2ANetAnchorGenerator, self).__init__() + self.base_size = base_size + self.scales = paddle.to_tensor(scales) + self.ratios = paddle.to_tensor(ratios) + self.scale_major = scale_major + self.ctr = ctr + self.base_anchors = self.gen_base_anchors() + + @property + def num_base_anchors(self): + return self.base_anchors.shape[0] + + def gen_base_anchors(self): + w = self.base_size + h = self.base_size + if self.ctr is None: + x_ctr = 0.5 * (w - 1) + y_ctr = 0.5 * (h - 1) + else: + x_ctr, y_ctr = self.ctr + + h_ratios = paddle.sqrt(self.ratios) + w_ratios = 1 / h_ratios + if self.scale_major: + ws = (w * w_ratios[:] * self.scales[:]).reshape([-1]) + hs = (h * h_ratios[:] * self.scales[:]).reshape([-1]) + else: + ws = (w * self.scales[:] * w_ratios[:]).reshape([-1]) + hs = (h * self.scales[:] * h_ratios[:]).reshape([-1]) + + base_anchors = paddle.stack( + [ + x_ctr - 0.5 * (ws - 1), y_ctr - 0.5 * (hs - 1), + x_ctr + 0.5 * (ws - 1), y_ctr + 0.5 * (hs - 1) + ], + axis=-1) + base_anchors = paddle.round(base_anchors) + return base_anchors + + def _meshgrid(self, x, y, row_major=True): + yy, xx = paddle.meshgrid(y, x) + yy = yy.reshape([-1]) + xx = xx.reshape([-1]) + if row_major: + return xx, yy + else: + return yy, xx + + def forward(self, featmap_size, stride=16): + # featmap_size*stride project it to original area + + feat_h = featmap_size[0] + feat_w = featmap_size[1] + shift_x = paddle.arange(0, feat_w, 1, 'int32') * stride + shift_y = paddle.arange(0, feat_h, 1, 'int32') * stride + shift_xx, shift_yy = self._meshgrid(shift_x, shift_y) + shifts = paddle.stack([shift_xx, shift_yy, shift_xx, shift_yy], axis=-1) + + all_anchors = self.base_anchors[:, :] + shifts[:, :] + all_anchors = all_anchors.reshape([feat_h * feat_w, 4]) + return all_anchors + + def valid_flags(self, featmap_size, valid_size): + feat_h, feat_w = featmap_size + valid_h, valid_w = valid_size + assert valid_h <= feat_h and valid_w <= feat_w + valid_x = paddle.zeros([feat_w], dtype='int32') + valid_y = paddle.zeros([feat_h], dtype='int32') + valid_x[:valid_w] = 1 + valid_y[:valid_h] = 1 + valid_xx, valid_yy = self._meshgrid(valid_x, valid_y) + valid = valid_xx & valid_yy + valid = paddle.reshape(valid, [-1, 1]) + valid = paddle.expand(valid, [-1, self.num_base_anchors]).reshape([-1]) + return valid + + +class AlignConv(nn.Layer): + def __init__(self, in_channels, out_channels, kernel_size=3, groups=1): + super(AlignConv, self).__init__() + self.kernel_size = kernel_size + self.align_conv = paddle.vision.ops.DeformConv2D( + in_channels, + out_channels, + kernel_size=self.kernel_size, + padding=(self.kernel_size - 1) // 2, + groups=groups, + weight_attr=ParamAttr(initializer=Normal(0, 0.01)), + bias_attr=None) + + @paddle.no_grad() + def get_offset(self, anchors, featmap_size, stride): + """ + Args: + anchors: [M,5] xc,yc,w,h,angle + featmap_size: (feat_h, feat_w) + stride: 8 + Returns: + + """ + anchors = paddle.reshape(anchors, [-1, 5]) # (NA,5) + dtype = anchors.dtype + feat_h = featmap_size[0] + feat_w = featmap_size[1] + pad = (self.kernel_size - 1) // 2 + idx = paddle.arange(-pad, pad + 1, dtype=dtype) + + yy, xx = paddle.meshgrid(idx, idx) + xx = paddle.reshape(xx, [-1]) + yy = paddle.reshape(yy, [-1]) + + # get sampling locations of default conv + xc = paddle.arange(0, feat_w, dtype=dtype) + yc = paddle.arange(0, feat_h, dtype=dtype) + yc, xc = paddle.meshgrid(yc, xc) + + xc = paddle.reshape(xc, [-1, 1]) + yc = paddle.reshape(yc, [-1, 1]) + x_conv = xc + xx + y_conv = yc + yy + + # get sampling locations of anchors + # x_ctr, y_ctr, w, h, a = np.unbind(anchors, dim=1) + x_ctr = anchors[:, 0] + y_ctr = anchors[:, 1] + w = anchors[:, 2] + h = anchors[:, 3] + a = anchors[:, 4] + + x_ctr = paddle.reshape(x_ctr, [-1, 1]) + y_ctr = paddle.reshape(y_ctr, [-1, 1]) + w = paddle.reshape(w, [-1, 1]) + h = paddle.reshape(h, [-1, 1]) + a = paddle.reshape(a, [-1, 1]) + + x_ctr = x_ctr / stride + y_ctr = y_ctr / stride + w_s = w / stride + h_s = h / stride + cos, sin = paddle.cos(a), paddle.sin(a) + dw, dh = w_s / self.kernel_size, h_s / self.kernel_size + x, y = dw * xx, dh * yy + xr = cos * x - sin * y + yr = sin * x + cos * y + x_anchor, y_anchor = xr + x_ctr, yr + y_ctr + # get offset filed + offset_x = x_anchor - x_conv + offset_y = y_anchor - y_conv + offset = paddle.stack([offset_y, offset_x], axis=-1) + offset = paddle.reshape( + offset, [feat_h * feat_w, self.kernel_size * self.kernel_size * 2]) + offset = paddle.transpose(offset, [1, 0]) + offset = paddle.reshape( + offset, + [1, self.kernel_size * self.kernel_size * 2, feat_h, feat_w]) + return offset + + def forward(self, x, refine_anchors, featmap_size, stride): + offset = self.get_offset(refine_anchors, featmap_size, stride) + x = F.relu(self.align_conv(x, offset)) + return x + + +@register +class S2ANetHead(nn.Layer): + """ + S2Anet head + Args: + stacked_convs (int): number of stacked_convs + feat_in (int): input channels of feat + feat_out (int): output channels of feat + num_classes (int): num_classes + anchor_strides (list): stride of anchors + anchor_scales (list): scale of anchors + anchor_ratios (list): ratios of anchors + target_means (list): target_means + target_stds (list): target_stds + align_conv_type (str): align_conv_type ['Conv', 'AlignConv'] + align_conv_size (int): kernel size of align_conv + use_sigmoid_cls (bool): use sigmoid_cls or not + reg_loss_weight (list): loss weight for regression + """ + __shared__ = ['num_classes'] + __inject__ = ['anchor_assign'] + + def __init__(self, + stacked_convs=2, + feat_in=256, + feat_out=256, + num_classes=15, + anchor_strides=[8, 16, 32, 64, 128], + anchor_scales=[4], + anchor_ratios=[1.0], + target_means=0.0, + target_stds=1.0, + align_conv_type='AlignConv', + align_conv_size=3, + use_sigmoid_cls=True, + anchor_assign=RBoxAssigner().__dict__, + reg_loss_weight=[1.0, 1.0, 1.0, 1.0, 1.1], + cls_loss_weight=[1.1, 1.05], + reg_loss_type='l1'): + super(S2ANetHead, self).__init__() + self.stacked_convs = stacked_convs + self.feat_in = feat_in + self.feat_out = feat_out + self.anchor_list = None + self.anchor_scales = anchor_scales + self.anchor_ratios = anchor_ratios + self.anchor_strides = anchor_strides + self.anchor_strides = paddle.to_tensor(anchor_strides) + self.anchor_base_sizes = list(anchor_strides) + self.means = paddle.ones(shape=[5]) * target_means + self.stds = paddle.ones(shape=[5]) * target_stds + assert align_conv_type in ['AlignConv', 'Conv', 'DCN'] + self.align_conv_type = align_conv_type + self.align_conv_size = align_conv_size + + self.use_sigmoid_cls = use_sigmoid_cls + self.cls_out_channels = num_classes if self.use_sigmoid_cls else 1 + self.sampling = False + self.anchor_assign = anchor_assign + self.reg_loss_weight = reg_loss_weight + self.cls_loss_weight = cls_loss_weight + self.alpha = 1.0 + self.beta = 1.0 + self.reg_loss_type = reg_loss_type + self.s2anet_head_out = None + + # anchor + self.anchor_generators = [] + for anchor_base in self.anchor_base_sizes: + self.anchor_generators.append( + S2ANetAnchorGenerator(anchor_base, anchor_scales, + anchor_ratios)) + + self.anchor_generators = nn.LayerList(self.anchor_generators) + self.fam_cls_convs = nn.Sequential() + self.fam_reg_convs = nn.Sequential() + + for i in range(self.stacked_convs): + chan_in = self.feat_in if i == 0 else self.feat_out + + self.fam_cls_convs.add_sublayer( + 'fam_cls_conv_{}'.format(i), + nn.Conv2D( + in_channels=chan_in, + out_channels=self.feat_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0)))) + + self.fam_cls_convs.add_sublayer('fam_cls_conv_{}_act'.format(i), + nn.ReLU()) + + self.fam_reg_convs.add_sublayer( + 'fam_reg_conv_{}'.format(i), + nn.Conv2D( + in_channels=chan_in, + out_channels=self.feat_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0)))) + + self.fam_reg_convs.add_sublayer('fam_reg_conv_{}_act'.format(i), + nn.ReLU()) + + self.fam_reg = nn.Conv2D( + self.feat_out, + 5, + 1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0))) + prior_prob = 0.01 + bias_init = float(-np.log((1 - prior_prob) / prior_prob)) + self.fam_cls = nn.Conv2D( + self.feat_out, + self.cls_out_channels, + 1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(bias_init))) + + if self.align_conv_type == "AlignConv": + self.align_conv = AlignConv(self.feat_out, self.feat_out, + self.align_conv_size) + elif self.align_conv_type == "Conv": + self.align_conv = nn.Conv2D( + self.feat_out, + self.feat_out, + self.align_conv_size, + padding=(self.align_conv_size - 1) // 2, + bias_attr=ParamAttr(initializer=Constant(0))) + + elif self.align_conv_type == "DCN": + self.align_conv_offset = nn.Conv2D( + self.feat_out, + 2 * self.align_conv_size**2, + 1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0))) + + self.align_conv = paddle.vision.ops.DeformConv2D( + self.feat_out, + self.feat_out, + self.align_conv_size, + padding=(self.align_conv_size - 1) // 2, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=False) + + self.or_conv = nn.Conv2D( + self.feat_out, + self.feat_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0))) + + # ODM + self.odm_cls_convs = nn.Sequential() + self.odm_reg_convs = nn.Sequential() + + for i in range(self.stacked_convs): + ch_in = self.feat_out + # ch_in = int(self.feat_out / 8) if i == 0 else self.feat_out + + self.odm_cls_convs.add_sublayer( + 'odm_cls_conv_{}'.format(i), + nn.Conv2D( + in_channels=ch_in, + out_channels=self.feat_out, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0)))) + + self.odm_cls_convs.add_sublayer('odm_cls_conv_{}_act'.format(i), + nn.ReLU()) + + self.odm_reg_convs.add_sublayer( + 'odm_reg_conv_{}'.format(i), + nn.Conv2D( + in_channels=self.feat_out, + out_channels=self.feat_out, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0)))) + + self.odm_reg_convs.add_sublayer('odm_reg_conv_{}_act'.format(i), + nn.ReLU()) + + self.odm_cls = nn.Conv2D( + self.feat_out, + self.cls_out_channels, + 3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(bias_init))) + self.odm_reg = nn.Conv2D( + self.feat_out, + 5, + 3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0.0, 0.01)), + bias_attr=ParamAttr(initializer=Constant(0))) + + self.featmap_sizes = [] + self.base_anchors_list = [] + self.refine_anchor_list = [] + + def forward(self, feats): + fam_reg_branch_list = [] + fam_cls_branch_list = [] + + odm_reg_branch_list = [] + odm_cls_branch_list = [] + + self.featmap_sizes_list = [] + self.base_anchors_list = [] + self.refine_anchor_list = [] + + for feat_idx in range(len(feats)): + feat = feats[feat_idx] + fam_cls_feat = self.fam_cls_convs(feat) + + fam_cls = self.fam_cls(fam_cls_feat) + # [N, CLS, H, W] --> [N, H, W, CLS] + fam_cls = fam_cls.transpose([0, 2, 3, 1]) + fam_cls_reshape = paddle.reshape( + fam_cls, [fam_cls.shape[0], -1, self.cls_out_channels]) + fam_cls_branch_list.append(fam_cls_reshape) + + fam_reg_feat = self.fam_reg_convs(feat) + + fam_reg = self.fam_reg(fam_reg_feat) + # [N, 5, H, W] --> [N, H, W, 5] + fam_reg = fam_reg.transpose([0, 2, 3, 1]) + fam_reg_reshape = paddle.reshape(fam_reg, [fam_reg.shape[0], -1, 5]) + fam_reg_branch_list.append(fam_reg_reshape) + + # prepare anchor + featmap_size = (paddle.shape(feat)[2], paddle.shape(feat)[3]) + self.featmap_sizes_list.append(featmap_size) + init_anchors = self.anchor_generators[feat_idx]( + featmap_size, self.anchor_strides[feat_idx]) + + init_anchors = paddle.to_tensor(init_anchors, dtype='float32') + NA = featmap_size[0] * featmap_size[1] + init_anchors = paddle.reshape(init_anchors, [NA, 4]) + init_anchors = self.rect2rbox(init_anchors) + self.base_anchors_list.append(init_anchors) + + if self.training: + refine_anchor = self.bbox_decode(fam_reg.detach(), init_anchors) + else: + refine_anchor = self.bbox_decode(fam_reg, init_anchors) + + self.refine_anchor_list.append(refine_anchor) + + if self.align_conv_type == 'AlignConv': + align_feat = self.align_conv(feat, + refine_anchor.clone(), + featmap_size, + self.anchor_strides[feat_idx]) + elif self.align_conv_type == 'DCN': + align_offset = self.align_conv_offset(feat) + align_feat = self.align_conv(feat, align_offset) + elif self.align_conv_type == 'Conv': + align_feat = self.align_conv(feat) + + or_feat = self.or_conv(align_feat) + odm_reg_feat = or_feat + odm_cls_feat = or_feat + + odm_reg_feat = self.odm_reg_convs(odm_reg_feat) + odm_cls_feat = self.odm_cls_convs(odm_cls_feat) + + odm_cls_score = self.odm_cls(odm_cls_feat) + # [N, CLS, H, W] --> [N, H, W, CLS] + odm_cls_score = odm_cls_score.transpose([0, 2, 3, 1]) + odm_cls_score_shape = odm_cls_score.shape + odm_cls_score_reshape = paddle.reshape(odm_cls_score, [ + odm_cls_score_shape[0], odm_cls_score_shape[1] * + odm_cls_score_shape[2], self.cls_out_channels + ]) + + odm_cls_branch_list.append(odm_cls_score_reshape) + + odm_bbox_pred = self.odm_reg(odm_reg_feat) + # [N, 5, H, W] --> [N, H, W, 5] + odm_bbox_pred = odm_bbox_pred.transpose([0, 2, 3, 1]) + odm_bbox_pred_reshape = paddle.reshape(odm_bbox_pred, [-1, 5]) + odm_bbox_pred_reshape = paddle.unsqueeze( + odm_bbox_pred_reshape, axis=0) + odm_reg_branch_list.append(odm_bbox_pred_reshape) + + self.s2anet_head_out = (fam_cls_branch_list, fam_reg_branch_list, + odm_cls_branch_list, odm_reg_branch_list) + return self.s2anet_head_out + + def get_prediction(self, nms_pre=2000): + refine_anchors = self.refine_anchor_list + fam_cls_branch_list = self.s2anet_head_out[0] + fam_reg_branch_list = self.s2anet_head_out[1] + odm_cls_branch_list = self.s2anet_head_out[2] + odm_reg_branch_list = self.s2anet_head_out[3] + pred_scores, pred_bboxes = self.get_bboxes( + odm_cls_branch_list, odm_reg_branch_list, refine_anchors, nms_pre, + self.cls_out_channels, self.use_sigmoid_cls) + return pred_scores, pred_bboxes + + def smooth_l1_loss(self, pred, label, delta=1.0 / 9.0): + """ + Args: + pred: pred score + label: label + delta: delta + Returns: loss + """ + assert pred.shape == label.shape and label.numel() > 0 + assert delta > 0 + diff = paddle.abs(pred - label) + loss = paddle.where(diff < delta, 0.5 * diff * diff / delta, + diff - 0.5 * delta) + return loss + + def get_fam_loss(self, fam_target, s2anet_head_out, reg_loss_type='gwd'): + (labels, label_weights, bbox_targets, bbox_weights, bbox_gt_bboxes, + pos_inds, neg_inds) = fam_target + fam_cls_branch_list, fam_reg_branch_list, odm_cls_branch_list, odm_reg_branch_list = s2anet_head_out + + fam_cls_losses = [] + fam_bbox_losses = [] + st_idx = 0 + num_total_samples = len(pos_inds) + len( + neg_inds) if self.sampling else len(pos_inds) + num_total_samples = max(1, num_total_samples) + + for idx, feat_size in enumerate(self.featmap_sizes_list): + feat_anchor_num = feat_size[0] * feat_size[1] + + # step1: get data + feat_labels = labels[st_idx:st_idx + feat_anchor_num] + feat_label_weights = label_weights[st_idx:st_idx + feat_anchor_num] + + feat_bbox_targets = bbox_targets[st_idx:st_idx + feat_anchor_num, :] + feat_bbox_weights = bbox_weights[st_idx:st_idx + feat_anchor_num, :] + + # step2: calc cls loss + feat_labels = feat_labels.reshape(-1) + feat_label_weights = feat_label_weights.reshape(-1) + + fam_cls_score = fam_cls_branch_list[idx] + fam_cls_score = paddle.squeeze(fam_cls_score, axis=0) + fam_cls_score1 = fam_cls_score + + feat_labels = paddle.to_tensor(feat_labels) + feat_labels_one_hot = paddle.nn.functional.one_hot( + feat_labels, self.cls_out_channels + 1) + feat_labels_one_hot = feat_labels_one_hot[:, 1:] + feat_labels_one_hot.stop_gradient = True + + num_total_samples = paddle.to_tensor( + num_total_samples, dtype='float32', stop_gradient=True) + + fam_cls = F.sigmoid_focal_loss( + fam_cls_score1, + feat_labels_one_hot, + normalizer=num_total_samples, + reduction='none') + + feat_label_weights = feat_label_weights.reshape( + feat_label_weights.shape[0], 1) + feat_label_weights = np.repeat( + feat_label_weights, self.cls_out_channels, axis=1) + feat_label_weights = paddle.to_tensor( + feat_label_weights, stop_gradient=True) + + fam_cls = fam_cls * feat_label_weights + fam_cls_total = paddle.sum(fam_cls) + fam_cls_losses.append(fam_cls_total) + + # step3: regression loss + feat_bbox_targets = paddle.to_tensor( + feat_bbox_targets, dtype='float32', stop_gradient=True) + feat_bbox_targets = paddle.reshape(feat_bbox_targets, [-1, 5]) + + fam_bbox_pred = fam_reg_branch_list[idx] + fam_bbox_pred = paddle.squeeze(fam_bbox_pred, axis=0) + fam_bbox_pred = paddle.reshape(fam_bbox_pred, [-1, 5]) + fam_bbox = self.smooth_l1_loss(fam_bbox_pred, feat_bbox_targets) + loss_weight = paddle.to_tensor( + self.reg_loss_weight, dtype='float32', stop_gradient=True) + fam_bbox = paddle.multiply(fam_bbox, loss_weight) + feat_bbox_weights = paddle.to_tensor( + feat_bbox_weights, stop_gradient=True) + + if reg_loss_type == 'l1': + fam_bbox = fam_bbox * feat_bbox_weights + fam_bbox_total = paddle.sum(fam_bbox) / num_total_samples + elif reg_loss_type == 'iou' or reg_loss_type == 'gwd': + fam_bbox = paddle.sum(fam_bbox, axis=-1) + feat_bbox_weights = paddle.sum(feat_bbox_weights, axis=-1) + try: + from rbox_iou_ops import rbox_iou + except Exception as e: + print("import custom_ops error, try install rbox_iou_ops " \ + "following ppdet/ext_op/README.md", e) + sys.stdout.flush() + sys.exit(-1) + # calc iou + fam_bbox_decode = self.delta2rbox(self.base_anchors_list[idx], + fam_bbox_pred) + bbox_gt_bboxes = paddle.to_tensor( + bbox_gt_bboxes, + dtype=fam_bbox_decode.dtype, + place=fam_bbox_decode.place) + bbox_gt_bboxes.stop_gradient = True + iou = rbox_iou(fam_bbox_decode, bbox_gt_bboxes) + iou = paddle.diag(iou) + + if reg_loss_type == 'gwd': + bbox_gt_bboxes_level = bbox_gt_bboxes[st_idx:st_idx + + feat_anchor_num, :] + fam_bbox_total = self.gwd_loss(fam_bbox_decode, + bbox_gt_bboxes_level) + fam_bbox_total = fam_bbox_total * feat_bbox_weights + fam_bbox_total = paddle.sum( + fam_bbox_total) / num_total_samples + + fam_bbox_losses.append(fam_bbox_total) + st_idx += feat_anchor_num + + fam_cls_loss = paddle.add_n(fam_cls_losses) + fam_cls_loss_weight = paddle.to_tensor( + self.cls_loss_weight[0], dtype='float32', stop_gradient=True) + fam_cls_loss = fam_cls_loss * fam_cls_loss_weight + fam_reg_loss = paddle.add_n(fam_bbox_losses) + return fam_cls_loss, fam_reg_loss + + def get_odm_loss(self, odm_target, s2anet_head_out, reg_loss_type='gwd'): + (labels, label_weights, bbox_targets, bbox_weights, bbox_gt_bboxes, + pos_inds, neg_inds) = odm_target + fam_cls_branch_list, fam_reg_branch_list, odm_cls_branch_list, odm_reg_branch_list = s2anet_head_out + + odm_cls_losses = [] + odm_bbox_losses = [] + st_idx = 0 + num_total_samples = len(pos_inds) + len( + neg_inds) if self.sampling else len(pos_inds) + num_total_samples = max(1, num_total_samples) + + for idx, feat_size in enumerate(self.featmap_sizes_list): + feat_anchor_num = feat_size[0] * feat_size[1] + + # step1: get data + feat_labels = labels[st_idx:st_idx + feat_anchor_num] + feat_label_weights = label_weights[st_idx:st_idx + feat_anchor_num] + + feat_bbox_targets = bbox_targets[st_idx:st_idx + feat_anchor_num, :] + feat_bbox_weights = bbox_weights[st_idx:st_idx + feat_anchor_num, :] + + # step2: calc cls loss + feat_labels = feat_labels.reshape(-1) + feat_label_weights = feat_label_weights.reshape(-1) + + odm_cls_score = odm_cls_branch_list[idx] + odm_cls_score = paddle.squeeze(odm_cls_score, axis=0) + odm_cls_score1 = odm_cls_score + + feat_labels = paddle.to_tensor(feat_labels) + feat_labels_one_hot = paddle.nn.functional.one_hot( + feat_labels, self.cls_out_channels + 1) + feat_labels_one_hot = feat_labels_one_hot[:, 1:] + feat_labels_one_hot.stop_gradient = True + + num_total_samples = paddle.to_tensor( + num_total_samples, dtype='float32', stop_gradient=True) + odm_cls = F.sigmoid_focal_loss( + odm_cls_score1, + feat_labels_one_hot, + normalizer=num_total_samples, + reduction='none') + + feat_label_weights = feat_label_weights.reshape( + feat_label_weights.shape[0], 1) + feat_label_weights = np.repeat( + feat_label_weights, self.cls_out_channels, axis=1) + feat_label_weights = paddle.to_tensor(feat_label_weights) + feat_label_weights.stop_gradient = True + + odm_cls = odm_cls * feat_label_weights + odm_cls_total = paddle.sum(odm_cls) + odm_cls_losses.append(odm_cls_total) + + # # step3: regression loss + feat_bbox_targets = paddle.to_tensor( + feat_bbox_targets, dtype='float32') + feat_bbox_targets = paddle.reshape(feat_bbox_targets, [-1, 5]) + feat_bbox_targets.stop_gradient = True + + odm_bbox_pred = odm_reg_branch_list[idx] + odm_bbox_pred = paddle.squeeze(odm_bbox_pred, axis=0) + odm_bbox_pred = paddle.reshape(odm_bbox_pred, [-1, 5]) + odm_bbox = self.smooth_l1_loss(odm_bbox_pred, feat_bbox_targets) + + loss_weight = paddle.to_tensor( + self.reg_loss_weight, dtype='float32', stop_gradient=True) + odm_bbox = paddle.multiply(odm_bbox, loss_weight) + feat_bbox_weights = paddle.to_tensor( + feat_bbox_weights, stop_gradient=True) + + if reg_loss_type == 'l1': + odm_bbox = odm_bbox * feat_bbox_weights + odm_bbox_total = paddle.sum(odm_bbox) / num_total_samples + elif reg_loss_type == 'iou' or reg_loss_type == 'gwd': + odm_bbox = paddle.sum(odm_bbox, axis=-1) + feat_bbox_weights = paddle.sum(feat_bbox_weights, axis=-1) + try: + from rbox_iou_ops import rbox_iou + except Exception as e: + print("import custom_ops error, try install rbox_iou_ops " \ + "following ppdet/ext_op/README.md", e) + sys.stdout.flush() + sys.exit(-1) + # calc iou + odm_bbox_decode = self.delta2rbox(self.refine_anchor_list[idx], + odm_bbox_pred) + bbox_gt_bboxes = paddle.to_tensor( + bbox_gt_bboxes, + dtype=odm_bbox_decode.dtype, + place=odm_bbox_decode.place) + bbox_gt_bboxes.stop_gradient = True + iou = rbox_iou(odm_bbox_decode, bbox_gt_bboxes) + iou = paddle.diag(iou) + + if reg_loss_type == 'gwd': + bbox_gt_bboxes_level = bbox_gt_bboxes[st_idx:st_idx + + feat_anchor_num, :] + odm_bbox_total = self.gwd_loss(odm_bbox_decode, + bbox_gt_bboxes_level) + odm_bbox_total = odm_bbox_total * feat_bbox_weights + odm_bbox_total = paddle.sum( + odm_bbox_total) / num_total_samples + + odm_bbox_losses.append(odm_bbox_total) + st_idx += feat_anchor_num + + odm_cls_loss = paddle.add_n(odm_cls_losses) + odm_cls_loss_weight = paddle.to_tensor( + self.cls_loss_weight[1], dtype='float32', stop_gradient=True) + odm_cls_loss = odm_cls_loss * odm_cls_loss_weight + odm_reg_loss = paddle.add_n(odm_bbox_losses) + return odm_cls_loss, odm_reg_loss + + def get_loss(self, inputs): + # inputs: im_id image im_shape scale_factor gt_bbox gt_class is_crowd + + # compute loss + fam_cls_loss_lst = [] + fam_reg_loss_lst = [] + odm_cls_loss_lst = [] + odm_reg_loss_lst = [] + + im_shape = inputs['im_shape'] + for im_id in range(im_shape.shape[0]): + np_im_shape = inputs['im_shape'][im_id].numpy() + np_scale_factor = inputs['scale_factor'][im_id].numpy() + # data_format: (xc, yc, w, h, theta) + gt_bboxes = inputs['gt_rbox'][im_id].numpy() + gt_labels = inputs['gt_class'][im_id].numpy() + is_crowd = inputs['is_crowd'][im_id].numpy() + gt_labels = gt_labels + 1 + + # featmap_sizes + anchors_list_all = np.concatenate(self.base_anchors_list) + + # get im_feat + fam_cls_feats_list = [e[im_id] for e in self.s2anet_head_out[0]] + fam_reg_feats_list = [e[im_id] for e in self.s2anet_head_out[1]] + odm_cls_feats_list = [e[im_id] for e in self.s2anet_head_out[2]] + odm_reg_feats_list = [e[im_id] for e in self.s2anet_head_out[3]] + im_s2anet_head_out = (fam_cls_feats_list, fam_reg_feats_list, + odm_cls_feats_list, odm_reg_feats_list) + + # FAM + im_fam_target = self.anchor_assign(anchors_list_all, gt_bboxes, + gt_labels, is_crowd) + if im_fam_target is not None: + im_fam_cls_loss, im_fam_reg_loss = self.get_fam_loss( + im_fam_target, im_s2anet_head_out, self.reg_loss_type) + fam_cls_loss_lst.append(im_fam_cls_loss) + fam_reg_loss_lst.append(im_fam_reg_loss) + + # ODM + np_refine_anchors_list = paddle.concat( + self.refine_anchor_list).numpy() + np_refine_anchors_list = np.concatenate(np_refine_anchors_list) + np_refine_anchors_list = np_refine_anchors_list.reshape(-1, 5) + im_odm_target = self.anchor_assign(np_refine_anchors_list, + gt_bboxes, gt_labels, is_crowd) + + if im_odm_target is not None: + im_odm_cls_loss, im_odm_reg_loss = self.get_odm_loss( + im_odm_target, im_s2anet_head_out, self.reg_loss_type) + odm_cls_loss_lst.append(im_odm_cls_loss) + odm_reg_loss_lst.append(im_odm_reg_loss) + fam_cls_loss = paddle.add_n(fam_cls_loss_lst) + fam_reg_loss = paddle.add_n(fam_reg_loss_lst) + odm_cls_loss = paddle.add_n(odm_cls_loss_lst) + odm_reg_loss = paddle.add_n(odm_reg_loss_lst) + return { + 'fam_cls_loss': fam_cls_loss, + 'fam_reg_loss': fam_reg_loss, + 'odm_cls_loss': odm_cls_loss, + 'odm_reg_loss': odm_reg_loss + } + + def get_bboxes(self, cls_score_list, bbox_pred_list, mlvl_anchors, nms_pre, + cls_out_channels, use_sigmoid_cls): + assert len(cls_score_list) == len(bbox_pred_list) == len(mlvl_anchors) + + mlvl_bboxes = [] + mlvl_scores = [] + + idx = 0 + for cls_score, bbox_pred, anchors in zip(cls_score_list, bbox_pred_list, + mlvl_anchors): + cls_score = paddle.reshape(cls_score, [-1, cls_out_channels]) + if use_sigmoid_cls: + scores = F.sigmoid(cls_score) + else: + scores = F.softmax(cls_score, axis=-1) + + # bbox_pred = bbox_pred.permute(1, 2, 0).reshape(-1, 5) + bbox_pred = paddle.transpose(bbox_pred, [1, 2, 0]) + bbox_pred = paddle.reshape(bbox_pred, [-1, 5]) + anchors = paddle.reshape(anchors, [-1, 5]) + + if scores.shape[0] > nms_pre: + # Get maximum scores for foreground classes. + if use_sigmoid_cls: + max_scores = paddle.max(scores, axis=1) + else: + max_scores = paddle.max(scores[:, 1:], axis=1) + + topk_val, topk_inds = paddle.topk(max_scores, nms_pre) + anchors = paddle.gather(anchors, topk_inds) + bbox_pred = paddle.gather(bbox_pred, topk_inds) + scores = paddle.gather(scores, topk_inds) + + bbox_delta = paddle.reshape(bbox_pred, [-1, 5]) + bboxes = self.delta2rbox(anchors, bbox_delta) + mlvl_bboxes.append(bboxes) + mlvl_scores.append(scores) + + idx += 1 + + mlvl_bboxes = paddle.concat(mlvl_bboxes, axis=0) + mlvl_scores = paddle.concat(mlvl_scores) + + return mlvl_scores, mlvl_bboxes + + def rect2rbox(self, bboxes): + """ + :param bboxes: shape (n, 4) (xmin, ymin, xmax, ymax) + :return: dbboxes: shape (n, 5) (x_ctr, y_ctr, w, h, angle) + """ + bboxes = paddle.reshape(bboxes, [-1, 4]) + num_boxes = paddle.shape(bboxes)[0] + x_ctr = (bboxes[:, 2] + bboxes[:, 0]) / 2.0 + y_ctr = (bboxes[:, 3] + bboxes[:, 1]) / 2.0 + edges1 = paddle.abs(bboxes[:, 2] - bboxes[:, 0]) + edges2 = paddle.abs(bboxes[:, 3] - bboxes[:, 1]) + + rbox_w = paddle.maximum(edges1, edges2) + rbox_h = paddle.minimum(edges1, edges2) + + # set angle + inds = edges1 < edges2 + inds = paddle.cast(inds, 'int32') + rboxes_angle = inds * np.pi / 2.0 + + rboxes = paddle.stack( + (x_ctr, y_ctr, rbox_w, rbox_h, rboxes_angle), axis=1) + return rboxes + + # deltas to rbox + def delta2rbox(self, rrois, deltas, wh_ratio_clip=1e-6): + """ + :param rrois: (cx, cy, w, h, theta) + :param deltas: (dx, dy, dw, dh, dtheta) + :param means: means of anchor + :param stds: stds of anchor + :param wh_ratio_clip: clip threshold of wh_ratio + :return: + """ + deltas = paddle.reshape(deltas, [-1, 5]) + rrois = paddle.reshape(rrois, [-1, 5]) + # fix dy2st bug denorm_deltas = deltas * self.stds + self.means + denorm_deltas = paddle.add( + paddle.multiply(deltas, self.stds), self.means) + + dx = denorm_deltas[:, 0] + dy = denorm_deltas[:, 1] + dw = denorm_deltas[:, 2] + dh = denorm_deltas[:, 3] + dangle = denorm_deltas[:, 4] + max_ratio = np.abs(np.log(wh_ratio_clip)) + dw = paddle.clip(dw, min=-max_ratio, max=max_ratio) + dh = paddle.clip(dh, min=-max_ratio, max=max_ratio) + + rroi_x = rrois[:, 0] + rroi_y = rrois[:, 1] + rroi_w = rrois[:, 2] + rroi_h = rrois[:, 3] + rroi_angle = rrois[:, 4] + + gx = dx * rroi_w * paddle.cos(rroi_angle) - dy * rroi_h * paddle.sin( + rroi_angle) + rroi_x + gy = dx * rroi_w * paddle.sin(rroi_angle) + dy * rroi_h * paddle.cos( + rroi_angle) + rroi_y + gw = rroi_w * dw.exp() + gh = rroi_h * dh.exp() + ga = np.pi * dangle + rroi_angle + ga = (ga + np.pi / 4) % np.pi - np.pi / 4 + ga = paddle.to_tensor(ga) + gw = paddle.to_tensor(gw, dtype='float32') + gh = paddle.to_tensor(gh, dtype='float32') + bboxes = paddle.stack([gx, gy, gw, gh, ga], axis=-1) + return bboxes + + def bbox_decode(self, bbox_preds, anchors): + """decode bbox from deltas + Args: + bbox_preds: [N,H,W,5] + anchors: [H*W,5] + return: + bboxes: [N,H,W,5] + """ + num_imgs, H, W, _ = bbox_preds.shape + bbox_delta = paddle.reshape(bbox_preds, [-1, 5]) + bboxes = self.delta2rbox(anchors, bbox_delta) + return bboxes + + def trace(self, A): + tr = paddle.diagonal(A, axis1=-2, axis2=-1) + tr = paddle.sum(tr, axis=-1) + return tr + + def sqrt_newton_schulz_autograd(self, A, numIters): + A_shape = A.shape + batchSize = A_shape[0] + dim = A_shape[1] + + normA = A * A + normA = paddle.sum(normA, axis=1) + normA = paddle.sum(normA, axis=1) + normA = paddle.sqrt(normA) + normA1 = normA.reshape([batchSize, 1, 1]) + Y = paddle.divide(A, paddle.expand_as(normA1, A)) + I = paddle.eye(dim, dim).reshape([1, dim, dim]) + l0 = [] + for i in range(batchSize): + l0.append(I) + I = paddle.concat(l0, axis=0) + I.stop_gradient = False + Z = paddle.eye(dim, dim).reshape([1, dim, dim]) + l1 = [] + for i in range(batchSize): + l1.append(Z) + Z = paddle.concat(l1, axis=0) + Z.stop_gradient = False + + for i in range(numIters): + T = 0.5 * (3.0 * I - Z.bmm(Y)) + Y = Y.bmm(T) + Z = T.bmm(Z) + sA = Y * paddle.sqrt(normA1).reshape([batchSize, 1, 1]) + sA = paddle.expand_as(sA, A) + return sA + + def wasserstein_distance_sigma(sigma1, sigma2): + wasserstein_distance_item2 = paddle.matmul( + sigma1, sigma1) + paddle.matmul( + sigma2, sigma2) - 2 * self.sqrt_newton_schulz_autograd( + paddle.matmul( + paddle.matmul(sigma1, paddle.matmul(sigma2, sigma2)), + sigma1), 10) + wasserstein_distance_item2 = self.trace(wasserstein_distance_item2) + + return wasserstein_distance_item2 + + def xywhr2xyrs(self, xywhr): + xywhr = paddle.reshape(xywhr, [-1, 5]) + xy = xywhr[:, :2] + wh = paddle.clip(xywhr[:, 2:4], min=1e-7, max=1e7) + r = xywhr[:, 4] + cos_r = paddle.cos(r) + sin_r = paddle.sin(r) + R = paddle.stack( + (cos_r, -sin_r, sin_r, cos_r), axis=-1).reshape([-1, 2, 2]) + S = 0.5 * paddle.nn.functional.diag_embed(wh) + return xy, R, S + + def gwd_loss(self, + pred, + target, + fun='log', + tau=1.0, + alpha=1.0, + normalize=False): + + xy_p, R_p, S_p = self.xywhr2xyrs(pred) + xy_t, R_t, S_t = self.xywhr2xyrs(target) + + xy_distance = (xy_p - xy_t).square().sum(axis=-1) + + Sigma_p = R_p.matmul(S_p.square()).matmul(R_p.transpose([0, 2, 1])) + Sigma_t = R_t.matmul(S_t.square()).matmul(R_t.transpose([0, 2, 1])) + + whr_distance = paddle.diagonal( + S_p, axis1=-2, axis2=-1).square().sum(axis=-1) + + whr_distance = whr_distance + paddle.diagonal( + S_t, axis1=-2, axis2=-1).square().sum(axis=-1) + _t = Sigma_p.matmul(Sigma_t) + + _t_tr = paddle.diagonal(_t, axis1=-2, axis2=-1).sum(axis=-1) + _t_det_sqrt = paddle.diagonal(S_p, axis1=-2, axis2=-1).prod(axis=-1) + _t_det_sqrt = _t_det_sqrt * paddle.diagonal( + S_t, axis1=-2, axis2=-1).prod(axis=-1) + whr_distance = whr_distance + (-2) * ( + (_t_tr + 2 * _t_det_sqrt).clip(0).sqrt()) + + distance = (xy_distance + alpha * alpha * whr_distance).clip(0) + + if normalize: + wh_p = pred[..., 2:4].clip(min=1e-7, max=1e7) + wh_t = target[..., 2:4].clip(min=1e-7, max=1e7) + scale = ((wh_p.log() + wh_t.log()).sum(dim=-1) / 4).exp() + distance = distance / scale + + if fun == 'log': + distance = paddle.log1p(distance) + + if tau >= 1.0: + return 1 - 1 / (tau + distance) + + return distance diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/simota_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/simota_head.py new file mode 100644 index 000000000..a1485f390 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/simota_head.py @@ -0,0 +1,498 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/dense_heads/yolox_head.py + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +from functools import partial +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Normal, Constant + +from ppdet.core.workspace import register + +from ppdet.modeling.bbox_utils import distance2bbox, bbox2distance +from ppdet.data.transform.atss_assigner import bbox_overlaps + +from .gfl_head import GFLHead + + +@register +class OTAHead(GFLHead): + """ + OTAHead + Args: + conv_feat (object): Instance of 'FCOSFeat' + num_classes (int): Number of classes + fpn_stride (list): The stride of each FPN Layer + prior_prob (float): Used to set the bias init for the class prediction layer + loss_qfl (object): Instance of QualityFocalLoss. + loss_dfl (object): Instance of DistributionFocalLoss. + loss_bbox (object): Instance of bbox loss. + assigner (object): Instance of label assigner. + reg_max: Max value of integral set :math: `{0, ..., reg_max}` + n QFL setting. Default: 16. + """ + __inject__ = [ + 'conv_feat', 'dgqp_module', 'loss_class', 'loss_dfl', 'loss_bbox', + 'assigner', 'nms' + ] + __shared__ = ['num_classes'] + + def __init__(self, + conv_feat='FCOSFeat', + dgqp_module=None, + num_classes=80, + fpn_stride=[8, 16, 32, 64, 128], + prior_prob=0.01, + loss_class='QualityFocalLoss', + loss_dfl='DistributionFocalLoss', + loss_bbox='GIoULoss', + assigner='SimOTAAssigner', + reg_max=16, + feat_in_chan=256, + nms=None, + nms_pre=1000, + cell_offset=0): + super(OTAHead, self).__init__( + conv_feat=conv_feat, + dgqp_module=dgqp_module, + num_classes=num_classes, + fpn_stride=fpn_stride, + prior_prob=prior_prob, + loss_class=loss_class, + loss_dfl=loss_dfl, + loss_bbox=loss_bbox, + reg_max=reg_max, + feat_in_chan=feat_in_chan, + nms=nms, + nms_pre=nms_pre, + cell_offset=cell_offset) + self.conv_feat = conv_feat + self.dgqp_module = dgqp_module + self.num_classes = num_classes + self.fpn_stride = fpn_stride + self.prior_prob = prior_prob + self.loss_qfl = loss_class + self.loss_dfl = loss_dfl + self.loss_bbox = loss_bbox + self.reg_max = reg_max + self.feat_in_chan = feat_in_chan + self.nms = nms + self.nms_pre = nms_pre + self.cell_offset = cell_offset + self.use_sigmoid = self.loss_qfl.use_sigmoid + + self.assigner = assigner + + def _get_target_single(self, flatten_cls_pred, flatten_center_and_stride, + flatten_bbox, gt_bboxes, gt_labels): + """Compute targets for priors in a single image. + """ + pos_num, label, label_weight, bbox_target = self.assigner( + F.sigmoid(flatten_cls_pred), flatten_center_and_stride, + flatten_bbox, gt_bboxes, gt_labels) + + return (pos_num, label, label_weight, bbox_target) + + def get_loss(self, head_outs, gt_meta): + cls_scores, bbox_preds = head_outs + num_level_anchors = [ + featmap.shape[-2] * featmap.shape[-1] for featmap in cls_scores + ] + num_imgs = gt_meta['im_id'].shape[0] + featmap_sizes = [[featmap.shape[-2], featmap.shape[-1]] + for featmap in cls_scores] + + decode_bbox_preds = [] + center_and_strides = [] + for featmap_size, stride, bbox_pred in zip(featmap_sizes, + self.fpn_stride, bbox_preds): + + # center in origin image + yy, xx = self.get_single_level_center_point(featmap_size, stride, + self.cell_offset) + + center_and_stride = paddle.stack([xx, yy, stride, stride], -1).tile( + [num_imgs, 1, 1]) + center_and_strides.append(center_and_stride) + center_in_feature = center_and_stride.reshape( + [-1, 4])[:, :-2] / stride + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]).reshape( + [num_imgs, -1, 4 * (self.reg_max + 1)]) + pred_distances = self.distribution_project(bbox_pred) + decode_bbox_pred_wo_stride = distance2bbox( + center_in_feature, pred_distances).reshape([num_imgs, -1, 4]) + decode_bbox_preds.append(decode_bbox_pred_wo_stride * stride) + + flatten_cls_preds = [ + cls_pred.transpose([0, 2, 3, 1]).reshape( + [num_imgs, -1, self.cls_out_channels]) + for cls_pred in cls_scores + ] + flatten_cls_preds = paddle.concat(flatten_cls_preds, axis=1) + flatten_bboxes = paddle.concat(decode_bbox_preds, axis=1) + flatten_center_and_strides = paddle.concat(center_and_strides, axis=1) + + gt_boxes, gt_labels = gt_meta['gt_bbox'], gt_meta['gt_class'] + pos_num_l, label_l, label_weight_l, bbox_target_l = [], [], [], [] + for flatten_cls_pred,flatten_center_and_stride,flatten_bbox,gt_box, gt_label \ + in zip(flatten_cls_preds.detach(),flatten_center_and_strides.detach(), \ + flatten_bboxes.detach(),gt_boxes, gt_labels): + pos_num, label, label_weight, bbox_target = self._get_target_single( + flatten_cls_pred, flatten_center_and_stride, flatten_bbox, + gt_box, gt_label) + pos_num_l.append(pos_num) + label_l.append(label) + label_weight_l.append(label_weight) + bbox_target_l.append(bbox_target) + + labels = paddle.to_tensor(np.stack(label_l, axis=0)) + label_weights = paddle.to_tensor(np.stack(label_weight_l, axis=0)) + bbox_targets = paddle.to_tensor(np.stack(bbox_target_l, axis=0)) + + center_and_strides_list = self._images_to_levels( + flatten_center_and_strides, num_level_anchors) + labels_list = self._images_to_levels(labels, num_level_anchors) + label_weights_list = self._images_to_levels(label_weights, + num_level_anchors) + bbox_targets_list = self._images_to_levels(bbox_targets, + num_level_anchors) + num_total_pos = sum(pos_num_l) + try: + num_total_pos = paddle.distributed.all_reduce(num_total_pos.clone( + )) / paddle.distributed.get_world_size() + except: + num_total_pos = max(num_total_pos, 1) + + loss_bbox_list, loss_dfl_list, loss_qfl_list, avg_factor = [], [], [], [] + for cls_score, bbox_pred, center_and_strides, labels, label_weights, bbox_targets, stride in zip( + cls_scores, bbox_preds, center_and_strides_list, labels_list, + label_weights_list, bbox_targets_list, self.fpn_stride): + center_and_strides = center_and_strides.reshape([-1, 4]) + cls_score = cls_score.transpose([0, 2, 3, 1]).reshape( + [-1, self.cls_out_channels]) + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]).reshape( + [-1, 4 * (self.reg_max + 1)]) + bbox_targets = bbox_targets.reshape([-1, 4]) + labels = labels.reshape([-1]) + label_weights = label_weights.reshape([-1]) + + bg_class_ind = self.num_classes + pos_inds = paddle.nonzero( + paddle.logical_and((labels >= 0), (labels < bg_class_ind)), + as_tuple=False).squeeze(1) + score = np.zeros(labels.shape) + + if len(pos_inds) > 0: + pos_bbox_targets = paddle.gather(bbox_targets, pos_inds, axis=0) + pos_bbox_pred = paddle.gather(bbox_pred, pos_inds, axis=0) + pos_centers = paddle.gather( + center_and_strides[:, :-2], pos_inds, axis=0) / stride + + weight_targets = F.sigmoid(cls_score.detach()) + weight_targets = paddle.gather( + weight_targets.max(axis=1, keepdim=True), pos_inds, axis=0) + pos_bbox_pred_corners = self.distribution_project(pos_bbox_pred) + pos_decode_bbox_pred = distance2bbox(pos_centers, + pos_bbox_pred_corners) + pos_decode_bbox_targets = pos_bbox_targets / stride + bbox_iou = bbox_overlaps( + pos_decode_bbox_pred.detach().numpy(), + pos_decode_bbox_targets.detach().numpy(), + is_aligned=True) + score[pos_inds.numpy()] = bbox_iou + + pred_corners = pos_bbox_pred.reshape([-1, self.reg_max + 1]) + target_corners = bbox2distance(pos_centers, + pos_decode_bbox_targets, + self.reg_max).reshape([-1]) + # regression loss + loss_bbox = paddle.sum( + self.loss_bbox(pos_decode_bbox_pred, + pos_decode_bbox_targets) * weight_targets) + + # dfl loss + loss_dfl = self.loss_dfl( + pred_corners, + target_corners, + weight=weight_targets.expand([-1, 4]).reshape([-1]), + avg_factor=4.0) + else: + loss_bbox = bbox_pred.sum() * 0 + loss_dfl = bbox_pred.sum() * 0 + weight_targets = paddle.to_tensor([0], dtype='float32') + + # qfl loss + score = paddle.to_tensor(score) + loss_qfl = self.loss_qfl( + cls_score, (labels, score), + weight=label_weights, + avg_factor=num_total_pos) + loss_bbox_list.append(loss_bbox) + loss_dfl_list.append(loss_dfl) + loss_qfl_list.append(loss_qfl) + avg_factor.append(weight_targets.sum()) + + avg_factor = sum(avg_factor) + try: + avg_factor = paddle.distributed.all_reduce(avg_factor.clone()) + avg_factor = paddle.clip( + avg_factor / paddle.distributed.get_world_size(), min=1) + except: + avg_factor = max(avg_factor.item(), 1) + if avg_factor <= 0: + loss_qfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + loss_bbox = paddle.to_tensor( + 0, dtype='float32', stop_gradient=False) + loss_dfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + else: + losses_bbox = list(map(lambda x: x / avg_factor, loss_bbox_list)) + losses_dfl = list(map(lambda x: x / avg_factor, loss_dfl_list)) + loss_qfl = sum(loss_qfl_list) + loss_bbox = sum(losses_bbox) + loss_dfl = sum(losses_dfl) + + loss_states = dict( + loss_qfl=loss_qfl, loss_bbox=loss_bbox, loss_dfl=loss_dfl) + + return loss_states + + +@register +class OTAVFLHead(OTAHead): + __inject__ = [ + 'conv_feat', 'dgqp_module', 'loss_class', 'loss_dfl', 'loss_bbox', + 'assigner', 'nms' + ] + __shared__ = ['num_classes'] + + def __init__(self, + conv_feat='FCOSFeat', + dgqp_module=None, + num_classes=80, + fpn_stride=[8, 16, 32, 64, 128], + prior_prob=0.01, + loss_class='VarifocalLoss', + loss_dfl='DistributionFocalLoss', + loss_bbox='GIoULoss', + assigner='SimOTAAssigner', + reg_max=16, + feat_in_chan=256, + nms=None, + nms_pre=1000, + cell_offset=0): + super(OTAVFLHead, self).__init__( + conv_feat=conv_feat, + dgqp_module=dgqp_module, + num_classes=num_classes, + fpn_stride=fpn_stride, + prior_prob=prior_prob, + loss_class=loss_class, + loss_dfl=loss_dfl, + loss_bbox=loss_bbox, + reg_max=reg_max, + feat_in_chan=feat_in_chan, + nms=nms, + nms_pre=nms_pre, + cell_offset=cell_offset) + self.conv_feat = conv_feat + self.dgqp_module = dgqp_module + self.num_classes = num_classes + self.fpn_stride = fpn_stride + self.prior_prob = prior_prob + self.loss_vfl = loss_class + self.loss_dfl = loss_dfl + self.loss_bbox = loss_bbox + self.reg_max = reg_max + self.feat_in_chan = feat_in_chan + self.nms = nms + self.nms_pre = nms_pre + self.cell_offset = cell_offset + self.use_sigmoid = self.loss_vfl.use_sigmoid + + self.assigner = assigner + + def get_loss(self, head_outs, gt_meta): + cls_scores, bbox_preds = head_outs + num_level_anchors = [ + featmap.shape[-2] * featmap.shape[-1] for featmap in cls_scores + ] + num_imgs = gt_meta['im_id'].shape[0] + featmap_sizes = [[featmap.shape[-2], featmap.shape[-1]] + for featmap in cls_scores] + + decode_bbox_preds = [] + center_and_strides = [] + for featmap_size, stride, bbox_pred in zip(featmap_sizes, + self.fpn_stride, bbox_preds): + # center in origin image + yy, xx = self.get_single_level_center_point(featmap_size, stride, + self.cell_offset) + strides = paddle.full((len(xx), ), stride) + center_and_stride = paddle.stack([xx, yy, strides, strides], + -1).tile([num_imgs, 1, 1]) + center_and_strides.append(center_and_stride) + center_in_feature = center_and_stride.reshape( + [-1, 4])[:, :-2] / stride + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]).reshape( + [num_imgs, -1, 4 * (self.reg_max + 1)]) + pred_distances = self.distribution_project(bbox_pred) + decode_bbox_pred_wo_stride = distance2bbox( + center_in_feature, pred_distances).reshape([num_imgs, -1, 4]) + decode_bbox_preds.append(decode_bbox_pred_wo_stride * stride) + + flatten_cls_preds = [ + cls_pred.transpose([0, 2, 3, 1]).reshape( + [num_imgs, -1, self.cls_out_channels]) + for cls_pred in cls_scores + ] + flatten_cls_preds = paddle.concat(flatten_cls_preds, axis=1) + flatten_bboxes = paddle.concat(decode_bbox_preds, axis=1) + flatten_center_and_strides = paddle.concat(center_and_strides, axis=1) + + gt_boxes, gt_labels = gt_meta['gt_bbox'], gt_meta['gt_class'] + pos_num_l, label_l, label_weight_l, bbox_target_l = [], [], [], [] + for flatten_cls_pred, flatten_center_and_stride, flatten_bbox,gt_box,gt_label \ + in zip(flatten_cls_preds.detach(), flatten_center_and_strides.detach(), \ + flatten_bboxes.detach(),gt_boxes,gt_labels): + pos_num, label, label_weight, bbox_target = self._get_target_single( + flatten_cls_pred, flatten_center_and_stride, flatten_bbox, + gt_box, gt_label) + pos_num_l.append(pos_num) + label_l.append(label) + label_weight_l.append(label_weight) + bbox_target_l.append(bbox_target) + + labels = paddle.to_tensor(np.stack(label_l, axis=0)) + label_weights = paddle.to_tensor(np.stack(label_weight_l, axis=0)) + bbox_targets = paddle.to_tensor(np.stack(bbox_target_l, axis=0)) + + center_and_strides_list = self._images_to_levels( + flatten_center_and_strides, num_level_anchors) + labels_list = self._images_to_levels(labels, num_level_anchors) + label_weights_list = self._images_to_levels(label_weights, + num_level_anchors) + bbox_targets_list = self._images_to_levels(bbox_targets, + num_level_anchors) + num_total_pos = sum(pos_num_l) + try: + num_total_pos = paddle.distributed.all_reduce(num_total_pos.clone( + )) / paddle.distributed.get_world_size() + except: + num_total_pos = max(num_total_pos, 1) + + loss_bbox_list, loss_dfl_list, loss_vfl_list, avg_factor = [], [], [], [] + for cls_score, bbox_pred, center_and_strides, labels, label_weights, bbox_targets, stride in zip( + cls_scores, bbox_preds, center_and_strides_list, labels_list, + label_weights_list, bbox_targets_list, self.fpn_stride): + center_and_strides = center_and_strides.reshape([-1, 4]) + cls_score = cls_score.transpose([0, 2, 3, 1]).reshape( + [-1, self.cls_out_channels]) + bbox_pred = bbox_pred.transpose([0, 2, 3, 1]).reshape( + [-1, 4 * (self.reg_max + 1)]) + bbox_targets = bbox_targets.reshape([-1, 4]) + labels = labels.reshape([-1]) + + bg_class_ind = self.num_classes + pos_inds = paddle.nonzero( + paddle.logical_and((labels >= 0), (labels < bg_class_ind)), + as_tuple=False).squeeze(1) + # vfl + vfl_score = np.zeros(cls_score.shape) + + if len(pos_inds) > 0: + pos_bbox_targets = paddle.gather(bbox_targets, pos_inds, axis=0) + pos_bbox_pred = paddle.gather(bbox_pred, pos_inds, axis=0) + pos_centers = paddle.gather( + center_and_strides[:, :-2], pos_inds, axis=0) / stride + + weight_targets = F.sigmoid(cls_score.detach()) + weight_targets = paddle.gather( + weight_targets.max(axis=1, keepdim=True), pos_inds, axis=0) + pos_bbox_pred_corners = self.distribution_project(pos_bbox_pred) + pos_decode_bbox_pred = distance2bbox(pos_centers, + pos_bbox_pred_corners) + pos_decode_bbox_targets = pos_bbox_targets / stride + bbox_iou = bbox_overlaps( + pos_decode_bbox_pred.detach().numpy(), + pos_decode_bbox_targets.detach().numpy(), + is_aligned=True) + + # vfl + pos_labels = paddle.gather(labels, pos_inds, axis=0) + vfl_score[pos_inds.numpy(), pos_labels] = bbox_iou + + pred_corners = pos_bbox_pred.reshape([-1, self.reg_max + 1]) + target_corners = bbox2distance(pos_centers, + pos_decode_bbox_targets, + self.reg_max).reshape([-1]) + # regression loss + loss_bbox = paddle.sum( + self.loss_bbox(pos_decode_bbox_pred, + pos_decode_bbox_targets) * weight_targets) + + # dfl loss + loss_dfl = self.loss_dfl( + pred_corners, + target_corners, + weight=weight_targets.expand([-1, 4]).reshape([-1]), + avg_factor=4.0) + else: + loss_bbox = bbox_pred.sum() * 0 + loss_dfl = bbox_pred.sum() * 0 + weight_targets = paddle.to_tensor([0], dtype='float32') + + # vfl loss + num_pos_avg_per_gpu = num_total_pos + vfl_score = paddle.to_tensor(vfl_score) + loss_vfl = self.loss_vfl( + cls_score, vfl_score, avg_factor=num_pos_avg_per_gpu) + + loss_bbox_list.append(loss_bbox) + loss_dfl_list.append(loss_dfl) + loss_vfl_list.append(loss_vfl) + avg_factor.append(weight_targets.sum()) + + avg_factor = sum(avg_factor) + try: + avg_factor = paddle.distributed.all_reduce(avg_factor.clone()) + avg_factor = paddle.clip( + avg_factor / paddle.distributed.get_world_size(), min=1) + except: + avg_factor = max(avg_factor.item(), 1) + if avg_factor <= 0: + loss_vfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + loss_bbox = paddle.to_tensor( + 0, dtype='float32', stop_gradient=False) + loss_dfl = paddle.to_tensor(0, dtype='float32', stop_gradient=False) + else: + losses_bbox = list(map(lambda x: x / avg_factor, loss_bbox_list)) + losses_dfl = list(map(lambda x: x / avg_factor, loss_dfl_list)) + loss_vfl = sum(loss_vfl_list) + loss_bbox = sum(losses_bbox) + loss_dfl = sum(losses_dfl) + + loss_states = dict( + loss_vfl=loss_vfl, loss_bbox=loss_bbox, loss_dfl=loss_dfl) + + return loss_states diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/solov2_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/solov2_head.py new file mode 100644 index 000000000..6989abb3a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/solov2_head.py @@ -0,0 +1,554 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from paddle import ParamAttr +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, Constant + +from ppdet.modeling.layers import ConvNormLayer, MaskMatrixNMS, DropBlock +from ppdet.core.workspace import register + +from six.moves import zip +import numpy as np + +__all__ = ['SOLOv2Head'] + + +@register +class SOLOv2MaskHead(nn.Layer): + """ + MaskHead of SOLOv2. + The code of this function is based on: + https://github.com/WXinlong/SOLO/blob/master/mmdet/models/mask_heads/mask_feat_head.py + + Args: + in_channels (int): The channel number of input Tensor. + out_channels (int): The channel number of output Tensor. + start_level (int): The position where the input starts. + end_level (int): The position where the input ends. + use_dcn_in_tower (bool): Whether to use dcn in tower or not. + """ + __shared__ = ['norm_type'] + + def __init__(self, + in_channels=256, + mid_channels=128, + out_channels=256, + start_level=0, + end_level=3, + use_dcn_in_tower=False, + norm_type='gn'): + super(SOLOv2MaskHead, self).__init__() + assert start_level >= 0 and end_level >= start_level + self.in_channels = in_channels + self.out_channels = out_channels + self.mid_channels = mid_channels + self.use_dcn_in_tower = use_dcn_in_tower + self.range_level = end_level - start_level + 1 + self.use_dcn = True if self.use_dcn_in_tower else False + self.convs_all_levels = [] + self.norm_type = norm_type + for i in range(start_level, end_level + 1): + conv_feat_name = 'mask_feat_head.convs_all_levels.{}'.format(i) + conv_pre_feat = nn.Sequential() + if i == start_level: + conv_pre_feat.add_sublayer( + conv_feat_name + '.conv' + str(i), + ConvNormLayer( + ch_in=self.in_channels, + ch_out=self.mid_channels, + filter_size=3, + stride=1, + use_dcn=self.use_dcn, + norm_type=self.norm_type)) + self.add_sublayer('conv_pre_feat' + str(i), conv_pre_feat) + self.convs_all_levels.append(conv_pre_feat) + else: + for j in range(i): + ch_in = 0 + if j == 0: + ch_in = self.in_channels + 2 if i == end_level else self.in_channels + else: + ch_in = self.mid_channels + conv_pre_feat.add_sublayer( + conv_feat_name + '.conv' + str(j), + ConvNormLayer( + ch_in=ch_in, + ch_out=self.mid_channels, + filter_size=3, + stride=1, + use_dcn=self.use_dcn, + norm_type=self.norm_type)) + conv_pre_feat.add_sublayer( + conv_feat_name + '.conv' + str(j) + 'act', nn.ReLU()) + conv_pre_feat.add_sublayer( + 'upsample' + str(i) + str(j), + nn.Upsample( + scale_factor=2, mode='bilinear')) + self.add_sublayer('conv_pre_feat' + str(i), conv_pre_feat) + self.convs_all_levels.append(conv_pre_feat) + + conv_pred_name = 'mask_feat_head.conv_pred.0' + self.conv_pred = self.add_sublayer( + conv_pred_name, + ConvNormLayer( + ch_in=self.mid_channels, + ch_out=self.out_channels, + filter_size=1, + stride=1, + use_dcn=self.use_dcn, + norm_type=self.norm_type)) + + def forward(self, inputs): + """ + Get SOLOv2MaskHead output. + + Args: + inputs(list[Tensor]): feature map from each necks with shape of [N, C, H, W] + Returns: + ins_pred(Tensor): Output of SOLOv2MaskHead head + """ + feat_all_level = F.relu(self.convs_all_levels[0](inputs[0])) + for i in range(1, self.range_level): + input_p = inputs[i] + if i == (self.range_level - 1): + input_feat = input_p + x_range = paddle.linspace( + -1, 1, paddle.shape(input_feat)[-1], dtype='float32') + y_range = paddle.linspace( + -1, 1, paddle.shape(input_feat)[-2], dtype='float32') + y, x = paddle.meshgrid([y_range, x_range]) + x = paddle.unsqueeze(x, [0, 1]) + y = paddle.unsqueeze(y, [0, 1]) + y = paddle.expand( + y, shape=[paddle.shape(input_feat)[0], 1, -1, -1]) + x = paddle.expand( + x, shape=[paddle.shape(input_feat)[0], 1, -1, -1]) + coord_feat = paddle.concat([x, y], axis=1) + input_p = paddle.concat([input_p, coord_feat], axis=1) + feat_all_level = paddle.add(feat_all_level, + self.convs_all_levels[i](input_p)) + ins_pred = F.relu(self.conv_pred(feat_all_level)) + + return ins_pred + + +@register +class SOLOv2Head(nn.Layer): + """ + Head block for SOLOv2 network + + Args: + num_classes (int): Number of output classes. + in_channels (int): Number of input channels. + seg_feat_channels (int): Num_filters of kernel & categroy branch convolution operation. + stacked_convs (int): Times of convolution operation. + num_grids (list[int]): List of feature map grids size. + kernel_out_channels (int): Number of output channels in kernel branch. + dcn_v2_stages (list): Which stage use dcn v2 in tower. It is between [0, stacked_convs). + segm_strides (list[int]): List of segmentation area stride. + solov2_loss (object): SOLOv2Loss instance. + score_threshold (float): Threshold of categroy score. + mask_nms (object): MaskMatrixNMS instance. + """ + __inject__ = ['solov2_loss', 'mask_nms'] + __shared__ = ['norm_type', 'num_classes'] + + def __init__(self, + num_classes=80, + in_channels=256, + seg_feat_channels=256, + stacked_convs=4, + num_grids=[40, 36, 24, 16, 12], + kernel_out_channels=256, + dcn_v2_stages=[], + segm_strides=[8, 8, 16, 32, 32], + solov2_loss=None, + score_threshold=0.1, + mask_threshold=0.5, + mask_nms=None, + norm_type='gn', + drop_block=False): + super(SOLOv2Head, self).__init__() + self.num_classes = num_classes + self.in_channels = in_channels + self.seg_num_grids = num_grids + self.cate_out_channels = self.num_classes + self.seg_feat_channels = seg_feat_channels + self.stacked_convs = stacked_convs + self.kernel_out_channels = kernel_out_channels + self.dcn_v2_stages = dcn_v2_stages + self.segm_strides = segm_strides + self.solov2_loss = solov2_loss + self.mask_nms = mask_nms + self.score_threshold = score_threshold + self.mask_threshold = mask_threshold + self.norm_type = norm_type + self.drop_block = drop_block + + self.kernel_pred_convs = [] + self.cate_pred_convs = [] + for i in range(self.stacked_convs): + use_dcn = True if i in self.dcn_v2_stages else False + ch_in = self.in_channels + 2 if i == 0 else self.seg_feat_channels + kernel_conv = self.add_sublayer( + 'bbox_head.kernel_convs.' + str(i), + ConvNormLayer( + ch_in=ch_in, + ch_out=self.seg_feat_channels, + filter_size=3, + stride=1, + use_dcn=use_dcn, + norm_type=self.norm_type)) + self.kernel_pred_convs.append(kernel_conv) + ch_in = self.in_channels if i == 0 else self.seg_feat_channels + cate_conv = self.add_sublayer( + 'bbox_head.cate_convs.' + str(i), + ConvNormLayer( + ch_in=ch_in, + ch_out=self.seg_feat_channels, + filter_size=3, + stride=1, + use_dcn=use_dcn, + norm_type=self.norm_type)) + self.cate_pred_convs.append(cate_conv) + + self.solo_kernel = self.add_sublayer( + 'bbox_head.solo_kernel', + nn.Conv2D( + self.seg_feat_channels, + self.kernel_out_channels, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=True)) + self.solo_cate = self.add_sublayer( + 'bbox_head.solo_cate', + nn.Conv2D( + self.seg_feat_channels, + self.cate_out_channels, + kernel_size=3, + stride=1, + padding=1, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.01)), + bias_attr=ParamAttr(initializer=Constant( + value=float(-np.log((1 - 0.01) / 0.01)))))) + + if self.drop_block and self.training: + self.drop_block_fun = DropBlock( + block_size=3, keep_prob=0.9, name='solo_cate.dropblock') + + def _points_nms(self, heat, kernel_size=2): + hmax = F.max_pool2d(heat, kernel_size=kernel_size, stride=1, padding=1) + keep = paddle.cast((hmax[:, :, :-1, :-1] == heat), 'float32') + return heat * keep + + def _split_feats(self, feats): + return (F.interpolate( + feats[0], + scale_factor=0.5, + align_corners=False, + align_mode=0, + mode='bilinear'), feats[1], feats[2], feats[3], F.interpolate( + feats[4], + size=paddle.shape(feats[3])[-2:], + mode='bilinear', + align_corners=False, + align_mode=0)) + + def forward(self, input): + """ + Get SOLOv2 head output + + Args: + input (list): List of Tensors, output of backbone or neck stages + Returns: + cate_pred_list (list): Tensors of each category branch layer + kernel_pred_list (list): Tensors of each kernel branch layer + """ + feats = self._split_feats(input) + cate_pred_list = [] + kernel_pred_list = [] + for idx in range(len(self.seg_num_grids)): + cate_pred, kernel_pred = self._get_output_single(feats[idx], idx) + cate_pred_list.append(cate_pred) + kernel_pred_list.append(kernel_pred) + + return cate_pred_list, kernel_pred_list + + def _get_output_single(self, input, idx): + ins_kernel_feat = input + # CoordConv + x_range = paddle.linspace( + -1, 1, paddle.shape(ins_kernel_feat)[-1], dtype='float32') + y_range = paddle.linspace( + -1, 1, paddle.shape(ins_kernel_feat)[-2], dtype='float32') + y, x = paddle.meshgrid([y_range, x_range]) + x = paddle.unsqueeze(x, [0, 1]) + y = paddle.unsqueeze(y, [0, 1]) + y = paddle.expand( + y, shape=[paddle.shape(ins_kernel_feat)[0], 1, -1, -1]) + x = paddle.expand( + x, shape=[paddle.shape(ins_kernel_feat)[0], 1, -1, -1]) + coord_feat = paddle.concat([x, y], axis=1) + ins_kernel_feat = paddle.concat([ins_kernel_feat, coord_feat], axis=1) + + # kernel branch + kernel_feat = ins_kernel_feat + seg_num_grid = self.seg_num_grids[idx] + kernel_feat = F.interpolate( + kernel_feat, + size=[seg_num_grid, seg_num_grid], + mode='bilinear', + align_corners=False, + align_mode=0) + cate_feat = kernel_feat[:, :-2, :, :] + + for kernel_layer in self.kernel_pred_convs: + kernel_feat = F.relu(kernel_layer(kernel_feat)) + if self.drop_block and self.training: + kernel_feat = self.drop_block_fun(kernel_feat) + kernel_pred = self.solo_kernel(kernel_feat) + # cate branch + for cate_layer in self.cate_pred_convs: + cate_feat = F.relu(cate_layer(cate_feat)) + if self.drop_block and self.training: + cate_feat = self.drop_block_fun(cate_feat) + cate_pred = self.solo_cate(cate_feat) + + if not self.training: + cate_pred = self._points_nms(F.sigmoid(cate_pred), kernel_size=2) + cate_pred = paddle.transpose(cate_pred, [0, 2, 3, 1]) + return cate_pred, kernel_pred + + def get_loss(self, cate_preds, kernel_preds, ins_pred, ins_labels, + cate_labels, grid_order_list, fg_num): + """ + Get loss of network of SOLOv2. + + Args: + cate_preds (list): Tensor list of categroy branch output. + kernel_preds (list): Tensor list of kernel branch output. + ins_pred (list): Tensor list of instance branch output. + ins_labels (list): List of instance labels pre batch. + cate_labels (list): List of categroy labels pre batch. + grid_order_list (list): List of index in pre grid. + fg_num (int): Number of positive samples in a mini-batch. + Returns: + loss_ins (Tensor): The instance loss Tensor of SOLOv2 network. + loss_cate (Tensor): The category loss Tensor of SOLOv2 network. + """ + batch_size = paddle.shape(grid_order_list[0])[0] + ins_pred_list = [] + for kernel_preds_level, grid_orders_level in zip(kernel_preds, + grid_order_list): + if grid_orders_level.shape[1] == 0: + ins_pred_list.append(None) + continue + grid_orders_level = paddle.reshape(grid_orders_level, [-1]) + reshape_pred = paddle.reshape( + kernel_preds_level, + shape=(paddle.shape(kernel_preds_level)[0], + paddle.shape(kernel_preds_level)[1], -1)) + reshape_pred = paddle.transpose(reshape_pred, [0, 2, 1]) + reshape_pred = paddle.reshape( + reshape_pred, shape=(-1, paddle.shape(reshape_pred)[2])) + gathered_pred = paddle.gather(reshape_pred, index=grid_orders_level) + gathered_pred = paddle.reshape( + gathered_pred, + shape=[batch_size, -1, paddle.shape(gathered_pred)[1]]) + cur_ins_pred = ins_pred + cur_ins_pred = paddle.reshape( + cur_ins_pred, + shape=(paddle.shape(cur_ins_pred)[0], + paddle.shape(cur_ins_pred)[1], -1)) + ins_pred_conv = paddle.matmul(gathered_pred, cur_ins_pred) + cur_ins_pred = paddle.reshape( + ins_pred_conv, + shape=(-1, paddle.shape(ins_pred)[-2], + paddle.shape(ins_pred)[-1])) + ins_pred_list.append(cur_ins_pred) + + num_ins = paddle.sum(fg_num) + cate_preds = [ + paddle.reshape( + paddle.transpose(cate_pred, [0, 2, 3, 1]), + shape=(-1, self.cate_out_channels)) for cate_pred in cate_preds + ] + flatten_cate_preds = paddle.concat(cate_preds) + new_cate_labels = [] + for cate_label in cate_labels: + new_cate_labels.append(paddle.reshape(cate_label, shape=[-1])) + cate_labels = paddle.concat(new_cate_labels) + + loss_ins, loss_cate = self.solov2_loss( + ins_pred_list, ins_labels, flatten_cate_preds, cate_labels, num_ins) + + return {'loss_ins': loss_ins, 'loss_cate': loss_cate} + + def get_prediction(self, cate_preds, kernel_preds, seg_pred, im_shape, + scale_factor): + """ + Get prediction result of SOLOv2 network + + Args: + cate_preds (list): List of Variables, output of categroy branch. + kernel_preds (list): List of Variables, output of kernel branch. + seg_pred (list): List of Variables, output of mask head stages. + im_shape (Variables): [h, w] for input images. + scale_factor (Variables): [scale, scale] for input images. + Returns: + seg_masks (Tensor): The prediction segmentation. + cate_labels (Tensor): The prediction categroy label of each segmentation. + seg_masks (Tensor): The prediction score of each segmentation. + """ + num_levels = len(cate_preds) + featmap_size = paddle.shape(seg_pred)[-2:] + seg_masks_list = [] + cate_labels_list = [] + cate_scores_list = [] + cate_preds = [cate_pred * 1.0 for cate_pred in cate_preds] + kernel_preds = [kernel_pred * 1.0 for kernel_pred in kernel_preds] + # Currently only supports batch size == 1 + for idx in range(1): + cate_pred_list = [ + paddle.reshape( + cate_preds[i][idx], shape=(-1, self.cate_out_channels)) + for i in range(num_levels) + ] + seg_pred_list = seg_pred + kernel_pred_list = [ + paddle.reshape( + paddle.transpose(kernel_preds[i][idx], [1, 2, 0]), + shape=(-1, self.kernel_out_channels)) + for i in range(num_levels) + ] + cate_pred_list = paddle.concat(cate_pred_list, axis=0) + kernel_pred_list = paddle.concat(kernel_pred_list, axis=0) + + seg_masks, cate_labels, cate_scores = self.get_seg_single( + cate_pred_list, seg_pred_list, kernel_pred_list, featmap_size, + im_shape[idx], scale_factor[idx][0]) + bbox_num = paddle.shape(cate_labels)[0] + return seg_masks, cate_labels, cate_scores, bbox_num + + def get_seg_single(self, cate_preds, seg_preds, kernel_preds, featmap_size, + im_shape, scale_factor): + """ + The code of this function is based on: + https://github.com/WXinlong/SOLO/blob/master/mmdet/models/anchor_heads/solov2_head.py#L385 + """ + h = paddle.cast(im_shape[0], 'int32')[0] + w = paddle.cast(im_shape[1], 'int32')[0] + upsampled_size_out = [featmap_size[0] * 4, featmap_size[1] * 4] + + y = paddle.zeros(shape=paddle.shape(cate_preds), dtype='float32') + inds = paddle.where(cate_preds > self.score_threshold, cate_preds, y) + inds = paddle.nonzero(inds) + cate_preds = paddle.reshape(cate_preds, shape=[-1]) + # Prevent empty and increase fake data + ind_a = paddle.cast(paddle.shape(kernel_preds)[0], 'int64') + ind_b = paddle.zeros(shape=[1], dtype='int64') + inds_end = paddle.unsqueeze(paddle.concat([ind_a, ind_b]), 0) + inds = paddle.concat([inds, inds_end]) + kernel_preds_end = paddle.ones( + shape=[1, self.kernel_out_channels], dtype='float32') + kernel_preds = paddle.concat([kernel_preds, kernel_preds_end]) + cate_preds = paddle.concat( + [cate_preds, paddle.zeros( + shape=[1], dtype='float32')]) + + # cate_labels & kernel_preds + cate_labels = inds[:, 1] + kernel_preds = paddle.gather(kernel_preds, index=inds[:, 0]) + cate_score_idx = paddle.add(inds[:, 0] * self.cate_out_channels, + cate_labels) + cate_scores = paddle.gather(cate_preds, index=cate_score_idx) + + size_trans = np.power(self.seg_num_grids, 2) + strides = [] + for _ind in range(len(self.segm_strides)): + strides.append( + paddle.full( + shape=[int(size_trans[_ind])], + fill_value=self.segm_strides[_ind], + dtype="int32")) + strides = paddle.concat(strides) + strides = paddle.concat( + [strides, paddle.zeros( + shape=[1], dtype='int32')]) + strides = paddle.gather(strides, index=inds[:, 0]) + + # mask encoding. + kernel_preds = paddle.unsqueeze(kernel_preds, [2, 3]) + seg_preds = F.conv2d(seg_preds, kernel_preds) + seg_preds = F.sigmoid(paddle.squeeze(seg_preds, [0])) + seg_masks = seg_preds > self.mask_threshold + seg_masks = paddle.cast(seg_masks, 'float32') + sum_masks = paddle.sum(seg_masks, axis=[1, 2]) + + y = paddle.zeros(shape=paddle.shape(sum_masks), dtype='float32') + keep = paddle.where(sum_masks > strides, sum_masks, y) + keep = paddle.nonzero(keep) + keep = paddle.squeeze(keep, axis=[1]) + # Prevent empty and increase fake data + keep_other = paddle.concat( + [keep, paddle.cast(paddle.shape(sum_masks)[0] - 1, 'int64')]) + keep_scores = paddle.concat( + [keep, paddle.cast(paddle.shape(sum_masks)[0], 'int64')]) + cate_scores_end = paddle.zeros(shape=[1], dtype='float32') + cate_scores = paddle.concat([cate_scores, cate_scores_end]) + + seg_masks = paddle.gather(seg_masks, index=keep_other) + seg_preds = paddle.gather(seg_preds, index=keep_other) + sum_masks = paddle.gather(sum_masks, index=keep_other) + cate_labels = paddle.gather(cate_labels, index=keep_other) + cate_scores = paddle.gather(cate_scores, index=keep_scores) + + # mask scoring. + seg_mul = paddle.cast(seg_preds * seg_masks, 'float32') + seg_scores = paddle.sum(seg_mul, axis=[1, 2]) / sum_masks + cate_scores *= seg_scores + # Matrix NMS + seg_preds, cate_scores, cate_labels = self.mask_nms( + seg_preds, seg_masks, cate_labels, cate_scores, sum_masks=sum_masks) + ori_shape = im_shape[:2] / scale_factor + 0.5 + ori_shape = paddle.cast(ori_shape, 'int32') + seg_preds = F.interpolate( + paddle.unsqueeze(seg_preds, 0), + size=upsampled_size_out, + mode='bilinear', + align_corners=False, + align_mode=0) + seg_preds = paddle.slice( + seg_preds, axes=[2, 3], starts=[0, 0], ends=[h, w]) + seg_masks = paddle.squeeze( + F.interpolate( + seg_preds, + size=ori_shape[:2], + mode='bilinear', + align_corners=False, + align_mode=0), + axis=[0]) + seg_masks = paddle.cast(seg_masks > self.mask_threshold, 'uint8') + return seg_masks, cate_labels, cate_scores diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/sparsercnn_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/sparsercnn_head.py new file mode 100644 index 000000000..377cf27fc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/sparsercnn_head.py @@ -0,0 +1,375 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/PeizeSun/SparseR-CNN/blob/main/projects/SparseRCNN/sparsercnn/head.py +Ths copyright of PeizeSun/SparseR-CNN is as follows: +MIT License [see LICENSE for details] +""" + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import copy +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register +from ppdet.modeling.heads.roi_extractor import RoIAlign +from ppdet.modeling.bbox_utils import delta2bbox +from .. import initializer as init + +_DEFAULT_SCALE_CLAMP = math.log(100000. / 16) + + +class DynamicConv(nn.Layer): + def __init__( + self, + head_hidden_dim, + head_dim_dynamic, + head_num_dynamic, ): + super().__init__() + + self.hidden_dim = head_hidden_dim + self.dim_dynamic = head_dim_dynamic + self.num_dynamic = head_num_dynamic + self.num_params = self.hidden_dim * self.dim_dynamic + self.dynamic_layer = nn.Linear(self.hidden_dim, + self.num_dynamic * self.num_params) + + self.norm1 = nn.LayerNorm(self.dim_dynamic) + self.norm2 = nn.LayerNorm(self.hidden_dim) + + self.activation = nn.ReLU() + + pooler_resolution = 7 + num_output = self.hidden_dim * pooler_resolution**2 + self.out_layer = nn.Linear(num_output, self.hidden_dim) + self.norm3 = nn.LayerNorm(self.hidden_dim) + + def forward(self, pro_features, roi_features): + ''' + pro_features: (1, N * nr_boxes, self.d_model) + roi_features: (49, N * nr_boxes, self.d_model) + ''' + features = roi_features.transpose(perm=[1, 0, 2]) + parameters = self.dynamic_layer(pro_features).transpose(perm=[1, 0, 2]) + + param1 = parameters[:, :, :self.num_params].reshape( + [-1, self.hidden_dim, self.dim_dynamic]) + param2 = parameters[:, :, self.num_params:].reshape( + [-1, self.dim_dynamic, self.hidden_dim]) + + features = paddle.bmm(features, param1) + features = self.norm1(features) + features = self.activation(features) + + features = paddle.bmm(features, param2) + features = self.norm2(features) + features = self.activation(features) + + features = features.flatten(1) + features = self.out_layer(features) + features = self.norm3(features) + features = self.activation(features) + + return features + + +class RCNNHead(nn.Layer): + def __init__( + self, + d_model, + num_classes, + dim_feedforward, + nhead, + dropout, + head_cls, + head_reg, + head_dim_dynamic, + head_num_dynamic, + scale_clamp: float=_DEFAULT_SCALE_CLAMP, + bbox_weights=(2.0, 2.0, 1.0, 1.0), ): + super().__init__() + + self.d_model = d_model + + # dynamic. + self.self_attn = nn.MultiHeadAttention(d_model, nhead, dropout=dropout) + self.inst_interact = DynamicConv(d_model, head_dim_dynamic, + head_num_dynamic) + + self.linear1 = nn.Linear(d_model, dim_feedforward) + self.dropout = nn.Dropout(dropout) + self.linear2 = nn.Linear(dim_feedforward, d_model) + + self.norm1 = nn.LayerNorm(d_model) + self.norm2 = nn.LayerNorm(d_model) + self.norm3 = nn.LayerNorm(d_model) + self.dropout1 = nn.Dropout(dropout) + self.dropout2 = nn.Dropout(dropout) + self.dropout3 = nn.Dropout(dropout) + + self.activation = nn.ReLU() + + # cls. + num_cls = head_cls + cls_module = list() + for _ in range(num_cls): + cls_module.append(nn.Linear(d_model, d_model, bias_attr=False)) + cls_module.append(nn.LayerNorm(d_model)) + cls_module.append(nn.ReLU()) + self.cls_module = nn.LayerList(cls_module) + + # reg. + num_reg = head_reg + reg_module = list() + for _ in range(num_reg): + reg_module.append(nn.Linear(d_model, d_model, bias_attr=False)) + reg_module.append(nn.LayerNorm(d_model)) + reg_module.append(nn.ReLU()) + self.reg_module = nn.LayerList(reg_module) + + # pred. + self.class_logits = nn.Linear(d_model, num_classes) + self.bboxes_delta = nn.Linear(d_model, 4) + self.scale_clamp = scale_clamp + self.bbox_weights = bbox_weights + + def forward(self, features, bboxes, pro_features, pooler): + """ + :param bboxes: (N, nr_boxes, 4) + :param pro_features: (N, nr_boxes, d_model) + """ + + N, nr_boxes = bboxes.shape[:2] + + proposal_boxes = list() + for b in range(N): + proposal_boxes.append(bboxes[b]) + roi_num = paddle.full([N], nr_boxes).astype("int32") + + roi_features = pooler(features, proposal_boxes, roi_num) + roi_features = roi_features.reshape( + [N * nr_boxes, self.d_model, -1]).transpose(perm=[2, 0, 1]) + + # self_att. + pro_features = pro_features.reshape([N, nr_boxes, self.d_model]) + pro_features2 = self.self_attn( + pro_features, pro_features, value=pro_features) + pro_features = pro_features.transpose(perm=[1, 0, 2]) + self.dropout1( + pro_features2.transpose(perm=[1, 0, 2])) + pro_features = self.norm1(pro_features) + + # inst_interact. + pro_features = pro_features.reshape( + [nr_boxes, N, self.d_model]).transpose(perm=[1, 0, 2]).reshape( + [1, N * nr_boxes, self.d_model]) + pro_features2 = self.inst_interact(pro_features, roi_features) + pro_features = pro_features + self.dropout2(pro_features2) + obj_features = self.norm2(pro_features) + + # obj_feature. + obj_features2 = self.linear2( + self.dropout(self.activation(self.linear1(obj_features)))) + obj_features = obj_features + self.dropout3(obj_features2) + obj_features = self.norm3(obj_features) + + fc_feature = obj_features.transpose(perm=[1, 0, 2]).reshape( + [N * nr_boxes, -1]) + cls_feature = fc_feature.clone() + reg_feature = fc_feature.clone() + for cls_layer in self.cls_module: + cls_feature = cls_layer(cls_feature) + for reg_layer in self.reg_module: + reg_feature = reg_layer(reg_feature) + class_logits = self.class_logits(cls_feature) + bboxes_deltas = self.bboxes_delta(reg_feature) + pred_bboxes = delta2bbox(bboxes_deltas, + bboxes.reshape([-1, 4]), self.bbox_weights) + + return class_logits.reshape([N, nr_boxes, -1]), pred_bboxes.reshape( + [N, nr_boxes, -1]), obj_features + + +@register +class SparseRCNNHead(nn.Layer): + ''' + SparsercnnHead + Args: + roi_input_shape (list[ShapeSpec]): The output shape of fpn + num_classes (int): Number of classes, + head_hidden_dim (int): The param of MultiHeadAttention, + head_dim_feedforward (int): The param of MultiHeadAttention, + nhead (int): The param of MultiHeadAttention, + head_dropout (float): The p of dropout, + head_cls (int): The number of class head, + head_reg (int): The number of regressionhead, + head_num_dynamic (int): The number of DynamicConv's param, + head_num_heads (int): The number of RCNNHead, + deep_supervision (int): wheather supervise the intermediate results, + num_proposals (int): the number of proposals boxes and features + ''' + __inject__ = ['loss_func'] + __shared__ = ['num_classes'] + + def __init__( + self, + head_hidden_dim, + head_dim_feedforward, + nhead, + head_dropout, + head_cls, + head_reg, + head_dim_dynamic, + head_num_dynamic, + head_num_heads, + deep_supervision, + num_proposals, + num_classes=80, + loss_func="SparseRCNNLoss", + roi_input_shape=None, ): + super().__init__() + + # Build RoI. + box_pooler = self._init_box_pooler(roi_input_shape) + self.box_pooler = box_pooler + + # Build heads. + rcnn_head = RCNNHead( + head_hidden_dim, + num_classes, + head_dim_feedforward, + nhead, + head_dropout, + head_cls, + head_reg, + head_dim_dynamic, + head_num_dynamic, ) + self.head_series = nn.LayerList( + [copy.deepcopy(rcnn_head) for i in range(head_num_heads)]) + self.return_intermediate = deep_supervision + + self.num_classes = num_classes + + # build init proposal + self.init_proposal_features = nn.Embedding(num_proposals, + head_hidden_dim) + self.init_proposal_boxes = nn.Embedding(num_proposals, 4) + + self.lossfunc = loss_func + + # Init parameters. + init.reset_initialized_parameter(self) + self._reset_parameters() + + def _reset_parameters(self): + # init all parameters. + prior_prob = 0.01 + bias_value = -math.log((1 - prior_prob) / prior_prob) + + for m in self.sublayers(): + if isinstance(m, nn.Linear): + init.xavier_normal_(m.weight, reverse=True) + elif not isinstance(m, nn.Embedding) and hasattr( + m, "weight") and m.weight.dim() > 1: + init.xavier_normal_(m.weight, reverse=False) + + if hasattr(m, "bias") and m.bias is not None and m.bias.shape[ + -1] == self.num_classes: + init.constant_(m.bias, bias_value) + + init_bboxes = paddle.empty_like(self.init_proposal_boxes.weight) + init_bboxes[:, :2] = 0.5 + init_bboxes[:, 2:] = 1.0 + self.init_proposal_boxes.weight.set_value(init_bboxes) + + @staticmethod + def _init_box_pooler(input_shape): + + pooler_resolution = 7 + sampling_ratio = 2 + + if input_shape is not None: + pooler_scales = tuple(1.0 / input_shape[k].stride + for k in range(len(input_shape))) + in_channels = [ + input_shape[f].channels for f in range(len(input_shape)) + ] + end_level = len(input_shape) - 1 + # Check all channel counts are equal + assert len(set(in_channels)) == 1, in_channels + else: + pooler_scales = [1.0 / 4.0, 1.0 / 8.0, 1.0 / 16.0, 1.0 / 32.0] + end_level = 3 + + box_pooler = RoIAlign( + resolution=pooler_resolution, + spatial_scale=pooler_scales, + sampling_ratio=sampling_ratio, + end_level=end_level, + aligned=True) + return box_pooler + + def forward(self, features, input_whwh): + + bs = len(features[0]) + bboxes = box_cxcywh_to_xyxy(self.init_proposal_boxes.weight.clone( + )).unsqueeze(0) + bboxes = bboxes * input_whwh.unsqueeze(-2) + + init_features = self.init_proposal_features.weight.unsqueeze(0).tile( + [1, bs, 1]) + proposal_features = init_features.clone() + + inter_class_logits = [] + inter_pred_bboxes = [] + + for rcnn_head in self.head_series: + class_logits, pred_bboxes, proposal_features = rcnn_head( + features, bboxes, proposal_features, self.box_pooler) + + if self.return_intermediate: + inter_class_logits.append(class_logits) + inter_pred_bboxes.append(pred_bboxes) + bboxes = pred_bboxes.detach() + + output = { + 'pred_logits': inter_class_logits[-1], + 'pred_boxes': inter_pred_bboxes[-1] + } + if self.return_intermediate: + output['aux_outputs'] = [{ + 'pred_logits': a, + 'pred_boxes': b + } for a, b in zip(inter_class_logits[:-1], inter_pred_bboxes[:-1])] + + return output + + def get_loss(self, outputs, targets): + losses = self.lossfunc(outputs, targets) + weight_dict = self.lossfunc.weight_dict + + for k in losses.keys(): + if k in weight_dict: + losses[k] *= weight_dict[k] + + return losses + + +def box_cxcywh_to_xyxy(x): + x_c, y_c, w, h = x.unbind(-1) + b = [(x_c - 0.5 * w), (y_c - 0.5 * h), (x_c + 0.5 * w), (y_c + 0.5 * h)] + return paddle.stack(b, axis=-1) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ssd_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ssd_head.py new file mode 100644 index 000000000..07e7e92f9 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ssd_head.py @@ -0,0 +1,215 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +from paddle.regularizer import L2Decay +from paddle import ParamAttr + +from ..layers import AnchorGeneratorSSD + + +class SepConvLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size=3, + padding=1, + conv_decay=0.): + super(SepConvLayer, self).__init__() + self.dw_conv = nn.Conv2D( + in_channels=in_channels, + out_channels=in_channels, + kernel_size=kernel_size, + stride=1, + padding=padding, + groups=in_channels, + weight_attr=ParamAttr(regularizer=L2Decay(conv_decay)), + bias_attr=False) + + self.bn = nn.BatchNorm2D( + in_channels, + weight_attr=ParamAttr(regularizer=L2Decay(0.)), + bias_attr=ParamAttr(regularizer=L2Decay(0.))) + + self.pw_conv = nn.Conv2D( + in_channels=in_channels, + out_channels=out_channels, + kernel_size=1, + stride=1, + padding=0, + weight_attr=ParamAttr(regularizer=L2Decay(conv_decay)), + bias_attr=False) + + def forward(self, x): + x = self.dw_conv(x) + x = F.relu6(self.bn(x)) + x = self.pw_conv(x) + return x + + +class SSDExtraHead(nn.Layer): + def __init__(self, + in_channels=256, + out_channels=([256, 512], [256, 512], [128, 256], [128, 256], + [128, 256]), + strides=(2, 2, 2, 1, 1), + paddings=(1, 1, 1, 0, 0)): + super(SSDExtraHead, self).__init__() + self.convs = nn.LayerList() + for out_channel, stride, padding in zip(out_channels, strides, + paddings): + self.convs.append( + self._make_layers(in_channels, out_channel[0], out_channel[1], + stride, padding)) + in_channels = out_channel[-1] + + def _make_layers(self, c_in, c_hidden, c_out, stride_3x3, padding_3x3): + return nn.Sequential( + nn.Conv2D(c_in, c_hidden, 1), + nn.ReLU(), + nn.Conv2D(c_hidden, c_out, 3, stride_3x3, padding_3x3), nn.ReLU()) + + def forward(self, x): + out = [x] + for conv_layer in self.convs: + out.append(conv_layer(out[-1])) + return out + + +@register +class SSDHead(nn.Layer): + """ + SSDHead + + Args: + num_classes (int): Number of classes + in_channels (list): Number of channels per input feature + anchor_generator (dict): Configuration of 'AnchorGeneratorSSD' instance + kernel_size (int): Conv kernel size + padding (int): Conv padding + use_sepconv (bool): Use SepConvLayer if true + conv_decay (float): Conv regularization coeff + loss (object): 'SSDLoss' instance + use_extra_head (bool): If use ResNet34 as baskbone, you should set `use_extra_head`=True + """ + + __shared__ = ['num_classes'] + __inject__ = ['anchor_generator', 'loss'] + + def __init__(self, + num_classes=80, + in_channels=(512, 1024, 512, 256, 256, 256), + anchor_generator=AnchorGeneratorSSD().__dict__, + kernel_size=3, + padding=1, + use_sepconv=False, + conv_decay=0., + loss='SSDLoss', + use_extra_head=False): + super(SSDHead, self).__init__() + # add background class + self.num_classes = num_classes + 1 + self.in_channels = in_channels + self.anchor_generator = anchor_generator + self.loss = loss + self.use_extra_head = use_extra_head + + if self.use_extra_head: + self.ssd_extra_head = SSDExtraHead() + self.in_channels = [256, 512, 512, 256, 256, 256] + + if isinstance(anchor_generator, dict): + self.anchor_generator = AnchorGeneratorSSD(**anchor_generator) + + self.num_priors = self.anchor_generator.num_priors + self.box_convs = [] + self.score_convs = [] + for i, num_prior in enumerate(self.num_priors): + box_conv_name = "boxes{}".format(i) + if not use_sepconv: + box_conv = self.add_sublayer( + box_conv_name, + nn.Conv2D( + in_channels=self.in_channels[i], + out_channels=num_prior * 4, + kernel_size=kernel_size, + padding=padding)) + else: + box_conv = self.add_sublayer( + box_conv_name, + SepConvLayer( + in_channels=self.in_channels[i], + out_channels=num_prior * 4, + kernel_size=kernel_size, + padding=padding, + conv_decay=conv_decay)) + self.box_convs.append(box_conv) + + score_conv_name = "scores{}".format(i) + if not use_sepconv: + score_conv = self.add_sublayer( + score_conv_name, + nn.Conv2D( + in_channels=self.in_channels[i], + out_channels=num_prior * self.num_classes, + kernel_size=kernel_size, + padding=padding)) + else: + score_conv = self.add_sublayer( + score_conv_name, + SepConvLayer( + in_channels=self.in_channels[i], + out_channels=num_prior * self.num_classes, + kernel_size=kernel_size, + padding=padding, + conv_decay=conv_decay)) + self.score_convs.append(score_conv) + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + def forward(self, feats, image, gt_bbox=None, gt_class=None): + if self.use_extra_head: + assert len(feats) == 1, \ + ("If you set use_extra_head=True, backbone feature " + "list length should be 1.") + feats = self.ssd_extra_head(feats[0]) + box_preds = [] + cls_scores = [] + for feat, box_conv, score_conv in zip(feats, self.box_convs, + self.score_convs): + box_pred = box_conv(feat) + box_pred = paddle.transpose(box_pred, [0, 2, 3, 1]) + box_pred = paddle.reshape(box_pred, [0, -1, 4]) + box_preds.append(box_pred) + + cls_score = score_conv(feat) + cls_score = paddle.transpose(cls_score, [0, 2, 3, 1]) + cls_score = paddle.reshape(cls_score, [0, -1, self.num_classes]) + cls_scores.append(cls_score) + + prior_boxes = self.anchor_generator(feats, image) + + if self.training: + return self.get_loss(box_preds, cls_scores, gt_bbox, gt_class, + prior_boxes) + else: + return (box_preds, cls_scores), prior_boxes + + def get_loss(self, boxes, scores, gt_bbox, gt_class, prior_boxes): + return self.loss(boxes, scores, gt_bbox, gt_class, prior_boxes) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/tood_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/tood_head.py new file mode 100644 index 000000000..b9dbd17e3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/tood_head.py @@ -0,0 +1,425 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Constant + +from ppdet.core.workspace import register +from ..initializer import normal_, constant_, bias_init_with_prob +from ppdet.modeling.bbox_utils import bbox_center +from ..losses import GIoULoss +from paddle.vision.ops import deform_conv2d +from ppdet.modeling.layers import ConvNormLayer + + +class ScaleReg(nn.Layer): + """ + Parameter for scaling the regression outputs. + """ + + def __init__(self, init_scale=1.): + super(ScaleReg, self).__init__() + self.scale_reg = self.create_parameter( + shape=[1], + attr=ParamAttr(initializer=Constant(value=init_scale)), + dtype="float32") + + def forward(self, inputs): + out = inputs * self.scale_reg + return out + + +class TaskDecomposition(nn.Layer): + """This code is based on + https://github.com/fcjian/TOOD/blob/master/mmdet/models/dense_heads/tood_head.py + """ + + def __init__( + self, + feat_channels, + stacked_convs, + la_down_rate=8, + norm_type='gn', + norm_groups=32, ): + super(TaskDecomposition, self).__init__() + self.feat_channels = feat_channels + self.stacked_convs = stacked_convs + self.norm_type = norm_type + self.norm_groups = norm_groups + self.in_channels = self.feat_channels * self.stacked_convs + self.la_conv1 = nn.Conv2D(self.in_channels, + self.in_channels // la_down_rate, 1) + self.la_conv2 = nn.Conv2D(self.in_channels // la_down_rate, + self.stacked_convs, 1) + + self.reduction_conv = ConvNormLayer( + self.in_channels, + self.feat_channels, + filter_size=1, + stride=1, + norm_type=self.norm_type, + norm_groups=self.norm_groups) + + self._init_weights() + + def _init_weights(self): + normal_(self.la_conv1.weight, std=0.001) + normal_(self.la_conv2.weight, std=0.001) + + def forward(self, feat, avg_feat=None): + b, _, h, w = feat.shape + if avg_feat is None: + avg_feat = F.adaptive_avg_pool2d(feat, (1, 1)) + weight = F.relu(self.la_conv1(avg_feat)) + weight = F.sigmoid(self.la_conv2(weight)) + + # here new_conv_weight = layer_attention_weight * conv_weight + # in order to save memory and FLOPs. + conv_weight = weight.reshape([b, 1, self.stacked_convs, 1]) * \ + self.reduction_conv.conv.weight.reshape( + [1, self.feat_channels, self.stacked_convs, self.feat_channels]) + conv_weight = conv_weight.reshape( + [b, self.feat_channels, self.in_channels]) + feat = feat.reshape([b, self.in_channels, h * w]) + feat = paddle.bmm(conv_weight, feat).reshape( + [b, self.feat_channels, h, w]) + if self.norm_type is not None: + feat = self.reduction_conv.norm(feat) + feat = F.relu(feat) + return feat + + +@register +class TOODHead(nn.Layer): + """This code is based on + https://github.com/fcjian/TOOD/blob/master/mmdet/models/dense_heads/tood_head.py + """ + __inject__ = ['nms', 'static_assigner', 'assigner'] + __shared__ = ['num_classes'] + + def __init__(self, + num_classes=80, + feat_channels=256, + stacked_convs=6, + fpn_strides=(8, 16, 32, 64, 128), + grid_cell_scale=8, + grid_cell_offset=0.5, + norm_type='gn', + norm_groups=32, + static_assigner_epoch=4, + use_align_head=True, + loss_weight={ + 'class': 1.0, + 'bbox': 1.0, + 'iou': 2.0, + }, + nms='MultiClassNMS', + static_assigner='ATSSAssigner', + assigner='TaskAlignedAssigner'): + super(TOODHead, self).__init__() + self.num_classes = num_classes + self.feat_channels = feat_channels + self.stacked_convs = stacked_convs + self.fpn_strides = fpn_strides + self.grid_cell_scale = grid_cell_scale + self.grid_cell_offset = grid_cell_offset + self.static_assigner_epoch = static_assigner_epoch + self.use_align_head = use_align_head + self.nms = nms + self.static_assigner = static_assigner + self.assigner = assigner + self.loss_weight = loss_weight + self.giou_loss = GIoULoss() + + self.inter_convs = nn.LayerList() + for i in range(self.stacked_convs): + self.inter_convs.append( + ConvNormLayer( + self.feat_channels, + self.feat_channels, + filter_size=3, + stride=1, + norm_type=norm_type, + norm_groups=norm_groups)) + + self.cls_decomp = TaskDecomposition( + self.feat_channels, + self.stacked_convs, + self.stacked_convs * 8, + norm_type=norm_type, + norm_groups=norm_groups) + self.reg_decomp = TaskDecomposition( + self.feat_channels, + self.stacked_convs, + self.stacked_convs * 8, + norm_type=norm_type, + norm_groups=norm_groups) + + self.tood_cls = nn.Conv2D( + self.feat_channels, self.num_classes, 3, padding=1) + self.tood_reg = nn.Conv2D(self.feat_channels, 4, 3, padding=1) + + if self.use_align_head: + self.cls_prob_conv1 = nn.Conv2D(self.feat_channels * + self.stacked_convs, + self.feat_channels // 4, 1) + self.cls_prob_conv2 = nn.Conv2D( + self.feat_channels // 4, 1, 3, padding=1) + self.reg_offset_conv1 = nn.Conv2D(self.feat_channels * + self.stacked_convs, + self.feat_channels // 4, 1) + self.reg_offset_conv2 = nn.Conv2D( + self.feat_channels // 4, 4 * 2, 3, padding=1) + + self.scales_regs = nn.LayerList([ScaleReg() for _ in self.fpn_strides]) + + self._init_weights() + + @classmethod + def from_config(cls, cfg, input_shape): + return { + 'feat_channels': input_shape[0].channels, + 'fpn_strides': [i.stride for i in input_shape], + } + + def _init_weights(self): + bias_cls = bias_init_with_prob(0.01) + normal_(self.tood_cls.weight, std=0.01) + constant_(self.tood_cls.bias, bias_cls) + normal_(self.tood_reg.weight, std=0.01) + + if self.use_align_head: + normal_(self.cls_prob_conv1.weight, std=0.01) + normal_(self.cls_prob_conv2.weight, std=0.01) + constant_(self.cls_prob_conv2.bias, bias_cls) + normal_(self.reg_offset_conv1.weight, std=0.001) + normal_(self.reg_offset_conv2.weight, std=0.001) + constant_(self.reg_offset_conv2.bias) + + def _generate_anchors(self, feats): + anchors, num_anchors_list = [], [] + stride_tensor_list = [] + for feat, stride in zip(feats, self.fpn_strides): + _, _, h, w = feat.shape + cell_half_size = self.grid_cell_scale * stride * 0.5 + shift_x = (paddle.arange(end=w) + self.grid_cell_offset) * stride + shift_y = (paddle.arange(end=h) + self.grid_cell_offset) * stride + shift_y, shift_x = paddle.meshgrid(shift_y, shift_x) + anchor = paddle.stack( + [ + shift_x - cell_half_size, shift_y - cell_half_size, + shift_x + cell_half_size, shift_y + cell_half_size + ], + axis=-1) + anchors.append(anchor.reshape([-1, 4])) + num_anchors_list.append(len(anchors[-1])) + stride_tensor_list.append( + paddle.full([num_anchors_list[-1], 1], stride)) + return anchors, num_anchors_list, stride_tensor_list + + @staticmethod + def _batch_distance2bbox(points, distance, max_shapes=None): + """Decode distance prediction to bounding box. + Args: + points (Tensor): [B, l, 2] + distance (Tensor): [B, l, 4] + max_shapes (tuple): [B, 2], "h w" format, Shape of the image. + Returns: + Tensor: Decoded bboxes. + """ + x1 = points[:, :, 0] - distance[:, :, 0] + y1 = points[:, :, 1] - distance[:, :, 1] + x2 = points[:, :, 0] + distance[:, :, 2] + y2 = points[:, :, 1] + distance[:, :, 3] + bboxes = paddle.stack([x1, y1, x2, y2], -1) + if max_shapes is not None: + out_bboxes = [] + for bbox, max_shape in zip(bboxes, max_shapes): + bbox[:, 0] = bbox[:, 0].clip(min=0, max=max_shape[1]) + bbox[:, 1] = bbox[:, 1].clip(min=0, max=max_shape[0]) + bbox[:, 2] = bbox[:, 2].clip(min=0, max=max_shape[1]) + bbox[:, 3] = bbox[:, 3].clip(min=0, max=max_shape[0]) + out_bboxes.append(bbox) + out_bboxes = paddle.stack(out_bboxes) + return out_bboxes + return bboxes + + @staticmethod + def _deform_sampling(feat, offset): + """ Sampling the feature according to offset. + Args: + feat (Tensor): Feature + offset (Tensor): Spatial offset for for feature sampliing + """ + # it is an equivalent implementation of bilinear interpolation + # you can also use F.grid_sample instead + c = feat.shape[1] + weight = paddle.ones([c, 1, 1, 1]) + y = deform_conv2d(feat, offset, weight, deformable_groups=c, groups=c) + return y + + def forward(self, feats): + assert len(feats) == len(self.fpn_strides), \ + "The size of feats is not equal to size of fpn_strides" + + anchors, num_anchors_list, stride_tensor_list = self._generate_anchors( + feats) + cls_score_list, bbox_pred_list = [], [] + for feat, scale_reg, anchor, stride in zip(feats, self.scales_regs, + anchors, self.fpn_strides): + b, _, h, w = feat.shape + inter_feats = [] + for inter_conv in self.inter_convs: + feat = F.relu(inter_conv(feat)) + inter_feats.append(feat) + feat = paddle.concat(inter_feats, axis=1) + + # task decomposition + avg_feat = F.adaptive_avg_pool2d(feat, (1, 1)) + cls_feat = self.cls_decomp(feat, avg_feat) + reg_feat = self.reg_decomp(feat, avg_feat) + + # cls prediction and alignment + cls_logits = self.tood_cls(cls_feat) + if self.use_align_head: + cls_prob = F.relu(self.cls_prob_conv1(feat)) + cls_prob = F.sigmoid(self.cls_prob_conv2(cls_prob)) + cls_score = (F.sigmoid(cls_logits) * cls_prob).sqrt() + else: + cls_score = F.sigmoid(cls_logits) + cls_score_list.append(cls_score.flatten(2).transpose([0, 2, 1])) + + # reg prediction and alignment + reg_dist = scale_reg(self.tood_reg(reg_feat).exp()) + reg_dist = reg_dist.transpose([0, 2, 3, 1]).reshape([b, -1, 4]) + anchor_centers = bbox_center(anchor).unsqueeze(0) / stride + reg_bbox = self._batch_distance2bbox( + anchor_centers.tile([b, 1, 1]), reg_dist) + if self.use_align_head: + reg_bbox = reg_bbox.reshape([b, h, w, 4]).transpose( + [0, 3, 1, 2]) + reg_offset = F.relu(self.reg_offset_conv1(feat)) + reg_offset = self.reg_offset_conv2(reg_offset) + bbox_pred = self._deform_sampling(reg_bbox, reg_offset) + bbox_pred = bbox_pred.flatten(2).transpose([0, 2, 1]) + else: + bbox_pred = reg_bbox + + if not self.training: + bbox_pred *= stride + bbox_pred_list.append(bbox_pred) + cls_score_list = paddle.concat(cls_score_list, axis=1) + bbox_pred_list = paddle.concat(bbox_pred_list, axis=1) + anchors = paddle.concat(anchors) + anchors.stop_gradient = True + stride_tensor_list = paddle.concat(stride_tensor_list).unsqueeze(0) + stride_tensor_list.stop_gradient = True + + return cls_score_list, bbox_pred_list, anchors, num_anchors_list, stride_tensor_list + + @staticmethod + def _focal_loss(score, label, alpha=0.25, gamma=2.0): + weight = (score - label).pow(gamma) + if alpha > 0: + alpha_t = alpha * label + (1 - alpha) * (1 - label) + weight *= alpha_t + loss = F.binary_cross_entropy( + score, label, weight=weight, reduction='sum') + return loss + + def get_loss(self, head_outs, gt_meta): + pred_scores, pred_bboxes, anchors, num_anchors_list, stride_tensor_list = head_outs + gt_labels = gt_meta['gt_class'] + gt_bboxes = gt_meta['gt_bbox'] + # label assignment + if gt_meta['epoch_id'] < self.static_assigner_epoch: + assigned_labels, assigned_bboxes, assigned_scores = self.static_assigner( + anchors, + num_anchors_list, + gt_labels, + gt_bboxes, + bg_index=self.num_classes) + alpha_l = 0.25 + else: + assigned_labels, assigned_bboxes, assigned_scores = self.assigner( + pred_scores.detach(), + pred_bboxes.detach() * stride_tensor_list, + bbox_center(anchors), + gt_labels, + gt_bboxes, + bg_index=self.num_classes) + alpha_l = -1 + + # rescale bbox + assigned_bboxes /= stride_tensor_list + # classification loss + loss_cls = self._focal_loss(pred_scores, assigned_scores, alpha=alpha_l) + # select positive samples mask + mask_positive = (assigned_labels != self.num_classes) + num_pos = mask_positive.astype(paddle.float32).sum() + # bbox regression loss + if num_pos > 0: + bbox_mask = mask_positive.unsqueeze(-1).tile([1, 1, 4]) + pred_bboxes_pos = paddle.masked_select(pred_bboxes, + bbox_mask).reshape([-1, 4]) + assigned_bboxes_pos = paddle.masked_select( + assigned_bboxes, bbox_mask).reshape([-1, 4]) + bbox_weight = paddle.masked_select( + assigned_scores.sum(-1), mask_positive).unsqueeze(-1) + # iou loss + loss_iou = self.giou_loss(pred_bboxes_pos, + assigned_bboxes_pos) * bbox_weight + loss_iou = loss_iou.sum() / bbox_weight.sum() + # l1 loss + loss_l1 = F.l1_loss(pred_bboxes_pos, assigned_bboxes_pos) + else: + loss_iou = paddle.zeros([1]) + loss_l1 = paddle.zeros([1]) + + loss_cls /= assigned_scores.sum().clip(min=1) + loss = self.loss_weight['class'] * loss_cls + self.loss_weight[ + 'iou'] * loss_iou + + return { + 'loss': loss, + 'loss_class': loss_cls, + 'loss_iou': loss_iou, + 'loss_l1': loss_l1 + } + + def post_process(self, head_outs, img_shape, scale_factor): + pred_scores, pred_bboxes, _, _, _ = head_outs + pred_scores = pred_scores.transpose([0, 2, 1]) + + for i in range(len(pred_bboxes)): + pred_bboxes[i, :, 0] = pred_bboxes[i, :, 0].clip( + min=0, max=img_shape[i, 1]) + pred_bboxes[i, :, 1] = pred_bboxes[i, :, 1].clip( + min=0, max=img_shape[i, 0]) + pred_bboxes[i, :, 2] = pred_bboxes[i, :, 2].clip( + min=0, max=img_shape[i, 1]) + pred_bboxes[i, :, 3] = pred_bboxes[i, :, 3].clip( + min=0, max=img_shape[i, 0]) + # scale bbox to origin + scale_factor = scale_factor.flip([1]).tile([1, 2]).unsqueeze(1) + pred_bboxes /= scale_factor + bbox_pred, bbox_num, _ = self.nms(pred_bboxes, pred_scores) + return bbox_pred, bbox_num diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ttf_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ttf_head.py new file mode 100644 index 000000000..dfe97bdb7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/ttf_head.py @@ -0,0 +1,311 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Constant, Normal +from paddle.regularizer import L2Decay +from ppdet.core.workspace import register +from ppdet.modeling.layers import DeformableConvV2, LiteConv +import numpy as np + + +@register +class HMHead(nn.Layer): + """ + Args: + ch_in (int): The channel number of input Tensor. + ch_out (int): The channel number of output Tensor. + num_classes (int): Number of classes. + conv_num (int): The convolution number of hm_feat. + dcn_head(bool): whether use dcn in head. False by default. + lite_head(bool): whether use lite version. False by default. + norm_type (string): norm type, 'sync_bn', 'bn', 'gn' are optional. + bn by default + + Return: + Heatmap head output + """ + __shared__ = ['num_classes', 'norm_type'] + + def __init__( + self, + ch_in, + ch_out=128, + num_classes=80, + conv_num=2, + dcn_head=False, + lite_head=False, + norm_type='bn', ): + super(HMHead, self).__init__() + head_conv = nn.Sequential() + for i in range(conv_num): + name = 'conv.{}'.format(i) + if lite_head: + lite_name = 'hm.' + name + head_conv.add_sublayer( + lite_name, + LiteConv( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + norm_type=norm_type)) + else: + if dcn_head: + head_conv.add_sublayer( + name, + DeformableConvV2( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + kernel_size=3, + weight_attr=ParamAttr(initializer=Normal(0, 0.01)))) + else: + head_conv.add_sublayer( + name, + nn.Conv2D( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0, 0.01)), + bias_attr=ParamAttr( + learning_rate=2., regularizer=L2Decay(0.)))) + head_conv.add_sublayer(name + '.act', nn.ReLU()) + self.feat = head_conv + bias_init = float(-np.log((1 - 0.01) / 0.01)) + weight_attr = None if lite_head else ParamAttr(initializer=Normal(0, + 0.01)) + self.head = nn.Conv2D( + in_channels=ch_out, + out_channels=num_classes, + kernel_size=1, + weight_attr=weight_attr, + bias_attr=ParamAttr( + learning_rate=2., + regularizer=L2Decay(0.), + initializer=Constant(bias_init))) + + def forward(self, feat): + out = self.feat(feat) + out = self.head(out) + return out + + +@register +class WHHead(nn.Layer): + """ + Args: + ch_in (int): The channel number of input Tensor. + ch_out (int): The channel number of output Tensor. + conv_num (int): The convolution number of wh_feat. + dcn_head(bool): whether use dcn in head. False by default. + lite_head(bool): whether use lite version. False by default. + norm_type (string): norm type, 'sync_bn', 'bn', 'gn' are optional. + bn by default + Return: + Width & Height head output + """ + __shared__ = ['norm_type'] + + def __init__(self, + ch_in, + ch_out=64, + conv_num=2, + dcn_head=False, + lite_head=False, + norm_type='bn'): + super(WHHead, self).__init__() + head_conv = nn.Sequential() + for i in range(conv_num): + name = 'conv.{}'.format(i) + if lite_head: + lite_name = 'wh.' + name + head_conv.add_sublayer( + lite_name, + LiteConv( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + norm_type=norm_type)) + else: + if dcn_head: + head_conv.add_sublayer( + name, + DeformableConvV2( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + kernel_size=3, + weight_attr=ParamAttr(initializer=Normal(0, 0.01)))) + else: + head_conv.add_sublayer( + name, + nn.Conv2D( + in_channels=ch_in if i == 0 else ch_out, + out_channels=ch_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0, 0.01)), + bias_attr=ParamAttr( + learning_rate=2., regularizer=L2Decay(0.)))) + head_conv.add_sublayer(name + '.act', nn.ReLU()) + + weight_attr = None if lite_head else ParamAttr(initializer=Normal(0, + 0.01)) + self.feat = head_conv + self.head = nn.Conv2D( + in_channels=ch_out, + out_channels=4, + kernel_size=1, + weight_attr=weight_attr, + bias_attr=ParamAttr( + learning_rate=2., regularizer=L2Decay(0.))) + + def forward(self, feat): + out = self.feat(feat) + out = self.head(out) + out = F.relu(out) + return out + + +@register +class TTFHead(nn.Layer): + """ + TTFHead + Args: + in_channels (int): the channel number of input to TTFHead. + num_classes (int): the number of classes, 80 by default. + hm_head_planes (int): the channel number in heatmap head, + 128 by default. + wh_head_planes (int): the channel number in width & height head, + 64 by default. + hm_head_conv_num (int): the number of convolution in heatmap head, + 2 by default. + wh_head_conv_num (int): the number of convolution in width & height + head, 2 by default. + hm_loss (object): Instance of 'CTFocalLoss'. + wh_loss (object): Instance of 'GIoULoss'. + wh_offset_base (float): the base offset of width and height, + 16.0 by default. + down_ratio (int): the actual down_ratio is calculated by base_down_ratio + (default 16) and the number of upsample layers. + lite_head(bool): whether use lite version. False by default. + norm_type (string): norm type, 'sync_bn', 'bn', 'gn' are optional. + bn by default + ags_module(bool): whether use AGS module to reweight location feature. + false by default. + + """ + + __shared__ = ['num_classes', 'down_ratio', 'norm_type'] + __inject__ = ['hm_loss', 'wh_loss'] + + def __init__(self, + in_channels, + num_classes=80, + hm_head_planes=128, + wh_head_planes=64, + hm_head_conv_num=2, + wh_head_conv_num=2, + hm_loss='CTFocalLoss', + wh_loss='GIoULoss', + wh_offset_base=16., + down_ratio=4, + dcn_head=False, + lite_head=False, + norm_type='bn', + ags_module=False): + super(TTFHead, self).__init__() + self.in_channels = in_channels + self.hm_head = HMHead(in_channels, hm_head_planes, num_classes, + hm_head_conv_num, dcn_head, lite_head, norm_type) + self.wh_head = WHHead(in_channels, wh_head_planes, wh_head_conv_num, + dcn_head, lite_head, norm_type) + self.hm_loss = hm_loss + self.wh_loss = wh_loss + + self.wh_offset_base = wh_offset_base + self.down_ratio = down_ratio + self.ags_module = ags_module + + @classmethod + def from_config(cls, cfg, input_shape): + if isinstance(input_shape, (list, tuple)): + input_shape = input_shape[0] + return {'in_channels': input_shape.channels, } + + def forward(self, feats): + hm = self.hm_head(feats) + wh = self.wh_head(feats) * self.wh_offset_base + return hm, wh + + def filter_box_by_weight(self, pred, target, weight): + """ + Filter out boxes where ttf_reg_weight is 0, only keep positive samples. + """ + index = paddle.nonzero(weight > 0) + index.stop_gradient = True + weight = paddle.gather_nd(weight, index) + pred = paddle.gather_nd(pred, index) + target = paddle.gather_nd(target, index) + return pred, target, weight + + def filter_loc_by_weight(self, score, weight): + index = paddle.nonzero(weight > 0) + index.stop_gradient = True + score = paddle.gather_nd(score, index) + return score + + def get_loss(self, pred_hm, pred_wh, target_hm, box_target, target_weight): + pred_hm = paddle.clip(F.sigmoid(pred_hm), 1e-4, 1 - 1e-4) + hm_loss = self.hm_loss(pred_hm, target_hm) + H, W = target_hm.shape[2:] + mask = paddle.reshape(target_weight, [-1, H, W]) + avg_factor = paddle.sum(mask) + 1e-4 + + base_step = self.down_ratio + shifts_x = paddle.arange(0, W * base_step, base_step, dtype='int32') + shifts_y = paddle.arange(0, H * base_step, base_step, dtype='int32') + shift_y, shift_x = paddle.tensor.meshgrid([shifts_y, shifts_x]) + base_loc = paddle.stack([shift_x, shift_y], axis=0) + base_loc.stop_gradient = True + + pred_boxes = paddle.concat( + [0 - pred_wh[:, 0:2, :, :] + base_loc, pred_wh[:, 2:4] + base_loc], + axis=1) + pred_boxes = paddle.transpose(pred_boxes, [0, 2, 3, 1]) + boxes = paddle.transpose(box_target, [0, 2, 3, 1]) + boxes.stop_gradient = True + + if self.ags_module: + pred_hm_max = paddle.max(pred_hm, axis=1, keepdim=True) + pred_hm_max_softmax = F.softmax(pred_hm_max, axis=1) + pred_hm_max_softmax = paddle.transpose(pred_hm_max_softmax, + [0, 2, 3, 1]) + pred_hm_max_softmax = self.filter_loc_by_weight(pred_hm_max_softmax, + mask) + else: + pred_hm_max_softmax = None + + pred_boxes, boxes, mask = self.filter_box_by_weight(pred_boxes, boxes, + mask) + mask.stop_gradient = True + wh_loss = self.wh_loss( + pred_boxes, + boxes, + iou_weight=mask.unsqueeze(1), + loc_reweight=pred_hm_max_softmax) + wh_loss = wh_loss / avg_factor + + ttf_loss = {'hm_loss': hm_loss, 'wh_loss': wh_loss} + return ttf_loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/yolo_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/yolo_head.py new file mode 100644 index 000000000..7b4e9bc33 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/heads/yolo_head.py @@ -0,0 +1,124 @@ +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from ppdet.core.workspace import register + + +def _de_sigmoid(x, eps=1e-7): + x = paddle.clip(x, eps, 1. / eps) + x = paddle.clip(1. / x - 1., eps, 1. / eps) + x = -paddle.log(x) + return x + + +@register +class YOLOv3Head(nn.Layer): + __shared__ = ['num_classes', 'data_format'] + __inject__ = ['loss'] + + def __init__(self, + in_channels=[1024, 512, 256], + anchors=[[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], + [59, 119], [116, 90], [156, 198], [373, 326]], + anchor_masks=[[6, 7, 8], [3, 4, 5], [0, 1, 2]], + num_classes=80, + loss='YOLOv3Loss', + iou_aware=False, + iou_aware_factor=0.4, + data_format='NCHW'): + """ + Head for YOLOv3 network + + Args: + num_classes (int): number of foreground classes + anchors (list): anchors + anchor_masks (list): anchor masks + loss (object): YOLOv3Loss instance + iou_aware (bool): whether to use iou_aware + iou_aware_factor (float): iou aware factor + data_format (str): data format, NCHW or NHWC + """ + super(YOLOv3Head, self).__init__() + assert len(in_channels) > 0, "in_channels length should > 0" + self.in_channels = in_channels + self.num_classes = num_classes + self.loss = loss + + self.iou_aware = iou_aware + self.iou_aware_factor = iou_aware_factor + + self.parse_anchor(anchors, anchor_masks) + self.num_outputs = len(self.anchors) + self.data_format = data_format + + self.yolo_outputs = [] + for i in range(len(self.anchors)): + + if self.iou_aware: + num_filters = len(self.anchors[i]) * (self.num_classes + 6) + else: + num_filters = len(self.anchors[i]) * (self.num_classes + 5) + name = 'yolo_output.{}'.format(i) + conv = nn.Conv2D( + in_channels=self.in_channels[i], + out_channels=num_filters, + kernel_size=1, + stride=1, + padding=0, + data_format=data_format, + bias_attr=ParamAttr(regularizer=L2Decay(0.))) + conv.skip_quant = True + yolo_output = self.add_sublayer(name, conv) + self.yolo_outputs.append(yolo_output) + + def parse_anchor(self, anchors, anchor_masks): + self.anchors = [[anchors[i] for i in mask] for mask in anchor_masks] + self.mask_anchors = [] + anchor_num = len(anchors) + for masks in anchor_masks: + self.mask_anchors.append([]) + for mask in masks: + assert mask < anchor_num, "anchor mask index overflow" + self.mask_anchors[-1].extend(anchors[mask]) + + def forward(self, feats, targets=None): + assert len(feats) == len(self.anchors) + yolo_outputs = [] + for i, feat in enumerate(feats): + yolo_output = self.yolo_outputs[i](feat) + if self.data_format == 'NHWC': + yolo_output = paddle.transpose(yolo_output, [0, 3, 1, 2]) + yolo_outputs.append(yolo_output) + + if self.training: + return self.loss(yolo_outputs, targets, self.anchors) + else: + if self.iou_aware: + y = [] + for i, out in enumerate(yolo_outputs): + na = len(self.anchors[i]) + ioup, x = out[:, 0:na, :, :], out[:, na:, :, :] + b, c, h, w = x.shape + no = c // na + x = x.reshape((b, na, no, h * w)) + ioup = ioup.reshape((b, na, 1, h * w)) + obj = x[:, :, 4:5, :] + ioup = F.sigmoid(ioup) + obj = F.sigmoid(obj) + obj_t = (obj**(1 - self.iou_aware_factor)) * ( + ioup**self.iou_aware_factor) + obj_t = _de_sigmoid(obj_t) + loc_t = x[:, :, :4, :] + cls_t = x[:, :, 5:, :] + y_t = paddle.concat([loc_t, obj_t, cls_t], axis=2) + y_t = y_t.reshape((b, c, h, w)) + y.append(y_t) + return y + else: + return yolo_outputs + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/initializer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/initializer.py new file mode 100644 index 000000000..b7a135dcc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/initializer.py @@ -0,0 +1,317 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/pytorch/pytorch/blob/master/torch/nn/init.py +Ths copyright of pytorch/pytorch is a BSD-style license, as found in the LICENSE file. +""" + +import math +import numpy as np + +import paddle +import paddle.nn as nn + +__all__ = [ + 'uniform_', + 'normal_', + 'constant_', + 'ones_', + 'zeros_', + 'xavier_uniform_', + 'xavier_normal_', + 'kaiming_uniform_', + 'kaiming_normal_', + 'linear_init_', + 'conv_init_', + 'reset_initialized_parameter', +] + + +def _no_grad_uniform_(tensor, a, b): + with paddle.no_grad(): + tensor.set_value( + paddle.uniform( + shape=tensor.shape, dtype=tensor.dtype, min=a, max=b)) + return tensor + + +def _no_grad_normal_(tensor, mean=0., std=1.): + with paddle.no_grad(): + tensor.set_value(paddle.normal(mean=mean, std=std, shape=tensor.shape)) + return tensor + + +def _no_grad_fill_(tensor, value=0.): + with paddle.no_grad(): + tensor.set_value(paddle.full_like(tensor, value, dtype=tensor.dtype)) + return tensor + + +def uniform_(tensor, a, b): + """ + Modified tensor inspace using uniform_ + Args: + tensor (paddle.Tensor): paddle Tensor + a (float|int): min value. + b (float|int): max value. + Return: + tensor + """ + return _no_grad_uniform_(tensor, a, b) + + +def normal_(tensor, mean=0., std=1.): + """ + Modified tensor inspace using normal_ + Args: + tensor (paddle.Tensor): paddle Tensor + mean (float|int): mean value. + std (float|int): std value. + Return: + tensor + """ + return _no_grad_normal_(tensor, mean, std) + + +def constant_(tensor, value=0.): + """ + Modified tensor inspace using constant_ + Args: + tensor (paddle.Tensor): paddle Tensor + value (float|int): value to fill tensor. + Return: + tensor + """ + return _no_grad_fill_(tensor, value) + + +def ones_(tensor): + """ + Modified tensor inspace using ones_ + Args: + tensor (paddle.Tensor): paddle Tensor + Return: + tensor + """ + return _no_grad_fill_(tensor, 1) + + +def zeros_(tensor): + """ + Modified tensor inspace using zeros_ + Args: + tensor (paddle.Tensor): paddle Tensor + Return: + tensor + """ + return _no_grad_fill_(tensor, 0) + + +def _calculate_fan_in_and_fan_out(tensor, reverse=False): + """ + Calculate (fan_in, _fan_out) for tensor + + Args: + tensor (Tensor): paddle.Tensor + reverse (bool: False): tensor data format order, False by default as [fout, fin, ...]. e.g. : conv.weight [cout, cin, kh, kw] is False; linear.weight [cin, cout] is True + + Return: + Tuple[fan_in, fan_out] + """ + if tensor.ndim < 2: + raise ValueError( + "Fan in and fan out can not be computed for tensor with fewer than 2 dimensions" + ) + + if reverse: + num_input_fmaps, num_output_fmaps = tensor.shape[0], tensor.shape[1] + else: + num_input_fmaps, num_output_fmaps = tensor.shape[1], tensor.shape[0] + + receptive_field_size = 1 + if tensor.ndim > 2: + receptive_field_size = np.prod(tensor.shape[2:]) + + fan_in = num_input_fmaps * receptive_field_size + fan_out = num_output_fmaps * receptive_field_size + + return fan_in, fan_out + + +def xavier_uniform_(tensor, gain=1., reverse=False): + """ + Modified tensor inspace using xavier_uniform_ + Args: + tensor (paddle.Tensor): paddle Tensor + gain (float): super parameter, 1. default. + reverse (bool): reverse (bool: False): tensor data format order, False by default as [fout, fin, ...]. + Return: + tensor + """ + fan_in, fan_out = _calculate_fan_in_and_fan_out(tensor, reverse=reverse) + std = gain * math.sqrt(2.0 / float(fan_in + fan_out)) + k = math.sqrt(3.0) * std + return _no_grad_uniform_(tensor, -k, k) + + +def xavier_normal_(tensor, gain=1., reverse=False): + """ + Modified tensor inspace using xavier_normal_ + Args: + tensor (paddle.Tensor): paddle Tensor + gain (float): super parameter, 1. default. + reverse (bool): reverse (bool: False): tensor data format order, False by default as [fout, fin, ...]. + Return: + tensor + """ + fan_in, fan_out = _calculate_fan_in_and_fan_out(tensor, reverse=reverse) + std = gain * math.sqrt(2.0 / float(fan_in + fan_out)) + return _no_grad_normal_(tensor, 0, std) + + +# reference: https://pytorch.org/docs/stable/_modules/torch/nn/init.html +def _calculate_correct_fan(tensor, mode, reverse=False): + mode = mode.lower() + valid_modes = ['fan_in', 'fan_out'] + if mode not in valid_modes: + raise ValueError("Mode {} not supported, please use one of {}".format( + mode, valid_modes)) + + fan_in, fan_out = _calculate_fan_in_and_fan_out(tensor, reverse) + + return fan_in if mode == 'fan_in' else fan_out + + +def _calculate_gain(nonlinearity, param=None): + linear_fns = [ + 'linear', 'conv1d', 'conv2d', 'conv3d', 'conv_transpose1d', + 'conv_transpose2d', 'conv_transpose3d' + ] + if nonlinearity in linear_fns or nonlinearity == 'sigmoid': + return 1 + elif nonlinearity == 'tanh': + return 5.0 / 3 + elif nonlinearity == 'relu': + return math.sqrt(2.0) + elif nonlinearity == 'leaky_relu': + if param is None: + negative_slope = 0.01 + elif not isinstance(param, bool) and isinstance( + param, int) or isinstance(param, float): + # True/False are instances of int, hence check above + negative_slope = param + else: + raise ValueError("negative_slope {} not a valid number".format( + param)) + return math.sqrt(2.0 / (1 + negative_slope**2)) + elif nonlinearity == 'selu': + return 3.0 / 4 + else: + raise ValueError("Unsupported nonlinearity {}".format(nonlinearity)) + + +def kaiming_uniform_(tensor, + a=0, + mode='fan_in', + nonlinearity='leaky_relu', + reverse=False): + """ + Modified tensor inspace using kaiming_uniform method + Args: + tensor (paddle.Tensor): paddle Tensor + mode (str): ['fan_in', 'fan_out'], 'fin_in' defalut + nonlinearity (str): nonlinearity method name + reverse (bool): reverse (bool: False): tensor data format order, False by default as [fout, fin, ...]. + Return: + tensor + """ + fan = _calculate_correct_fan(tensor, mode, reverse) + gain = _calculate_gain(nonlinearity, a) + std = gain / math.sqrt(fan) + k = math.sqrt(3.0) * std + return _no_grad_uniform_(tensor, -k, k) + + +def kaiming_normal_(tensor, + a=0, + mode='fan_in', + nonlinearity='leaky_relu', + reverse=False): + """ + Modified tensor inspace using kaiming_normal_ + Args: + tensor (paddle.Tensor): paddle Tensor + mode (str): ['fan_in', 'fan_out'], 'fin_in' defalut + nonlinearity (str): nonlinearity method name + reverse (bool): reverse (bool: False): tensor data format order, False by default as [fout, fin, ...]. + Return: + tensor + """ + fan = _calculate_correct_fan(tensor, mode, reverse) + gain = _calculate_gain(nonlinearity, a) + std = gain / math.sqrt(fan) + return _no_grad_normal_(tensor, 0, std) + + +def linear_init_(module): + bound = 1 / math.sqrt(module.weight.shape[0]) + uniform_(module.weight, -bound, bound) + uniform_(module.bias, -bound, bound) + + +def conv_init_(module): + bound = 1 / np.sqrt(np.prod(module.weight.shape[1:])) + uniform_(module.weight, -bound, bound) + uniform_(module.bias, -bound, bound) + + +def bias_init_with_prob(prior_prob=0.01): + """initialize conv/fc bias value according to a given probability value.""" + bias_init = float(-np.log((1 - prior_prob) / prior_prob)) + return bias_init + + +@paddle.no_grad() +def reset_initialized_parameter(model, include_self=True): + """ + Reset initialized parameter using following method for [conv, linear, embedding, bn] + + Args: + model (paddle.Layer): paddle Layer + include_self (bool: False): include_self for Layer.named_sublayers method. Indicate whether including itself + Return: + None + """ + for _, m in model.named_sublayers(include_self=include_self): + if isinstance(m, nn.Conv2D): + k = float(m._groups) / (m._in_channels * m._kernel_size[0] * + m._kernel_size[1]) + k = math.sqrt(k) + _no_grad_uniform_(m.weight, -k, k) + if hasattr(m, 'bias') and getattr(m, 'bias') is not None: + _no_grad_uniform_(m.bias, -k, k) + + elif isinstance(m, nn.Linear): + k = math.sqrt(1. / m.weight.shape[0]) + _no_grad_uniform_(m.weight, -k, k) + if hasattr(m, 'bias') and getattr(m, 'bias') is not None: + _no_grad_uniform_(m.bias, -k, k) + + elif isinstance(m, nn.Embedding): + _no_grad_normal_(m.weight, mean=0., std=1.) + + elif isinstance(m, (nn.BatchNorm2D, nn.LayerNorm)): + _no_grad_fill_(m.weight, 1.) + if hasattr(m, 'bias') and getattr(m, 'bias') is not None: + _no_grad_fill_(m.bias, 0) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/keypoint_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/keypoint_utils.py new file mode 100644 index 000000000..b3f84da7d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/keypoint_utils.py @@ -0,0 +1,336 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import cv2 +import numpy as np + + +def get_affine_mat_kernel(h, w, s, inv=False): + if w < h: + w_ = s + h_ = int(np.ceil((s / w * h) / 64.) * 64) + scale_w = w + scale_h = h_ / w_ * w + + else: + h_ = s + w_ = int(np.ceil((s / h * w) / 64.) * 64) + scale_h = h + scale_w = w_ / h_ * h + + center = np.array([np.round(w / 2.), np.round(h / 2.)]) + + size_resized = (w_, h_) + trans = get_affine_transform( + center, np.array([scale_w, scale_h]), 0, size_resized, inv=inv) + + return trans, size_resized + + +def get_affine_transform(center, + input_size, + rot, + output_size, + shift=(0., 0.), + inv=False): + """Get the affine transform matrix, given the center/scale/rot/output_size. + + Args: + center (np.ndarray[2, ]): Center of the bounding box (x, y). + input_size (np.ndarray[2, ]): Size of input feature (width, height). + rot (float): Rotation angle (degree). + output_size (np.ndarray[2, ]): Size of the destination heatmaps. + shift (0-100%): Shift translation ratio wrt the width/height. + Default (0., 0.). + inv (bool): Option to inverse the affine transform direction. + (inv=False: src->dst or inv=True: dst->src) + + Returns: + np.ndarray: The transform matrix. + """ + assert len(center) == 2 + assert len(output_size) == 2 + assert len(shift) == 2 + + if not isinstance(input_size, (np.ndarray, list)): + input_size = np.array([input_size, input_size], dtype=np.float32) + scale_tmp = input_size + + shift = np.array(shift) + src_w = scale_tmp[0] + dst_w = output_size[0] + dst_h = output_size[1] + + rot_rad = np.pi * rot / 180 + src_dir = rotate_point([0., src_w * -0.5], rot_rad) + dst_dir = np.array([0., dst_w * -0.5]) + + src = np.zeros((3, 2), dtype=np.float32) + + src[0, :] = center + scale_tmp * shift + src[1, :] = center + src_dir + scale_tmp * shift + src[2, :] = _get_3rd_point(src[0, :], src[1, :]) + + dst = np.zeros((3, 2), dtype=np.float32) + dst[0, :] = [dst_w * 0.5, dst_h * 0.5] + dst[1, :] = np.array([dst_w * 0.5, dst_h * 0.5]) + dst_dir + dst[2, :] = _get_3rd_point(dst[0, :], dst[1, :]) + + if inv: + trans = cv2.getAffineTransform(np.float32(dst), np.float32(src)) + else: + trans = cv2.getAffineTransform(np.float32(src), np.float32(dst)) + + return trans + + +def get_warp_matrix(theta, size_input, size_dst, size_target): + """Calculate the transformation matrix under the constraint of unbiased. + Paper ref: Huang et al. The Devil is in the Details: Delving into Unbiased + Data Processing for Human Pose Estimation (CVPR 2020). + + Args: + theta (float): Rotation angle in degrees. + size_input (np.ndarray): Size of input image [w, h]. + size_dst (np.ndarray): Size of output image [w, h]. + size_target (np.ndarray): Size of ROI in input plane [w, h]. + + Returns: + matrix (np.ndarray): A matrix for transformation. + """ + theta = np.deg2rad(theta) + matrix = np.zeros((2, 3), dtype=np.float32) + scale_x = size_dst[0] / size_target[0] + scale_y = size_dst[1] / size_target[1] + matrix[0, 0] = np.cos(theta) * scale_x + matrix[0, 1] = -np.sin(theta) * scale_x + matrix[0, 2] = scale_x * ( + -0.5 * size_input[0] * np.cos(theta) + 0.5 * size_input[1] * + np.sin(theta) + 0.5 * size_target[0]) + matrix[1, 0] = np.sin(theta) * scale_y + matrix[1, 1] = np.cos(theta) * scale_y + matrix[1, 2] = scale_y * ( + -0.5 * size_input[0] * np.sin(theta) - 0.5 * size_input[1] * + np.cos(theta) + 0.5 * size_target[1]) + return matrix + + +def _get_3rd_point(a, b): + """To calculate the affine matrix, three pairs of points are required. This + function is used to get the 3rd point, given 2D points a & b. + + The 3rd point is defined by rotating vector `a - b` by 90 degrees + anticlockwise, using b as the rotation center. + + Args: + a (np.ndarray): point(x,y) + b (np.ndarray): point(x,y) + + Returns: + np.ndarray: The 3rd point. + """ + assert len( + a) == 2, 'input of _get_3rd_point should be point with length of 2' + assert len( + b) == 2, 'input of _get_3rd_point should be point with length of 2' + direction = a - b + third_pt = b + np.array([-direction[1], direction[0]], dtype=np.float32) + + return third_pt + + +def rotate_point(pt, angle_rad): + """Rotate a point by an angle. + + Args: + pt (list[float]): 2 dimensional point to be rotated + angle_rad (float): rotation angle by radian + + Returns: + list[float]: Rotated point. + """ + assert len(pt) == 2 + sn, cs = np.sin(angle_rad), np.cos(angle_rad) + new_x = pt[0] * cs - pt[1] * sn + new_y = pt[0] * sn + pt[1] * cs + rotated_pt = [new_x, new_y] + + return rotated_pt + + +def transpred(kpts, h, w, s): + trans, _ = get_affine_mat_kernel(h, w, s, inv=True) + + return warp_affine_joints(kpts[..., :2].copy(), trans) + + +def warp_affine_joints(joints, mat): + """Apply affine transformation defined by the transform matrix on the + joints. + + Args: + joints (np.ndarray[..., 2]): Origin coordinate of joints. + mat (np.ndarray[3, 2]): The affine matrix. + + Returns: + matrix (np.ndarray[..., 2]): Result coordinate of joints. + """ + joints = np.array(joints) + shape = joints.shape + joints = joints.reshape(-1, 2) + return np.dot(np.concatenate( + (joints, joints[:, 0:1] * 0 + 1), axis=1), + mat.T).reshape(shape) + + +def affine_transform(pt, t): + new_pt = np.array([pt[0], pt[1], 1.]).T + new_pt = np.dot(t, new_pt) + return new_pt[:2] + + +def transform_preds(coords, center, scale, output_size): + target_coords = np.zeros(coords.shape) + trans = get_affine_transform(center, scale * 200, 0, output_size, inv=1) + for p in range(coords.shape[0]): + target_coords[p, 0:2] = affine_transform(coords[p, 0:2], trans) + return target_coords + + +def oks_iou(g, d, a_g, a_d, sigmas=None, in_vis_thre=None): + if not isinstance(sigmas, np.ndarray): + sigmas = np.array([ + .26, .25, .25, .35, .35, .79, .79, .72, .72, .62, .62, 1.07, 1.07, + .87, .87, .89, .89 + ]) / 10.0 + vars = (sigmas * 2)**2 + xg = g[0::3] + yg = g[1::3] + vg = g[2::3] + ious = np.zeros((d.shape[0])) + for n_d in range(0, d.shape[0]): + xd = d[n_d, 0::3] + yd = d[n_d, 1::3] + vd = d[n_d, 2::3] + dx = xd - xg + dy = yd - yg + e = (dx**2 + dy**2) / vars / ((a_g + a_d[n_d]) / 2 + np.spacing(1)) / 2 + if in_vis_thre is not None: + ind = list(vg > in_vis_thre) and list(vd > in_vis_thre) + e = e[ind] + ious[n_d] = np.sum(np.exp(-e)) / e.shape[0] if e.shape[0] != 0 else 0.0 + return ious + + +def oks_nms(kpts_db, thresh, sigmas=None, in_vis_thre=None): + """greedily select boxes with high confidence and overlap with current maximum <= thresh + rule out overlap >= thresh + + Args: + kpts_db (list): The predicted keypoints within the image + thresh (float): The threshold to select the boxes + sigmas (np.array): The variance to calculate the oks iou + Default: None + in_vis_thre (float): The threshold to select the high confidence boxes + Default: None + + Return: + keep (list): indexes to keep + """ + + if len(kpts_db) == 0: + return [] + + scores = np.array([kpts_db[i]['score'] for i in range(len(kpts_db))]) + kpts = np.array( + [kpts_db[i]['keypoints'].flatten() for i in range(len(kpts_db))]) + areas = np.array([kpts_db[i]['area'] for i in range(len(kpts_db))]) + + order = scores.argsort()[::-1] + + keep = [] + while order.size > 0: + i = order[0] + keep.append(i) + + oks_ovr = oks_iou(kpts[i], kpts[order[1:]], areas[i], areas[order[1:]], + sigmas, in_vis_thre) + + inds = np.where(oks_ovr <= thresh)[0] + order = order[inds + 1] + + return keep + + +def rescore(overlap, scores, thresh, type='gaussian'): + assert overlap.shape[0] == scores.shape[0] + if type == 'linear': + inds = np.where(overlap >= thresh)[0] + scores[inds] = scores[inds] * (1 - overlap[inds]) + else: + scores = scores * np.exp(-overlap**2 / thresh) + + return scores + + +def soft_oks_nms(kpts_db, thresh, sigmas=None, in_vis_thre=None): + """greedily select boxes with high confidence and overlap with current maximum <= thresh + rule out overlap >= thresh + + Args: + kpts_db (list): The predicted keypoints within the image + thresh (float): The threshold to select the boxes + sigmas (np.array): The variance to calculate the oks iou + Default: None + in_vis_thre (float): The threshold to select the high confidence boxes + Default: None + + Return: + keep (list): indexes to keep + """ + + if len(kpts_db) == 0: + return [] + + scores = np.array([kpts_db[i]['score'] for i in range(len(kpts_db))]) + kpts = np.array( + [kpts_db[i]['keypoints'].flatten() for i in range(len(kpts_db))]) + areas = np.array([kpts_db[i]['area'] for i in range(len(kpts_db))]) + + order = scores.argsort()[::-1] + scores = scores[order] + + # max_dets = order.size + max_dets = 20 + keep = np.zeros(max_dets, dtype=np.intp) + keep_cnt = 0 + while order.size > 0 and keep_cnt < max_dets: + i = order[0] + + oks_ovr = oks_iou(kpts[i], kpts[order[1:]], areas[i], areas[order[1:]], + sigmas, in_vis_thre) + + order = order[1:] + scores = rescore(oks_ovr, scores[1:], thresh) + + tmp = scores.argsort()[::-1] + order = order[tmp] + scores = scores[tmp] + + keep[keep_cnt] = i + keep_cnt += 1 + + keep = keep[:keep_cnt] + + return keep diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/layers.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/layers.py new file mode 100644 index 000000000..73da16a14 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/layers.py @@ -0,0 +1,1424 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import math +import six +import numpy as np +from numbers import Integral + +import paddle +import paddle.nn as nn +from paddle import ParamAttr +from paddle import to_tensor +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, Constant, XavierUniform +from paddle.regularizer import L2Decay + +from ppdet.core.workspace import register, serializable +from ppdet.modeling.bbox_utils import delta2bbox +from . import ops +from .initializer import xavier_uniform_, constant_ + +from paddle.vision.ops import DeformConv2D + + +def _to_list(l): + if isinstance(l, (list, tuple)): + return list(l) + return [l] + + +class DeformableConvV2(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size, + stride=1, + padding=0, + dilation=1, + groups=1, + weight_attr=None, + bias_attr=None, + lr_scale=1, + regularizer=None, + skip_quant=False, + dcn_bias_regularizer=L2Decay(0.), + dcn_bias_lr_scale=2.): + super(DeformableConvV2, self).__init__() + self.offset_channel = 2 * kernel_size**2 + self.mask_channel = kernel_size**2 + + if lr_scale == 1 and regularizer is None: + offset_bias_attr = ParamAttr(initializer=Constant(0.)) + else: + offset_bias_attr = ParamAttr( + initializer=Constant(0.), + learning_rate=lr_scale, + regularizer=regularizer) + self.conv_offset = nn.Conv2D( + in_channels, + 3 * kernel_size**2, + kernel_size, + stride=stride, + padding=(kernel_size - 1) // 2, + weight_attr=ParamAttr(initializer=Constant(0.0)), + bias_attr=offset_bias_attr) + if skip_quant: + self.conv_offset.skip_quant = True + + if bias_attr: + # in FCOS-DCN head, specifically need learning_rate and regularizer + dcn_bias_attr = ParamAttr( + initializer=Constant(value=0), + regularizer=dcn_bias_regularizer, + learning_rate=dcn_bias_lr_scale) + else: + # in ResNet backbone, do not need bias + dcn_bias_attr = False + self.conv_dcn = DeformConv2D( + in_channels, + out_channels, + kernel_size, + stride=stride, + padding=(kernel_size - 1) // 2 * dilation, + dilation=dilation, + groups=groups, + weight_attr=weight_attr, + bias_attr=dcn_bias_attr) + + def forward(self, x): + offset_mask = self.conv_offset(x) + offset, mask = paddle.split( + offset_mask, + num_or_sections=[self.offset_channel, self.mask_channel], + axis=1) + mask = F.sigmoid(mask) + y = self.conv_dcn(x, offset, mask=mask) + return y + + +class ConvNormLayer(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + stride, + groups=1, + norm_type='bn', + norm_decay=0., + norm_groups=32, + use_dcn=False, + bias_on=False, + lr_scale=1., + freeze_norm=False, + initializer=Normal( + mean=0., std=0.01), + skip_quant=False, + dcn_lr_scale=2., + dcn_regularizer=L2Decay(0.)): + super(ConvNormLayer, self).__init__() + assert norm_type in ['bn', 'sync_bn', 'gn'] + + if bias_on: + bias_attr = ParamAttr( + initializer=Constant(value=0.), learning_rate=lr_scale) + else: + bias_attr = False + + if not use_dcn: + self.conv = nn.Conv2D( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=groups, + weight_attr=ParamAttr( + initializer=initializer, learning_rate=1.), + bias_attr=bias_attr) + if skip_quant: + self.conv.skip_quant = True + else: + # in FCOS-DCN head, specifically need learning_rate and regularizer + self.conv = DeformableConvV2( + in_channels=ch_in, + out_channels=ch_out, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=groups, + weight_attr=ParamAttr( + initializer=initializer, learning_rate=1.), + bias_attr=True, + lr_scale=dcn_lr_scale, + regularizer=dcn_regularizer, + dcn_bias_regularizer=dcn_regularizer, + dcn_bias_lr_scale=dcn_lr_scale, + skip_quant=skip_quant) + + norm_lr = 0. if freeze_norm else 1. + param_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay) if norm_decay is not None else None) + bias_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay) if norm_decay is not None else None) + if norm_type == 'bn': + self.norm = nn.BatchNorm2D( + ch_out, weight_attr=param_attr, bias_attr=bias_attr) + elif norm_type == 'sync_bn': + self.norm = nn.SyncBatchNorm( + ch_out, weight_attr=param_attr, bias_attr=bias_attr) + elif norm_type == 'gn': + self.norm = nn.GroupNorm( + num_groups=norm_groups, + num_channels=ch_out, + weight_attr=param_attr, + bias_attr=bias_attr) + + def forward(self, inputs): + out = self.conv(inputs) + out = self.norm(out) + return out + + +class LiteConv(nn.Layer): + def __init__(self, + in_channels, + out_channels, + stride=1, + with_act=True, + norm_type='sync_bn', + name=None): + super(LiteConv, self).__init__() + self.lite_conv = nn.Sequential() + conv1 = ConvNormLayer( + in_channels, + in_channels, + filter_size=5, + stride=stride, + groups=in_channels, + norm_type=norm_type, + initializer=XavierUniform()) + conv2 = ConvNormLayer( + in_channels, + out_channels, + filter_size=1, + stride=stride, + norm_type=norm_type, + initializer=XavierUniform()) + conv3 = ConvNormLayer( + out_channels, + out_channels, + filter_size=1, + stride=stride, + norm_type=norm_type, + initializer=XavierUniform()) + conv4 = ConvNormLayer( + out_channels, + out_channels, + filter_size=5, + stride=stride, + groups=out_channels, + norm_type=norm_type, + initializer=XavierUniform()) + conv_list = [conv1, conv2, conv3, conv4] + self.lite_conv.add_sublayer('conv1', conv1) + self.lite_conv.add_sublayer('relu6_1', nn.ReLU6()) + self.lite_conv.add_sublayer('conv2', conv2) + if with_act: + self.lite_conv.add_sublayer('relu6_2', nn.ReLU6()) + self.lite_conv.add_sublayer('conv3', conv3) + self.lite_conv.add_sublayer('relu6_3', nn.ReLU6()) + self.lite_conv.add_sublayer('conv4', conv4) + if with_act: + self.lite_conv.add_sublayer('relu6_4', nn.ReLU6()) + + def forward(self, inputs): + out = self.lite_conv(inputs) + return out + + +class DropBlock(nn.Layer): + def __init__(self, block_size, keep_prob, name, data_format='NCHW'): + """ + DropBlock layer, see https://arxiv.org/abs/1810.12890 + + Args: + block_size (int): block size + keep_prob (int): keep probability + name (str): layer name + data_format (str): data format, NCHW or NHWC + """ + super(DropBlock, self).__init__() + self.block_size = block_size + self.keep_prob = keep_prob + self.name = name + self.data_format = data_format + + def forward(self, x): + if not self.training or self.keep_prob == 1: + return x + else: + gamma = (1. - self.keep_prob) / (self.block_size**2) + if self.data_format == 'NCHW': + shape = x.shape[2:] + else: + shape = x.shape[1:3] + for s in shape: + gamma *= s / (s - self.block_size + 1) + + matrix = paddle.cast(paddle.rand(x.shape) < gamma, x.dtype) + mask_inv = F.max_pool2d( + matrix, + self.block_size, + stride=1, + padding=self.block_size // 2, + data_format=self.data_format) + mask = 1. - mask_inv + y = x * mask * (mask.numel() / mask.sum()) + return y + + +@register +@serializable +class AnchorGeneratorSSD(object): + def __init__(self, + steps=[8, 16, 32, 64, 100, 300], + aspect_ratios=[[2.], [2., 3.], [2., 3.], [2., 3.], [2.], [2.]], + min_ratio=15, + max_ratio=90, + base_size=300, + min_sizes=[30.0, 60.0, 111.0, 162.0, 213.0, 264.0], + max_sizes=[60.0, 111.0, 162.0, 213.0, 264.0, 315.0], + offset=0.5, + flip=True, + clip=False, + min_max_aspect_ratios_order=False): + self.steps = steps + self.aspect_ratios = aspect_ratios + self.min_ratio = min_ratio + self.max_ratio = max_ratio + self.base_size = base_size + self.min_sizes = min_sizes + self.max_sizes = max_sizes + self.offset = offset + self.flip = flip + self.clip = clip + self.min_max_aspect_ratios_order = min_max_aspect_ratios_order + + if self.min_sizes == [] and self.max_sizes == []: + num_layer = len(aspect_ratios) + step = int( + math.floor(((self.max_ratio - self.min_ratio)) / (num_layer - 2 + ))) + for ratio in six.moves.range(self.min_ratio, self.max_ratio + 1, + step): + self.min_sizes.append(self.base_size * ratio / 100.) + self.max_sizes.append(self.base_size * (ratio + step) / 100.) + self.min_sizes = [self.base_size * .10] + self.min_sizes + self.max_sizes = [self.base_size * .20] + self.max_sizes + + self.num_priors = [] + for aspect_ratio, min_size, max_size in zip( + aspect_ratios, self.min_sizes, self.max_sizes): + if isinstance(min_size, (list, tuple)): + self.num_priors.append( + len(_to_list(min_size)) + len(_to_list(max_size))) + else: + self.num_priors.append((len(aspect_ratio) * 2 + 1) * len( + _to_list(min_size)) + len(_to_list(max_size))) + + def __call__(self, inputs, image): + boxes = [] + for input, min_size, max_size, aspect_ratio, step in zip( + inputs, self.min_sizes, self.max_sizes, self.aspect_ratios, + self.steps): + box, _ = ops.prior_box( + input=input, + image=image, + min_sizes=_to_list(min_size), + max_sizes=_to_list(max_size), + aspect_ratios=aspect_ratio, + flip=self.flip, + clip=self.clip, + steps=[step, step], + offset=self.offset, + min_max_aspect_ratios_order=self.min_max_aspect_ratios_order) + boxes.append(paddle.reshape(box, [-1, 4])) + return boxes + + +@register +@serializable +class RCNNBox(object): + __shared__ = ['num_classes'] + + def __init__(self, + prior_box_var=[10., 10., 5., 5.], + code_type="decode_center_size", + box_normalized=False, + num_classes=80): + super(RCNNBox, self).__init__() + self.prior_box_var = prior_box_var + self.code_type = code_type + self.box_normalized = box_normalized + self.num_classes = num_classes + + def __call__(self, bbox_head_out, rois, im_shape, scale_factor): + bbox_pred = bbox_head_out[0] + cls_prob = bbox_head_out[1] + roi = rois[0] + rois_num = rois[1] + + origin_shape = paddle.floor(im_shape / scale_factor + 0.5) + scale_list = [] + origin_shape_list = [] + + batch_size = 1 + if isinstance(roi, list): + batch_size = len(roi) + else: + batch_size = paddle.slice(paddle.shape(im_shape), [0], [0], [1]) + # bbox_pred.shape: [N, C*4] + for idx in range(batch_size): + roi_per_im = roi[idx] + rois_num_per_im = rois_num[idx] + expand_im_shape = paddle.expand(im_shape[idx, :], + [rois_num_per_im, 2]) + origin_shape_list.append(expand_im_shape) + + origin_shape = paddle.concat(origin_shape_list) + + # bbox_pred.shape: [N, C*4] + # C=num_classes in faster/mask rcnn(bbox_head), C=1 in cascade rcnn(cascade_head) + bbox = paddle.concat(roi) + if bbox.shape[0] == 0: + bbox = paddle.zeros([0, bbox_pred.shape[1]], dtype='float32') + else: + bbox = delta2bbox(bbox_pred, bbox, self.prior_box_var) + scores = cls_prob[:, :-1] + + # bbox.shape: [N, C, 4] + # bbox.shape[1] must be equal to scores.shape[1] + bbox_num_class = bbox.shape[1] + if bbox_num_class == 1: + bbox = paddle.tile(bbox, [1, self.num_classes, 1]) + + origin_h = paddle.unsqueeze(origin_shape[:, 0], axis=1) + origin_w = paddle.unsqueeze(origin_shape[:, 1], axis=1) + zeros = paddle.zeros_like(origin_h) + x1 = paddle.maximum(paddle.minimum(bbox[:, :, 0], origin_w), zeros) + y1 = paddle.maximum(paddle.minimum(bbox[:, :, 1], origin_h), zeros) + x2 = paddle.maximum(paddle.minimum(bbox[:, :, 2], origin_w), zeros) + y2 = paddle.maximum(paddle.minimum(bbox[:, :, 3], origin_h), zeros) + bbox = paddle.stack([x1, y1, x2, y2], axis=-1) + bboxes = (bbox, rois_num) + return bboxes, scores + + +@register +@serializable +class MultiClassNMS(object): + def __init__(self, + score_threshold=.05, + nms_top_k=-1, + keep_top_k=100, + nms_threshold=.5, + normalized=True, + nms_eta=1.0, + return_index=False, + return_rois_num=True): + super(MultiClassNMS, self).__init__() + self.score_threshold = score_threshold + self.nms_top_k = nms_top_k + self.keep_top_k = keep_top_k + self.nms_threshold = nms_threshold + self.normalized = normalized + self.nms_eta = nms_eta + self.return_index = return_index + self.return_rois_num = return_rois_num + + def __call__(self, bboxes, score, background_label=-1): + """ + bboxes (Tensor|List[Tensor]): 1. (Tensor) Predicted bboxes with shape + [N, M, 4], N is the batch size and M + is the number of bboxes + 2. (List[Tensor]) bboxes and bbox_num, + bboxes have shape of [M, C, 4], C + is the class number and bbox_num means + the number of bboxes of each batch with + shape [N,] + score (Tensor): Predicted scores with shape [N, C, M] or [M, C] + background_label (int): Ignore the background label; For example, RCNN + is num_classes and YOLO is -1. + """ + kwargs = self.__dict__.copy() + if isinstance(bboxes, tuple): + bboxes, bbox_num = bboxes + kwargs.update({'rois_num': bbox_num}) + if background_label > -1: + kwargs.update({'background_label': background_label}) + return ops.multiclass_nms(bboxes, score, **kwargs) + + +@register +@serializable +class MatrixNMS(object): + __append_doc__ = True + + def __init__(self, + score_threshold=.05, + post_threshold=.05, + nms_top_k=-1, + keep_top_k=100, + use_gaussian=False, + gaussian_sigma=2., + normalized=False, + background_label=0): + super(MatrixNMS, self).__init__() + self.score_threshold = score_threshold + self.post_threshold = post_threshold + self.nms_top_k = nms_top_k + self.keep_top_k = keep_top_k + self.normalized = normalized + self.use_gaussian = use_gaussian + self.gaussian_sigma = gaussian_sigma + self.background_label = background_label + + def __call__(self, bbox, score, *args): + return ops.matrix_nms( + bboxes=bbox, + scores=score, + score_threshold=self.score_threshold, + post_threshold=self.post_threshold, + nms_top_k=self.nms_top_k, + keep_top_k=self.keep_top_k, + use_gaussian=self.use_gaussian, + gaussian_sigma=self.gaussian_sigma, + background_label=self.background_label, + normalized=self.normalized) + + +@register +@serializable +class YOLOBox(object): + __shared__ = ['num_classes'] + + def __init__(self, + num_classes=80, + conf_thresh=0.005, + downsample_ratio=32, + clip_bbox=True, + scale_x_y=1.): + self.num_classes = num_classes + self.conf_thresh = conf_thresh + self.downsample_ratio = downsample_ratio + self.clip_bbox = clip_bbox + self.scale_x_y = scale_x_y + + def __call__(self, + yolo_head_out, + anchors, + im_shape, + scale_factor, + var_weight=None): + boxes_list = [] + scores_list = [] + origin_shape = im_shape / scale_factor + origin_shape = paddle.cast(origin_shape, 'int32') + for i, head_out in enumerate(yolo_head_out): + boxes, scores = ops.yolo_box(head_out, origin_shape, anchors[i], + self.num_classes, self.conf_thresh, + self.downsample_ratio // 2**i, + self.clip_bbox, self.scale_x_y) + boxes_list.append(boxes) + scores_list.append(paddle.transpose(scores, perm=[0, 2, 1])) + yolo_boxes = paddle.concat(boxes_list, axis=1) + yolo_scores = paddle.concat(scores_list, axis=2) + return yolo_boxes, yolo_scores + + +@register +@serializable +class SSDBox(object): + def __init__(self, is_normalized=True): + self.is_normalized = is_normalized + self.norm_delta = float(not self.is_normalized) + + def __call__(self, + preds, + prior_boxes, + im_shape, + scale_factor, + var_weight=None): + boxes, scores = preds + outputs = [] + for box, score, prior_box in zip(boxes, scores, prior_boxes): + pb_w = prior_box[:, 2] - prior_box[:, 0] + self.norm_delta + pb_h = prior_box[:, 3] - prior_box[:, 1] + self.norm_delta + pb_x = prior_box[:, 0] + pb_w * 0.5 + pb_y = prior_box[:, 1] + pb_h * 0.5 + out_x = pb_x + box[:, :, 0] * pb_w * 0.1 + out_y = pb_y + box[:, :, 1] * pb_h * 0.1 + out_w = paddle.exp(box[:, :, 2] * 0.2) * pb_w + out_h = paddle.exp(box[:, :, 3] * 0.2) * pb_h + + if self.is_normalized: + h = paddle.unsqueeze( + im_shape[:, 0] / scale_factor[:, 0], axis=-1) + w = paddle.unsqueeze( + im_shape[:, 1] / scale_factor[:, 1], axis=-1) + output = paddle.stack( + [(out_x - out_w / 2.) * w, (out_y - out_h / 2.) * h, + (out_x + out_w / 2.) * w, (out_y + out_h / 2.) * h], + axis=-1) + else: + output = paddle.stack( + [ + out_x - out_w / 2., out_y - out_h / 2., + out_x + out_w / 2. - 1., out_y + out_h / 2. - 1. + ], + axis=-1) + outputs.append(output) + boxes = paddle.concat(outputs, axis=1) + + scores = F.softmax(paddle.concat(scores, axis=1)) + scores = paddle.transpose(scores, [0, 2, 1]) + + return boxes, scores + + +@register +@serializable +class AnchorGrid(object): + """Generate anchor grid + + Args: + image_size (int or list): input image size, may be a single integer or + list of [h, w]. Default: 512 + min_level (int): min level of the feature pyramid. Default: 3 + max_level (int): max level of the feature pyramid. Default: 7 + anchor_base_scale: base anchor scale. Default: 4 + num_scales: number of anchor scales. Default: 3 + aspect_ratios: aspect ratios. default: [[1, 1], [1.4, 0.7], [0.7, 1.4]] + """ + + def __init__(self, + image_size=512, + min_level=3, + max_level=7, + anchor_base_scale=4, + num_scales=3, + aspect_ratios=[[1, 1], [1.4, 0.7], [0.7, 1.4]]): + super(AnchorGrid, self).__init__() + if isinstance(image_size, Integral): + self.image_size = [image_size, image_size] + else: + self.image_size = image_size + for dim in self.image_size: + assert dim % 2 ** max_level == 0, \ + "image size should be multiple of the max level stride" + self.min_level = min_level + self.max_level = max_level + self.anchor_base_scale = anchor_base_scale + self.num_scales = num_scales + self.aspect_ratios = aspect_ratios + + @property + def base_cell(self): + if not hasattr(self, '_base_cell'): + self._base_cell = self.make_cell() + return self._base_cell + + def make_cell(self): + scales = [2**(i / self.num_scales) for i in range(self.num_scales)] + scales = np.array(scales) + ratios = np.array(self.aspect_ratios) + ws = np.outer(scales, ratios[:, 0]).reshape(-1, 1) + hs = np.outer(scales, ratios[:, 1]).reshape(-1, 1) + anchors = np.hstack((-0.5 * ws, -0.5 * hs, 0.5 * ws, 0.5 * hs)) + return anchors + + def make_grid(self, stride): + cell = self.base_cell * stride * self.anchor_base_scale + x_steps = np.arange(stride // 2, self.image_size[1], stride) + y_steps = np.arange(stride // 2, self.image_size[0], stride) + offset_x, offset_y = np.meshgrid(x_steps, y_steps) + offset_x = offset_x.flatten() + offset_y = offset_y.flatten() + offsets = np.stack((offset_x, offset_y, offset_x, offset_y), axis=-1) + offsets = offsets[:, np.newaxis, :] + return (cell + offsets).reshape(-1, 4) + + def generate(self): + return [ + self.make_grid(2**l) + for l in range(self.min_level, self.max_level + 1) + ] + + def __call__(self): + if not hasattr(self, '_anchor_vars'): + anchor_vars = [] + helper = LayerHelper('anchor_grid') + for idx, l in enumerate(range(self.min_level, self.max_level + 1)): + stride = 2**l + anchors = self.make_grid(stride) + var = helper.create_parameter( + attr=ParamAttr(name='anchors_{}'.format(idx)), + shape=anchors.shape, + dtype='float32', + stop_gradient=True, + default_initializer=NumpyArrayInitializer(anchors)) + anchor_vars.append(var) + var.persistable = True + self._anchor_vars = anchor_vars + + return self._anchor_vars + + +@register +@serializable +class FCOSBox(object): + __shared__ = ['num_classes'] + + def __init__(self, num_classes=80): + super(FCOSBox, self).__init__() + self.num_classes = num_classes + + def _merge_hw(self, inputs, ch_type="channel_first"): + """ + Merge h and w of the feature map into one dimension. + Args: + inputs (Tensor): Tensor of the input feature map + ch_type (str): "channel_first" or "channel_last" style + Return: + new_shape (Tensor): The new shape after h and w merged + """ + shape_ = paddle.shape(inputs) + bs, ch, hi, wi = shape_[0], shape_[1], shape_[2], shape_[3] + img_size = hi * wi + img_size.stop_gradient = True + if ch_type == "channel_first": + new_shape = paddle.concat([bs, ch, img_size]) + elif ch_type == "channel_last": + new_shape = paddle.concat([bs, img_size, ch]) + else: + raise KeyError("Wrong ch_type %s" % ch_type) + new_shape.stop_gradient = True + return new_shape + + def _postprocessing_by_level(self, locations, box_cls, box_reg, box_ctn, + scale_factor): + """ + Postprocess each layer of the output with corresponding locations. + Args: + locations (Tensor): anchor points for current layer, [H*W, 2] + box_cls (Tensor): categories prediction, [N, C, H, W], + C is the number of classes + box_reg (Tensor): bounding box prediction, [N, 4, H, W] + box_ctn (Tensor): centerness prediction, [N, 1, H, W] + scale_factor (Tensor): [h_scale, w_scale] for input images + Return: + box_cls_ch_last (Tensor): score for each category, in [N, C, M] + C is the number of classes and M is the number of anchor points + box_reg_decoding (Tensor): decoded bounding box, in [N, M, 4] + last dimension is [x1, y1, x2, y2] + """ + act_shape_cls = self._merge_hw(box_cls) + box_cls_ch_last = paddle.reshape(x=box_cls, shape=act_shape_cls) + box_cls_ch_last = F.sigmoid(box_cls_ch_last) + + act_shape_reg = self._merge_hw(box_reg) + box_reg_ch_last = paddle.reshape(x=box_reg, shape=act_shape_reg) + box_reg_ch_last = paddle.transpose(box_reg_ch_last, perm=[0, 2, 1]) + box_reg_decoding = paddle.stack( + [ + locations[:, 0] - box_reg_ch_last[:, :, 0], + locations[:, 1] - box_reg_ch_last[:, :, 1], + locations[:, 0] + box_reg_ch_last[:, :, 2], + locations[:, 1] + box_reg_ch_last[:, :, 3] + ], + axis=1) + box_reg_decoding = paddle.transpose(box_reg_decoding, perm=[0, 2, 1]) + + act_shape_ctn = self._merge_hw(box_ctn) + box_ctn_ch_last = paddle.reshape(x=box_ctn, shape=act_shape_ctn) + box_ctn_ch_last = F.sigmoid(box_ctn_ch_last) + + # recover the location to original image + im_scale = paddle.concat([scale_factor, scale_factor], axis=1) + im_scale = paddle.expand(im_scale, [box_reg_decoding.shape[0], 4]) + im_scale = paddle.reshape(im_scale, [box_reg_decoding.shape[0], -1, 4]) + box_reg_decoding = box_reg_decoding / im_scale + box_cls_ch_last = box_cls_ch_last * box_ctn_ch_last + return box_cls_ch_last, box_reg_decoding + + def __call__(self, locations, cls_logits, bboxes_reg, centerness, + scale_factor): + pred_boxes_ = [] + pred_scores_ = [] + for pts, cls, box, ctn in zip(locations, cls_logits, bboxes_reg, + centerness): + pred_scores_lvl, pred_boxes_lvl = self._postprocessing_by_level( + pts, cls, box, ctn, scale_factor) + pred_boxes_.append(pred_boxes_lvl) + pred_scores_.append(pred_scores_lvl) + pred_boxes = paddle.concat(pred_boxes_, axis=1) + pred_scores = paddle.concat(pred_scores_, axis=2) + return pred_boxes, pred_scores + + +@register +class TTFBox(object): + __shared__ = ['down_ratio'] + + def __init__(self, max_per_img=100, score_thresh=0.01, down_ratio=4): + super(TTFBox, self).__init__() + self.max_per_img = max_per_img + self.score_thresh = score_thresh + self.down_ratio = down_ratio + + def _simple_nms(self, heat, kernel=3): + """ + Use maxpool to filter the max score, get local peaks. + """ + pad = (kernel - 1) // 2 + hmax = F.max_pool2d(heat, kernel, stride=1, padding=pad) + keep = paddle.cast(hmax == heat, 'float32') + return heat * keep + + def _topk(self, scores): + """ + Select top k scores and decode to get xy coordinates. + """ + k = self.max_per_img + shape_fm = paddle.shape(scores) + shape_fm.stop_gradient = True + cat, height, width = shape_fm[1], shape_fm[2], shape_fm[3] + # batch size is 1 + scores_r = paddle.reshape(scores, [cat, -1]) + topk_scores, topk_inds = paddle.topk(scores_r, k) + topk_scores, topk_inds = paddle.topk(scores_r, k) + topk_ys = topk_inds // width + topk_xs = topk_inds % width + + topk_score_r = paddle.reshape(topk_scores, [-1]) + topk_score, topk_ind = paddle.topk(topk_score_r, k) + k_t = paddle.full(paddle.shape(topk_ind), k, dtype='int64') + topk_clses = paddle.cast(paddle.floor_divide(topk_ind, k_t), 'float32') + + topk_inds = paddle.reshape(topk_inds, [-1]) + topk_ys = paddle.reshape(topk_ys, [-1, 1]) + topk_xs = paddle.reshape(topk_xs, [-1, 1]) + topk_inds = paddle.gather(topk_inds, topk_ind) + topk_ys = paddle.gather(topk_ys, topk_ind) + topk_xs = paddle.gather(topk_xs, topk_ind) + + return topk_score, topk_inds, topk_clses, topk_ys, topk_xs + + def _decode(self, hm, wh, im_shape, scale_factor): + heatmap = F.sigmoid(hm) + heat = self._simple_nms(heatmap) + scores, inds, clses, ys, xs = self._topk(heat) + ys = paddle.cast(ys, 'float32') * self.down_ratio + xs = paddle.cast(xs, 'float32') * self.down_ratio + scores = paddle.tensor.unsqueeze(scores, [1]) + clses = paddle.tensor.unsqueeze(clses, [1]) + + wh_t = paddle.transpose(wh, [0, 2, 3, 1]) + wh = paddle.reshape(wh_t, [-1, paddle.shape(wh_t)[-1]]) + wh = paddle.gather(wh, inds) + + x1 = xs - wh[:, 0:1] + y1 = ys - wh[:, 1:2] + x2 = xs + wh[:, 2:3] + y2 = ys + wh[:, 3:4] + + bboxes = paddle.concat([x1, y1, x2, y2], axis=1) + + scale_y = scale_factor[:, 0:1] + scale_x = scale_factor[:, 1:2] + scale_expand = paddle.concat( + [scale_x, scale_y, scale_x, scale_y], axis=1) + boxes_shape = paddle.shape(bboxes) + boxes_shape.stop_gradient = True + scale_expand = paddle.expand(scale_expand, shape=boxes_shape) + bboxes = paddle.divide(bboxes, scale_expand) + results = paddle.concat([clses, scores, bboxes], axis=1) + # hack: append result with cls=-1 and score=1. to avoid all scores + # are less than score_thresh which may cause error in gather. + fill_r = paddle.to_tensor(np.array([[-1, 1, 0, 0, 0, 0]])) + fill_r = paddle.cast(fill_r, results.dtype) + results = paddle.concat([results, fill_r]) + scores = results[:, 1] + valid_ind = paddle.nonzero(scores > self.score_thresh) + results = paddle.gather(results, valid_ind) + return results, paddle.shape(results)[0:1] + + def __call__(self, hm, wh, im_shape, scale_factor): + results = [] + results_num = [] + for i in range(scale_factor.shape[0]): + result, num = self._decode(hm[i:i + 1, ], wh[i:i + 1, ], + im_shape[i:i + 1, ], + scale_factor[i:i + 1, ]) + results.append(result) + results_num.append(num) + results = paddle.concat(results, axis=0) + results_num = paddle.concat(results_num, axis=0) + return results, results_num + + +@register +@serializable +class JDEBox(object): + __shared__ = ['num_classes'] + + def __init__(self, num_classes=1, conf_thresh=0.3, downsample_ratio=32): + self.num_classes = num_classes + self.conf_thresh = conf_thresh + self.downsample_ratio = downsample_ratio + + def generate_anchor(self, nGh, nGw, anchor_wh): + nA = len(anchor_wh) + yv, xv = paddle.meshgrid([paddle.arange(nGh), paddle.arange(nGw)]) + mesh = paddle.stack( + (xv, yv), axis=0).cast(dtype='float32') # 2 x nGh x nGw + meshs = paddle.tile(mesh, [nA, 1, 1, 1]) + + anchor_offset_mesh = anchor_wh[:, :, None][:, :, :, None].repeat( + int(nGh), axis=-2).repeat( + int(nGw), axis=-1) + anchor_offset_mesh = paddle.to_tensor( + anchor_offset_mesh.astype(np.float32)) + # nA x 2 x nGh x nGw + + anchor_mesh = paddle.concat([meshs, anchor_offset_mesh], axis=1) + anchor_mesh = paddle.transpose(anchor_mesh, + [0, 2, 3, 1]) # (nA x nGh x nGw) x 4 + return anchor_mesh + + def decode_delta(self, delta, fg_anchor_list): + px, py, pw, ph = fg_anchor_list[:, 0], fg_anchor_list[:,1], \ + fg_anchor_list[:, 2], fg_anchor_list[:,3] + dx, dy, dw, dh = delta[:, 0], delta[:, 1], delta[:, 2], delta[:, 3] + gx = pw * dx + px + gy = ph * dy + py + gw = pw * paddle.exp(dw) + gh = ph * paddle.exp(dh) + gx1 = gx - gw * 0.5 + gy1 = gy - gh * 0.5 + gx2 = gx + gw * 0.5 + gy2 = gy + gh * 0.5 + return paddle.stack([gx1, gy1, gx2, gy2], axis=1) + + def decode_delta_map(self, nA, nGh, nGw, delta_map, anchor_vec): + anchor_mesh = self.generate_anchor(nGh, nGw, anchor_vec) + anchor_mesh = paddle.unsqueeze(anchor_mesh, 0) + pred_list = self.decode_delta( + paddle.reshape( + delta_map, shape=[-1, 4]), + paddle.reshape( + anchor_mesh, shape=[-1, 4])) + pred_map = paddle.reshape(pred_list, shape=[nA * nGh * nGw, 4]) + return pred_map + + def _postprocessing_by_level(self, nA, stride, head_out, anchor_vec): + boxes_shape = head_out.shape # [nB, nA*6, nGh, nGw] + nGh, nGw = boxes_shape[-2], boxes_shape[-1] + nB = 1 # TODO: only support bs=1 now + boxes_list, scores_list = [], [] + for idx in range(nB): + p = paddle.reshape( + head_out[idx], shape=[nA, self.num_classes + 5, nGh, nGw]) + p = paddle.transpose(p, perm=[0, 2, 3, 1]) # [nA, nGh, nGw, 6] + delta_map = p[:, :, :, :4] + boxes = self.decode_delta_map(nA, nGh, nGw, delta_map, anchor_vec) + # [nA * nGh * nGw, 4] + boxes_list.append(boxes * stride) + + p_conf = paddle.transpose( + p[:, :, :, 4:6], perm=[3, 0, 1, 2]) # [2, nA, nGh, nGw] + p_conf = F.softmax( + p_conf, axis=0)[1, :, :, :].unsqueeze(-1) # [nA, nGh, nGw, 1] + scores = paddle.reshape(p_conf, shape=[nA * nGh * nGw, 1]) + scores_list.append(scores) + + boxes_results = paddle.stack(boxes_list) + scores_results = paddle.stack(scores_list) + return boxes_results, scores_results + + def __call__(self, yolo_head_out, anchors): + bbox_pred_list = [] + for i, head_out in enumerate(yolo_head_out): + stride = self.downsample_ratio // 2**i + anc_w, anc_h = anchors[i][0::2], anchors[i][1::2] + anchor_vec = np.stack((anc_w, anc_h), axis=1) / stride + nA = len(anc_w) + boxes, scores = self._postprocessing_by_level(nA, stride, head_out, + anchor_vec) + bbox_pred_list.append(paddle.concat([boxes, scores], axis=-1)) + + yolo_boxes_scores = paddle.concat(bbox_pred_list, axis=1) + boxes_idx_over_conf_thr = paddle.nonzero( + yolo_boxes_scores[:, :, -1] > self.conf_thresh) + boxes_idx_over_conf_thr.stop_gradient = True + + return boxes_idx_over_conf_thr, yolo_boxes_scores + + +@register +@serializable +class MaskMatrixNMS(object): + """ + Matrix NMS for multi-class masks. + Args: + update_threshold (float): Updated threshold of categroy score in second time. + pre_nms_top_n (int): Number of total instance to be kept per image before NMS + post_nms_top_n (int): Number of total instance to be kept per image after NMS. + kernel (str): 'linear' or 'gaussian'. + sigma (float): std in gaussian method. + Input: + seg_preds (Variable): shape (n, h, w), segmentation feature maps + seg_masks (Variable): shape (n, h, w), segmentation feature maps + cate_labels (Variable): shape (n), mask labels in descending order + cate_scores (Variable): shape (n), mask scores in descending order + sum_masks (Variable): a float tensor of the sum of seg_masks + Returns: + Variable: cate_scores, tensors of shape (n) + """ + + def __init__(self, + update_threshold=0.05, + pre_nms_top_n=500, + post_nms_top_n=100, + kernel='gaussian', + sigma=2.0): + super(MaskMatrixNMS, self).__init__() + self.update_threshold = update_threshold + self.pre_nms_top_n = pre_nms_top_n + self.post_nms_top_n = post_nms_top_n + self.kernel = kernel + self.sigma = sigma + + def _sort_score(self, scores, top_num): + if paddle.shape(scores)[0] > top_num: + return paddle.topk(scores, top_num)[1] + else: + return paddle.argsort(scores, descending=True) + + def __call__(self, + seg_preds, + seg_masks, + cate_labels, + cate_scores, + sum_masks=None): + # sort and keep top nms_pre + sort_inds = self._sort_score(cate_scores, self.pre_nms_top_n) + seg_masks = paddle.gather(seg_masks, index=sort_inds) + seg_preds = paddle.gather(seg_preds, index=sort_inds) + sum_masks = paddle.gather(sum_masks, index=sort_inds) + cate_scores = paddle.gather(cate_scores, index=sort_inds) + cate_labels = paddle.gather(cate_labels, index=sort_inds) + + seg_masks = paddle.flatten(seg_masks, start_axis=1, stop_axis=-1) + # inter. + inter_matrix = paddle.mm(seg_masks, paddle.transpose(seg_masks, [1, 0])) + n_samples = paddle.shape(cate_labels) + # union. + sum_masks_x = paddle.expand(sum_masks, shape=[n_samples, n_samples]) + # iou. + iou_matrix = (inter_matrix / ( + sum_masks_x + paddle.transpose(sum_masks_x, [1, 0]) - inter_matrix)) + iou_matrix = paddle.triu(iou_matrix, diagonal=1) + # label_specific matrix. + cate_labels_x = paddle.expand(cate_labels, shape=[n_samples, n_samples]) + label_matrix = paddle.cast( + (cate_labels_x == paddle.transpose(cate_labels_x, [1, 0])), + 'float32') + label_matrix = paddle.triu(label_matrix, diagonal=1) + + # IoU compensation + compensate_iou = paddle.max((iou_matrix * label_matrix), axis=0) + compensate_iou = paddle.expand( + compensate_iou, shape=[n_samples, n_samples]) + compensate_iou = paddle.transpose(compensate_iou, [1, 0]) + + # IoU decay + decay_iou = iou_matrix * label_matrix + + # matrix nms + if self.kernel == 'gaussian': + decay_matrix = paddle.exp(-1 * self.sigma * (decay_iou**2)) + compensate_matrix = paddle.exp(-1 * self.sigma * + (compensate_iou**2)) + decay_coefficient = paddle.min(decay_matrix / compensate_matrix, + axis=0) + elif self.kernel == 'linear': + decay_matrix = (1 - decay_iou) / (1 - compensate_iou) + decay_coefficient = paddle.min(decay_matrix, axis=0) + else: + raise NotImplementedError + + # update the score. + cate_scores = cate_scores * decay_coefficient + y = paddle.zeros(shape=paddle.shape(cate_scores), dtype='float32') + keep = paddle.where(cate_scores >= self.update_threshold, cate_scores, + y) + keep = paddle.nonzero(keep) + keep = paddle.squeeze(keep, axis=[1]) + # Prevent empty and increase fake data + keep = paddle.concat( + [keep, paddle.cast(paddle.shape(cate_scores)[0] - 1, 'int64')]) + + seg_preds = paddle.gather(seg_preds, index=keep) + cate_scores = paddle.gather(cate_scores, index=keep) + cate_labels = paddle.gather(cate_labels, index=keep) + + # sort and keep top_k + sort_inds = self._sort_score(cate_scores, self.post_nms_top_n) + seg_preds = paddle.gather(seg_preds, index=sort_inds) + cate_scores = paddle.gather(cate_scores, index=sort_inds) + cate_labels = paddle.gather(cate_labels, index=sort_inds) + return seg_preds, cate_scores, cate_labels + + +def Conv2d(in_channels, + out_channels, + kernel_size, + stride=1, + padding=0, + dilation=1, + groups=1, + bias=True, + weight_init=Normal(std=0.001), + bias_init=Constant(0.)): + weight_attr = paddle.framework.ParamAttr(initializer=weight_init) + if bias: + bias_attr = paddle.framework.ParamAttr(initializer=bias_init) + else: + bias_attr = False + conv = nn.Conv2D( + in_channels, + out_channels, + kernel_size, + stride, + padding, + dilation, + groups, + weight_attr=weight_attr, + bias_attr=bias_attr) + return conv + + +def ConvTranspose2d(in_channels, + out_channels, + kernel_size, + stride=1, + padding=0, + output_padding=0, + groups=1, + bias=True, + dilation=1, + weight_init=Normal(std=0.001), + bias_init=Constant(0.)): + weight_attr = paddle.framework.ParamAttr(initializer=weight_init) + if bias: + bias_attr = paddle.framework.ParamAttr(initializer=bias_init) + else: + bias_attr = False + conv = nn.Conv2DTranspose( + in_channels, + out_channels, + kernel_size, + stride, + padding, + output_padding, + dilation, + groups, + weight_attr=weight_attr, + bias_attr=bias_attr) + return conv + + +def BatchNorm2d(num_features, eps=1e-05, momentum=0.9, affine=True): + if not affine: + weight_attr = False + bias_attr = False + else: + weight_attr = None + bias_attr = None + batchnorm = nn.BatchNorm2D( + num_features, + momentum, + eps, + weight_attr=weight_attr, + bias_attr=bias_attr) + return batchnorm + + +def ReLU(): + return nn.ReLU() + + +def Upsample(scale_factor=None, mode='nearest', align_corners=False): + return nn.Upsample(None, scale_factor, mode, align_corners) + + +def MaxPool(kernel_size, stride, padding, ceil_mode=False): + return nn.MaxPool2D(kernel_size, stride, padding, ceil_mode=ceil_mode) + + +class Concat(nn.Layer): + def __init__(self, dim=0): + super(Concat, self).__init__() + self.dim = dim + + def forward(self, inputs): + return paddle.concat(inputs, axis=self.dim) + + def extra_repr(self): + return 'dim={}'.format(self.dim) + + +def _convert_attention_mask(attn_mask, dtype): + """ + Convert the attention mask to the target dtype we expect. + Parameters: + attn_mask (Tensor, optional): A tensor used in multi-head attention + to prevents attention to some unwanted positions, usually the + paddings or the subsequent positions. It is a tensor with shape + broadcasted to `[batch_size, n_head, sequence_length, sequence_length]`. + When the data type is bool, the unwanted positions have `False` + values and the others have `True` values. When the data type is + int, the unwanted positions have 0 values and the others have 1 + values. When the data type is float, the unwanted positions have + `-INF` values and the others have 0 values. It can be None when + nothing wanted or needed to be prevented attention to. Default None. + dtype (VarType): The target type of `attn_mask` we expect. + Returns: + Tensor: A Tensor with shape same as input `attn_mask`, with data type `dtype`. + """ + return nn.layer.transformer._convert_attention_mask(attn_mask, dtype) + + +class MultiHeadAttention(nn.Layer): + """ + Attention mapps queries and a set of key-value pairs to outputs, and + Multi-Head Attention performs multiple parallel attention to jointly attending + to information from different representation subspaces. + + Please refer to `Attention Is All You Need `_ + for more details. + + Parameters: + embed_dim (int): The expected feature size in the input and output. + num_heads (int): The number of heads in multi-head attention. + dropout (float, optional): The dropout probability used on attention + weights to drop some attention targets. 0 for no dropout. Default 0 + kdim (int, optional): The feature size in key. If None, assumed equal to + `embed_dim`. Default None. + vdim (int, optional): The feature size in value. If None, assumed equal to + `embed_dim`. Default None. + need_weights (bool, optional): Indicate whether to return the attention + weights. Default False. + + Examples: + + .. code-block:: python + + import paddle + + # encoder input: [batch_size, sequence_length, d_model] + query = paddle.rand((2, 4, 128)) + # self attention mask: [batch_size, num_heads, query_len, query_len] + attn_mask = paddle.rand((2, 2, 4, 4)) + multi_head_attn = paddle.nn.MultiHeadAttention(128, 2) + output = multi_head_attn(query, None, None, attn_mask=attn_mask) # [2, 4, 128] + """ + + def __init__(self, + embed_dim, + num_heads, + dropout=0., + kdim=None, + vdim=None, + need_weights=False): + super(MultiHeadAttention, self).__init__() + self.embed_dim = embed_dim + self.kdim = kdim if kdim is not None else embed_dim + self.vdim = vdim if vdim is not None else embed_dim + self._qkv_same_embed_dim = self.kdim == embed_dim and self.vdim == embed_dim + + self.num_heads = num_heads + self.dropout = dropout + self.need_weights = need_weights + + self.head_dim = embed_dim // num_heads + assert self.head_dim * num_heads == self.embed_dim, "embed_dim must be divisible by num_heads" + + if self._qkv_same_embed_dim: + self.in_proj_weight = self.create_parameter( + shape=[embed_dim, 3 * embed_dim], + attr=None, + dtype=self._dtype, + is_bias=False) + self.in_proj_bias = self.create_parameter( + shape=[3 * embed_dim], + attr=None, + dtype=self._dtype, + is_bias=True) + else: + self.q_proj = nn.Linear(embed_dim, embed_dim) + self.k_proj = nn.Linear(self.kdim, embed_dim) + self.v_proj = nn.Linear(self.vdim, embed_dim) + + self.out_proj = nn.Linear(embed_dim, embed_dim) + self._type_list = ('q_proj', 'k_proj', 'v_proj') + + self._reset_parameters() + + def _reset_parameters(self): + for p in self.parameters(): + if p.dim() > 1: + xavier_uniform_(p) + else: + constant_(p) + + def compute_qkv(self, tensor, index): + if self._qkv_same_embed_dim: + tensor = F.linear( + x=tensor, + weight=self.in_proj_weight[:, index * self.embed_dim:(index + 1) + * self.embed_dim], + bias=self.in_proj_bias[index * self.embed_dim:(index + 1) * + self.embed_dim] + if self.in_proj_bias is not None else None) + else: + tensor = getattr(self, self._type_list[index])(tensor) + tensor = tensor.reshape( + [0, 0, self.num_heads, self.head_dim]).transpose([0, 2, 1, 3]) + return tensor + + def forward(self, query, key=None, value=None, attn_mask=None): + r""" + Applies multi-head attention to map queries and a set of key-value pairs + to outputs. + + Parameters: + query (Tensor): The queries for multi-head attention. It is a + tensor with shape `[batch_size, query_length, embed_dim]`. The + data type should be float32 or float64. + key (Tensor, optional): The keys for multi-head attention. It is + a tensor with shape `[batch_size, key_length, kdim]`. The + data type should be float32 or float64. If None, use `query` as + `key`. Default None. + value (Tensor, optional): The values for multi-head attention. It + is a tensor with shape `[batch_size, value_length, vdim]`. + The data type should be float32 or float64. If None, use `query` as + `value`. Default None. + attn_mask (Tensor, optional): A tensor used in multi-head attention + to prevents attention to some unwanted positions, usually the + paddings or the subsequent positions. It is a tensor with shape + broadcasted to `[batch_size, n_head, sequence_length, sequence_length]`. + When the data type is bool, the unwanted positions have `False` + values and the others have `True` values. When the data type is + int, the unwanted positions have 0 values and the others have 1 + values. When the data type is float, the unwanted positions have + `-INF` values and the others have 0 values. It can be None when + nothing wanted or needed to be prevented attention to. Default None. + + Returns: + Tensor|tuple: It is a tensor that has the same shape and data type \ + as `query`, representing attention output. Or a tuple if \ + `need_weights` is True or `cache` is not None. If `need_weights` \ + is True, except for attention output, the tuple also includes \ + the attention weights tensor shaped `[batch_size, num_heads, query_length, key_length]`. \ + If `cache` is not None, the tuple then includes the new cache \ + having the same type as `cache`, and if it is `StaticCache`, it \ + is same as the input `cache`, if it is `Cache`, the new cache \ + reserves tensors concatanating raw tensors with intermediate \ + results of current query. + """ + key = query if key is None else key + value = query if value is None else value + # compute q ,k ,v + q, k, v = (self.compute_qkv(t, i) + for i, t in enumerate([query, key, value])) + + # scale dot product attention + product = paddle.matmul(x=q, y=k, transpose_y=True) + scaling = float(self.head_dim)**-0.5 + product = product * scaling + + if attn_mask is not None: + # Support bool or int mask + attn_mask = _convert_attention_mask(attn_mask, product.dtype) + product = product + attn_mask + weights = F.softmax(product) + if self.dropout: + weights = F.dropout( + weights, + self.dropout, + training=self.training, + mode="upscale_in_train") + + out = paddle.matmul(weights, v) + + # combine heads + out = paddle.transpose(out, perm=[0, 2, 1, 3]) + out = paddle.reshape(x=out, shape=[0, 0, out.shape[2] * out.shape[3]]) + + # project to output + out = self.out_proj(out) + + outs = [out] + if self.need_weights: + outs.append(weights) + return out if len(outs) == 1 else tuple(outs) + + +@register +class ConvMixer(nn.Layer): + def __init__( + self, + dim, + depth, + kernel_size=3, ): + super().__init__() + self.dim = dim + self.depth = depth + self.kernel_size = kernel_size + + self.mixer = self.conv_mixer(dim, depth, kernel_size) + + def forward(self, x): + return self.mixer(x) + + @staticmethod + def conv_mixer( + dim, + depth, + kernel_size, ): + Seq, ActBn = nn.Sequential, lambda x: Seq(x, nn.GELU(), nn.BatchNorm2D(dim)) + Residual = type('Residual', (Seq, ), + {'forward': lambda self, x: self[0](x) + x}) + return Seq(*[ + Seq(Residual( + ActBn( + nn.Conv2D( + dim, dim, kernel_size, groups=dim, padding="same"))), + ActBn(nn.Conv2D(dim, dim, 1))) for i in range(depth) + ]) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__init__.py new file mode 100644 index 000000000..83389c08e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__init__.py @@ -0,0 +1,41 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import yolo_loss +from . import iou_aware_loss +from . import iou_loss +from . import ssd_loss +from . import fcos_loss +from . import solov2_loss +from . import ctfocal_loss +from . import keypoint_loss +from . import jde_loss +from . import fairmot_loss +from . import gfocal_loss +from . import detr_loss +from . import sparsercnn_loss + +from .yolo_loss import * +from .iou_aware_loss import * +from .iou_loss import * +from .ssd_loss import * +from .fcos_loss import * +from .solov2_loss import * +from .ctfocal_loss import * +from .keypoint_loss import * +from .jde_loss import * +from .fairmot_loss import * +from .gfocal_loss import * +from .detr_loss import * +from .sparsercnn_loss import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..f19224ba0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ctfocal_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ctfocal_loss.cpython-37.pyc new file mode 100644 index 000000000..093a6a7a2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ctfocal_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/detr_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/detr_loss.cpython-37.pyc new file mode 100644 index 000000000..998b42ff4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/detr_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fairmot_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fairmot_loss.cpython-37.pyc new file mode 100644 index 000000000..1b031f7a9 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fairmot_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fcos_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fcos_loss.cpython-37.pyc new file mode 100644 index 000000000..4aa367a0f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/fcos_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/gfocal_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/gfocal_loss.cpython-37.pyc new file mode 100644 index 000000000..bea48b9bc Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/gfocal_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_aware_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_aware_loss.cpython-37.pyc new file mode 100644 index 000000000..9f06e288c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_aware_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_loss.cpython-37.pyc new file mode 100644 index 000000000..524270db2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/iou_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/jde_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/jde_loss.cpython-37.pyc new file mode 100644 index 000000000..aae1c7aef Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/jde_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/keypoint_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/keypoint_loss.cpython-37.pyc new file mode 100644 index 000000000..696189238 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/keypoint_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/solov2_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/solov2_loss.cpython-37.pyc new file mode 100644 index 000000000..84bc2fcda Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/solov2_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/sparsercnn_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/sparsercnn_loss.cpython-37.pyc new file mode 100644 index 000000000..c791e49cb Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/sparsercnn_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ssd_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ssd_loss.cpython-37.pyc new file mode 100644 index 000000000..f2fd2b9c3 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/ssd_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/varifocal_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/varifocal_loss.cpython-37.pyc new file mode 100644 index 000000000..bf1a6509e Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/varifocal_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/yolo_loss.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/yolo_loss.cpython-37.pyc new file mode 100644 index 000000000..18a2ae010 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/__pycache__/yolo_loss.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ctfocal_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ctfocal_loss.py new file mode 100644 index 000000000..dd00eb854 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ctfocal_loss.py @@ -0,0 +1,68 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle + +from ppdet.core.workspace import register, serializable + +__all__ = ['CTFocalLoss'] + + +@register +@serializable +class CTFocalLoss(object): + """ + CTFocalLoss: CornerNet & CenterNet Focal Loss + Args: + loss_weight (float): loss weight + gamma (float): gamma parameter for Focal Loss + """ + + def __init__(self, loss_weight=1., gamma=2.0): + self.loss_weight = loss_weight + self.gamma = gamma + + def __call__(self, pred, target): + """ + Calculate the loss + Args: + pred (Tensor): heatmap prediction + target (Tensor): target for positive samples + Return: + ct_focal_loss (Tensor): Focal Loss used in CornerNet & CenterNet. + Note that the values in target are in [0, 1] since gaussian is + used to reduce the punishment and we treat [0, 1) as neg example. + """ + fg_map = paddle.cast(target == 1, 'float32') + fg_map.stop_gradient = True + bg_map = paddle.cast(target < 1, 'float32') + bg_map.stop_gradient = True + + neg_weights = paddle.pow(1 - target, 4) + pos_loss = 0 - paddle.log(pred) * paddle.pow(1 - pred, + self.gamma) * fg_map + + neg_loss = 0 - paddle.log(1 - pred) * paddle.pow( + pred, self.gamma) * neg_weights * bg_map + pos_loss = paddle.sum(pos_loss) + neg_loss = paddle.sum(neg_loss) + + fg_num = paddle.sum(fg_map) + ct_focal_loss = (pos_loss + neg_loss) / ( + fg_num + paddle.cast(fg_num == 0, 'float32')) + return ct_focal_loss * self.loss_weight diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/detr_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/detr_loss.py new file mode 100644 index 000000000..5a589d4a2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/detr_loss.py @@ -0,0 +1,230 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +from .iou_loss import GIoULoss +from ..transformers import bbox_cxcywh_to_xyxy, sigmoid_focal_loss + +__all__ = ['DETRLoss'] + + +@register +class DETRLoss(nn.Layer): + __shared__ = ['num_classes', 'use_focal_loss'] + __inject__ = ['matcher'] + + def __init__(self, + num_classes=80, + matcher='HungarianMatcher', + loss_coeff={ + 'class': 1, + 'bbox': 5, + 'giou': 2, + 'no_object': 0.1, + 'mask': 1, + 'dice': 1 + }, + aux_loss=True, + use_focal_loss=False): + r""" + Args: + num_classes (int): The number of classes. + matcher (HungarianMatcher): It computes an assignment between the targets + and the predictions of the network. + loss_coeff (dict): The coefficient of loss. + aux_loss (bool): If 'aux_loss = True', loss at each decoder layer are to be used. + use_focal_loss (bool): Use focal loss or not. + """ + super(DETRLoss, self).__init__() + self.num_classes = num_classes + + self.matcher = matcher + self.loss_coeff = loss_coeff + self.aux_loss = aux_loss + self.use_focal_loss = use_focal_loss + + if not self.use_focal_loss: + self.loss_coeff['class'] = paddle.full([num_classes + 1], + loss_coeff['class']) + self.loss_coeff['class'][-1] = loss_coeff['no_object'] + self.giou_loss = GIoULoss() + + def _get_loss_class(self, logits, gt_class, match_indices, bg_index, + num_gts): + # logits: [b, query, num_classes], gt_class: list[[n, 1]] + target_label = paddle.full(logits.shape[:2], bg_index, dtype='int64') + bs, num_query_objects = target_label.shape + if sum(len(a) for a in gt_class) > 0: + index, updates = self._get_index_updates(num_query_objects, + gt_class, match_indices) + target_label = paddle.scatter( + target_label.reshape([-1, 1]), index, updates.astype('int64')) + target_label = target_label.reshape([bs, num_query_objects]) + if self.use_focal_loss: + target_label = F.one_hot(target_label, + self.num_classes + 1)[:, :, :-1] + return { + 'loss_class': self.loss_coeff['class'] * sigmoid_focal_loss( + logits, target_label, num_gts / num_query_objects) + if self.use_focal_loss else F.cross_entropy( + logits, target_label, weight=self.loss_coeff['class']) + } + + def _get_loss_bbox(self, boxes, gt_bbox, match_indices, num_gts): + # boxes: [b, query, 4], gt_bbox: list[[n, 4]] + loss = dict() + if sum(len(a) for a in gt_bbox) == 0: + loss['loss_bbox'] = paddle.to_tensor([0.]) + loss['loss_giou'] = paddle.to_tensor([0.]) + return loss + + src_bbox, target_bbox = self._get_src_target_assign(boxes, gt_bbox, + match_indices) + loss['loss_bbox'] = self.loss_coeff['bbox'] * F.l1_loss( + src_bbox, target_bbox, reduction='sum') / num_gts + loss['loss_giou'] = self.giou_loss( + bbox_cxcywh_to_xyxy(src_bbox), bbox_cxcywh_to_xyxy(target_bbox)) + loss['loss_giou'] = loss['loss_giou'].sum() / num_gts + loss['loss_giou'] = self.loss_coeff['giou'] * loss['loss_giou'] + return loss + + def _get_loss_mask(self, masks, gt_mask, match_indices, num_gts): + # masks: [b, query, h, w], gt_mask: list[[n, H, W]] + loss = dict() + if sum(len(a) for a in gt_mask) == 0: + loss['loss_mask'] = paddle.to_tensor([0.]) + loss['loss_dice'] = paddle.to_tensor([0.]) + return loss + + src_masks, target_masks = self._get_src_target_assign(masks, gt_mask, + match_indices) + src_masks = F.interpolate( + src_masks.unsqueeze(0), + size=target_masks.shape[-2:], + mode="bilinear")[0] + loss['loss_mask'] = self.loss_coeff['mask'] * F.sigmoid_focal_loss( + src_masks, + target_masks, + paddle.to_tensor( + [num_gts], dtype='float32')) + loss['loss_dice'] = self.loss_coeff['dice'] * self._dice_loss( + src_masks, target_masks, num_gts) + return loss + + def _dice_loss(self, inputs, targets, num_gts): + inputs = F.sigmoid(inputs) + inputs = inputs.flatten(1) + targets = targets.flatten(1) + numerator = 2 * (inputs * targets).sum(1) + denominator = inputs.sum(-1) + targets.sum(-1) + loss = 1 - (numerator + 1) / (denominator + 1) + return loss.sum() / num_gts + + def _get_loss_aux(self, boxes, logits, gt_bbox, gt_class, bg_index, + num_gts): + loss_class = [] + loss_bbox = [] + loss_giou = [] + for aux_boxes, aux_logits in zip(boxes, logits): + match_indices = self.matcher(aux_boxes, aux_logits, gt_bbox, + gt_class) + loss_class.append( + self._get_loss_class(aux_logits, gt_class, match_indices, + bg_index, num_gts)['loss_class']) + loss_ = self._get_loss_bbox(aux_boxes, gt_bbox, match_indices, + num_gts) + loss_bbox.append(loss_['loss_bbox']) + loss_giou.append(loss_['loss_giou']) + loss = { + 'loss_class_aux': paddle.add_n(loss_class), + 'loss_bbox_aux': paddle.add_n(loss_bbox), + 'loss_giou_aux': paddle.add_n(loss_giou) + } + return loss + + def _get_index_updates(self, num_query_objects, target, match_indices): + batch_idx = paddle.concat([ + paddle.full_like(src, i) for i, (src, _) in enumerate(match_indices) + ]) + src_idx = paddle.concat([src for (src, _) in match_indices]) + src_idx += (batch_idx * num_query_objects) + target_assign = paddle.concat([ + paddle.gather( + t, dst, axis=0) for t, (_, dst) in zip(target, match_indices) + ]) + return src_idx, target_assign + + def _get_src_target_assign(self, src, target, match_indices): + src_assign = paddle.concat([ + paddle.gather( + t, I, axis=0) if len(I) > 0 else paddle.zeros([0, t.shape[-1]]) + for t, (I, _) in zip(src, match_indices) + ]) + target_assign = paddle.concat([ + paddle.gather( + t, J, axis=0) if len(J) > 0 else paddle.zeros([0, t.shape[-1]]) + for t, (_, J) in zip(target, match_indices) + ]) + return src_assign, target_assign + + def forward(self, + boxes, + logits, + gt_bbox, + gt_class, + masks=None, + gt_mask=None): + r""" + Args: + boxes (Tensor): [l, b, query, 4] + logits (Tensor): [l, b, query, num_classes] + gt_bbox (List(Tensor)): list[[n, 4]] + gt_class (List(Tensor)): list[[n, 1]] + masks (Tensor, optional): [b, query, h, w] + gt_mask (List(Tensor), optional): list[[n, H, W]] + """ + match_indices = self.matcher(boxes[-1].detach(), logits[-1].detach(), + gt_bbox, gt_class) + num_gts = sum(len(a) for a in gt_bbox) + try: + # TODO: Paddle does not have a "paddle.distributed.is_initialized()" + num_gts = paddle.to_tensor([num_gts], dtype=paddle.float32) + paddle.distributed.all_reduce(num_gts) + num_gts = paddle.clip( + num_gts / paddle.distributed.get_world_size(), min=1).item() + except: + num_gts = max(num_gts.item(), 1) + total_loss = dict() + total_loss.update( + self._get_loss_class(logits[-1], gt_class, match_indices, + self.num_classes, num_gts)) + total_loss.update( + self._get_loss_bbox(boxes[-1], gt_bbox, match_indices, num_gts)) + if masks is not None and gt_mask is not None: + total_loss.update( + self._get_loss_mask(masks, gt_mask, match_indices, num_gts)) + + if self.aux_loss: + total_loss.update( + self._get_loss_aux(boxes[:-1], logits[:-1], gt_bbox, gt_class, + self.num_classes, num_gts)) + + return total_loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fairmot_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fairmot_loss.py new file mode 100644 index 000000000..e24ff33fe --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fairmot_loss.py @@ -0,0 +1,41 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +from paddle.nn.initializer import Constant +from ppdet.core.workspace import register + +__all__ = ['FairMOTLoss'] + + +@register +class FairMOTLoss(nn.Layer): + def __init__(self): + super(FairMOTLoss, self).__init__() + self.det_weight = self.create_parameter( + shape=[1], default_initializer=Constant(-1.85)) + self.reid_weight = self.create_parameter( + shape=[1], default_initializer=Constant(-1.05)) + + def forward(self, det_loss, reid_loss): + loss = paddle.exp(-self.det_weight) * det_loss + paddle.exp( + -self.reid_weight) * reid_loss + (self.det_weight + self.reid_weight + ) + loss *= 0.5 + return {'loss': loss} diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fcos_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fcos_loss.py new file mode 100644 index 000000000..c8d600573 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/fcos_loss.py @@ -0,0 +1,225 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +from ppdet.modeling import ops + +__all__ = ['FCOSLoss'] + + +def flatten_tensor(inputs, channel_first=False): + """ + Flatten a Tensor + Args: + inputs (Tensor): 4-D Tensor with shape [N, C, H, W] or [N, H, W, C] + channel_first (bool): If true the dimension order of Tensor is + [N, C, H, W], otherwise is [N, H, W, C] + Return: + output_channel_last (Tensor): The flattened Tensor in channel_last style + """ + if channel_first: + input_channel_last = paddle.transpose(inputs, perm=[0, 2, 3, 1]) + else: + input_channel_last = inputs + output_channel_last = paddle.flatten( + input_channel_last, start_axis=0, stop_axis=2) + return output_channel_last + + +@register +class FCOSLoss(nn.Layer): + """ + FCOSLoss + Args: + loss_alpha (float): alpha in focal loss + loss_gamma (float): gamma in focal loss + iou_loss_type (str): location loss type, IoU/GIoU/LINEAR_IoU + reg_weights (float): weight for location loss + """ + + def __init__(self, + loss_alpha=0.25, + loss_gamma=2.0, + iou_loss_type="giou", + reg_weights=1.0): + super(FCOSLoss, self).__init__() + self.loss_alpha = loss_alpha + self.loss_gamma = loss_gamma + self.iou_loss_type = iou_loss_type + self.reg_weights = reg_weights + + def __iou_loss(self, pred, targets, positive_mask, weights=None): + """ + Calculate the loss for location prediction + Args: + pred (Tensor): bounding boxes prediction + targets (Tensor): targets for positive samples + positive_mask (Tensor): mask of positive samples + weights (Tensor): weights for each positive samples + Return: + loss (Tensor): location loss + """ + plw = pred[:, 0] * positive_mask + pth = pred[:, 1] * positive_mask + prw = pred[:, 2] * positive_mask + pbh = pred[:, 3] * positive_mask + + tlw = targets[:, 0] * positive_mask + tth = targets[:, 1] * positive_mask + trw = targets[:, 2] * positive_mask + tbh = targets[:, 3] * positive_mask + tlw.stop_gradient = True + trw.stop_gradient = True + tth.stop_gradient = True + tbh.stop_gradient = True + + ilw = paddle.minimum(plw, tlw) + irw = paddle.minimum(prw, trw) + ith = paddle.minimum(pth, tth) + ibh = paddle.minimum(pbh, tbh) + + clw = paddle.maximum(plw, tlw) + crw = paddle.maximum(prw, trw) + cth = paddle.maximum(pth, tth) + cbh = paddle.maximum(pbh, tbh) + + area_predict = (plw + prw) * (pth + pbh) + area_target = (tlw + trw) * (tth + tbh) + area_inter = (ilw + irw) * (ith + ibh) + ious = (area_inter + 1.0) / ( + area_predict + area_target - area_inter + 1.0) + ious = ious * positive_mask + + if self.iou_loss_type.lower() == "linear_iou": + loss = 1.0 - ious + elif self.iou_loss_type.lower() == "giou": + area_uniou = area_predict + area_target - area_inter + area_circum = (clw + crw) * (cth + cbh) + 1e-7 + giou = ious - (area_circum - area_uniou) / area_circum + loss = 1.0 - giou + elif self.iou_loss_type.lower() == "iou": + loss = 0.0 - paddle.log(ious) + else: + raise KeyError + if weights is not None: + loss = loss * weights + return loss + + def forward(self, cls_logits, bboxes_reg, centerness, tag_labels, + tag_bboxes, tag_center): + """ + Calculate the loss for classification, location and centerness + Args: + cls_logits (list): list of Tensor, which is predicted + score for all anchor points with shape [N, M, C] + bboxes_reg (list): list of Tensor, which is predicted + offsets for all anchor points with shape [N, M, 4] + centerness (list): list of Tensor, which is predicted + centerness for all anchor points with shape [N, M, 1] + tag_labels (list): list of Tensor, which is category + targets for each anchor point + tag_bboxes (list): list of Tensor, which is bounding + boxes targets for positive samples + tag_center (list): list of Tensor, which is centerness + targets for positive samples + Return: + loss (dict): loss composed by classification loss, bounding box + """ + cls_logits_flatten_list = [] + bboxes_reg_flatten_list = [] + centerness_flatten_list = [] + tag_labels_flatten_list = [] + tag_bboxes_flatten_list = [] + tag_center_flatten_list = [] + num_lvl = len(cls_logits) + for lvl in range(num_lvl): + cls_logits_flatten_list.append( + flatten_tensor(cls_logits[lvl], True)) + bboxes_reg_flatten_list.append( + flatten_tensor(bboxes_reg[lvl], True)) + centerness_flatten_list.append( + flatten_tensor(centerness[lvl], True)) + + tag_labels_flatten_list.append( + flatten_tensor(tag_labels[lvl], False)) + tag_bboxes_flatten_list.append( + flatten_tensor(tag_bboxes[lvl], False)) + tag_center_flatten_list.append( + flatten_tensor(tag_center[lvl], False)) + + cls_logits_flatten = paddle.concat(cls_logits_flatten_list, axis=0) + bboxes_reg_flatten = paddle.concat(bboxes_reg_flatten_list, axis=0) + centerness_flatten = paddle.concat(centerness_flatten_list, axis=0) + + tag_labels_flatten = paddle.concat(tag_labels_flatten_list, axis=0) + tag_bboxes_flatten = paddle.concat(tag_bboxes_flatten_list, axis=0) + tag_center_flatten = paddle.concat(tag_center_flatten_list, axis=0) + tag_labels_flatten.stop_gradient = True + tag_bboxes_flatten.stop_gradient = True + tag_center_flatten.stop_gradient = True + + mask_positive_bool = tag_labels_flatten > 0 + mask_positive_bool.stop_gradient = True + mask_positive_float = paddle.cast(mask_positive_bool, dtype="float32") + mask_positive_float.stop_gradient = True + + num_positive_fp32 = paddle.sum(mask_positive_float) + num_positive_fp32.stop_gradient = True + num_positive_int32 = paddle.cast(num_positive_fp32, dtype="int32") + num_positive_int32 = num_positive_int32 * 0 + 1 + num_positive_int32.stop_gradient = True + + normalize_sum = paddle.sum(tag_center_flatten * mask_positive_float) + normalize_sum.stop_gradient = True + + # 1. cls_logits: sigmoid_focal_loss + # expand onehot labels + num_classes = cls_logits_flatten.shape[-1] + tag_labels_flatten = paddle.squeeze(tag_labels_flatten, axis=-1) + tag_labels_flatten_bin = F.one_hot( + tag_labels_flatten, num_classes=1 + num_classes) + tag_labels_flatten_bin = tag_labels_flatten_bin[:, 1:] + # sigmoid_focal_loss + cls_loss = F.sigmoid_focal_loss( + cls_logits_flatten, tag_labels_flatten_bin) / num_positive_fp32 + + # 2. bboxes_reg: giou_loss + mask_positive_float = paddle.squeeze(mask_positive_float, axis=-1) + tag_center_flatten = paddle.squeeze(tag_center_flatten, axis=-1) + reg_loss = self.__iou_loss( + bboxes_reg_flatten, + tag_bboxes_flatten, + mask_positive_float, + weights=tag_center_flatten) + reg_loss = reg_loss * mask_positive_float / normalize_sum + + # 3. centerness: sigmoid_cross_entropy_with_logits_loss + centerness_flatten = paddle.squeeze(centerness_flatten, axis=-1) + ctn_loss = ops.sigmoid_cross_entropy_with_logits(centerness_flatten, + tag_center_flatten) + ctn_loss = ctn_loss * mask_positive_float / num_positive_fp32 + + loss_all = { + "loss_centerness": paddle.sum(ctn_loss), + "loss_cls": paddle.sum(cls_loss), + "loss_box": paddle.sum(reg_loss) + } + return loss_all diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/gfocal_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/gfocal_loss.py new file mode 100644 index 000000000..37e27f084 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/gfocal_loss.py @@ -0,0 +1,217 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/losses/gfocal_loss.py + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ppdet.modeling import ops + +__all__ = ['QualityFocalLoss', 'DistributionFocalLoss'] + + +def quality_focal_loss(pred, target, beta=2.0, use_sigmoid=True): + """ + Quality Focal Loss (QFL) is from `Generalized Focal Loss: Learning + Qualified and Distributed Bounding Boxes for Dense Object Detection + `_. + Args: + pred (Tensor): Predicted joint representation of classification + and quality (IoU) estimation with shape (N, C), C is the number of + classes. + target (tuple([Tensor])): Target category label with shape (N,) + and target quality label with shape (N,). + beta (float): The beta parameter for calculating the modulating factor. + Defaults to 2.0. + Returns: + Tensor: Loss tensor with shape (N,). + """ + assert len(target) == 2, """target for QFL must be a tuple of two elements, + including category label and quality label, respectively""" + # label denotes the category id, score denotes the quality score + label, score = target + if use_sigmoid: + func = F.binary_cross_entropy_with_logits + else: + func = F.binary_cross_entropy + + # negatives are supervised by 0 quality score + pred_sigmoid = F.sigmoid(pred) if use_sigmoid else pred + scale_factor = pred_sigmoid + zerolabel = paddle.zeros(pred.shape, dtype='float32') + loss = func(pred, zerolabel, reduction='none') * scale_factor.pow(beta) + + # FG cat_id: [0, num_classes -1], BG cat_id: num_classes + bg_class_ind = pred.shape[1] + pos = paddle.logical_and((label >= 0), + (label < bg_class_ind)).nonzero().squeeze(1) + if pos.shape[0] == 0: + return loss.sum(axis=1) + pos_label = paddle.gather(label, pos, axis=0) + pos_mask = np.zeros(pred.shape, dtype=np.int32) + pos_mask[pos.numpy(), pos_label.numpy()] = 1 + pos_mask = paddle.to_tensor(pos_mask, dtype='bool') + score = score.unsqueeze(-1).expand([-1, pred.shape[1]]).cast('float32') + # positives are supervised by bbox quality (IoU) score + scale_factor_new = score - pred_sigmoid + + loss_pos = func( + pred, score, reduction='none') * scale_factor_new.abs().pow(beta) + loss = loss * paddle.logical_not(pos_mask) + loss_pos * pos_mask + loss = loss.sum(axis=1) + return loss + + +def distribution_focal_loss(pred, label): + """Distribution Focal Loss (DFL) is from `Generalized Focal Loss: Learning + Qualified and Distributed Bounding Boxes for Dense Object Detection + `_. + Args: + pred (Tensor): Predicted general distribution of bounding boxes + (before softmax) with shape (N, n+1), n is the max value of the + integral set `{0, ..., n}` in paper. + label (Tensor): Target distance label for bounding boxes with + shape (N,). + Returns: + Tensor: Loss tensor with shape (N,). + """ + dis_left = label.cast('int64') + dis_right = dis_left + 1 + weight_left = dis_right.cast('float32') - label + weight_right = label - dis_left.cast('float32') + loss = F.cross_entropy(pred, dis_left, reduction='none') * weight_left \ + + F.cross_entropy(pred, dis_right, reduction='none') * weight_right + return loss + + +@register +@serializable +class QualityFocalLoss(nn.Layer): + r"""Quality Focal Loss (QFL) is a variant of `Generalized Focal Loss: + Learning Qualified and Distributed Bounding Boxes for Dense Object + Detection `_. + Args: + use_sigmoid (bool): Whether sigmoid operation is conducted in QFL. + Defaults to True. + beta (float): The beta parameter for calculating the modulating factor. + Defaults to 2.0. + reduction (str): Options are "none", "mean" and "sum". + loss_weight (float): Loss weight of current loss. + """ + + def __init__(self, + use_sigmoid=True, + beta=2.0, + reduction='mean', + loss_weight=1.0): + super(QualityFocalLoss, self).__init__() + self.use_sigmoid = use_sigmoid + self.beta = beta + assert reduction in ('none', 'mean', 'sum') + self.reduction = reduction + self.loss_weight = loss_weight + + def forward(self, pred, target, weight=None, avg_factor=None): + """Forward function. + Args: + pred (Tensor): Predicted joint representation of + classification and quality (IoU) estimation with shape (N, C), + C is the number of classes. + target (tuple([Tensor])): Target category label with shape + (N,) and target quality label with shape (N,). + weight (Tensor, optional): The weight of loss for each + prediction. Defaults to None. + avg_factor (int, optional): Average factor that is used to average + the loss. Defaults to None. + """ + + loss = self.loss_weight * quality_focal_loss( + pred, target, beta=self.beta, use_sigmoid=self.use_sigmoid) + + if weight is not None: + loss = loss * weight + if avg_factor is None: + if self.reduction == 'none': + return loss + elif self.reduction == 'mean': + return loss.mean() + elif self.reduction == 'sum': + return loss.sum() + else: + # if reduction is mean, then average the loss by avg_factor + if self.reduction == 'mean': + loss = loss.sum() / avg_factor + # if reduction is 'none', then do nothing, otherwise raise an error + elif self.reduction != 'none': + raise ValueError( + 'avg_factor can not be used with reduction="sum"') + return loss + + +@register +@serializable +class DistributionFocalLoss(nn.Layer): + """Distribution Focal Loss (DFL) is a variant of `Generalized Focal Loss: + Learning Qualified and Distributed Bounding Boxes for Dense Object + Detection `_. + Args: + reduction (str): Options are `'none'`, `'mean'` and `'sum'`. + loss_weight (float): Loss weight of current loss. + """ + + def __init__(self, reduction='mean', loss_weight=1.0): + super(DistributionFocalLoss, self).__init__() + assert reduction in ('none', 'mean', 'sum') + self.reduction = reduction + self.loss_weight = loss_weight + + def forward(self, pred, target, weight=None, avg_factor=None): + """Forward function. + Args: + pred (Tensor): Predicted general distribution of bounding + boxes (before softmax) with shape (N, n+1), n is the max value + of the integral set `{0, ..., n}` in paper. + target (Tensor): Target distance label for bounding boxes + with shape (N,). + weight (Tensor, optional): The weight of loss for each + prediction. Defaults to None. + avg_factor (int, optional): Average factor that is used to average + the loss. Defaults to None. + """ + loss = self.loss_weight * distribution_focal_loss(pred, target) + if weight is not None: + loss = loss * weight + if avg_factor is None: + if self.reduction == 'none': + return loss + elif self.reduction == 'mean': + return loss.mean() + elif self.reduction == 'sum': + return loss.sum() + else: + # if reduction is mean, then average the loss by avg_factor + if self.reduction == 'mean': + loss = loss.sum() / avg_factor + # if reduction is 'none', then do nothing, otherwise raise an error + elif self.reduction != 'none': + raise ValueError( + 'avg_factor can not be used with reduction="sum"') + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_aware_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_aware_loss.py new file mode 100644 index 000000000..4a9e904dd --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_aware_loss.py @@ -0,0 +1,47 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from .iou_loss import IouLoss +from ..bbox_utils import bbox_iou + + +@register +@serializable +class IouAwareLoss(IouLoss): + """ + iou aware loss, see https://arxiv.org/abs/1912.05992 + Args: + loss_weight (float): iou aware loss weight, default is 1.0 + max_height (int): max height of input to support random shape input + max_width (int): max width of input to support random shape input + """ + + def __init__(self, loss_weight=1.0, giou=False, diou=False, ciou=False): + super(IouAwareLoss, self).__init__( + loss_weight=loss_weight, giou=giou, diou=diou, ciou=ciou) + + def __call__(self, ioup, pbox, gbox): + iou = bbox_iou( + pbox, gbox, giou=self.giou, diou=self.diou, ciou=self.ciou) + iou.stop_gradient = True + loss_iou_aware = F.binary_cross_entropy_with_logits( + ioup, iou, reduction='none') + loss_iou_aware = loss_iou_aware * self.loss_weight + return loss_iou_aware diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_loss.py new file mode 100644 index 000000000..9b8da6c05 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/iou_loss.py @@ -0,0 +1,210 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import numpy as np + +import paddle + +from ppdet.core.workspace import register, serializable +from ..bbox_utils import bbox_iou + +__all__ = ['IouLoss', 'GIoULoss', 'DIouLoss'] + + +@register +@serializable +class IouLoss(object): + """ + iou loss, see https://arxiv.org/abs/1908.03851 + loss = 1.0 - iou * iou + Args: + loss_weight (float): iou loss weight, default is 2.5 + max_height (int): max height of input to support random shape input + max_width (int): max width of input to support random shape input + ciou_term (bool): whether to add ciou_term + loss_square (bool): whether to square the iou term + """ + + def __init__(self, + loss_weight=2.5, + giou=False, + diou=False, + ciou=False, + loss_square=True): + self.loss_weight = loss_weight + self.giou = giou + self.diou = diou + self.ciou = ciou + self.loss_square = loss_square + + def __call__(self, pbox, gbox): + iou = bbox_iou( + pbox, gbox, giou=self.giou, diou=self.diou, ciou=self.ciou) + if self.loss_square: + loss_iou = 1 - iou * iou + else: + loss_iou = 1 - iou + + loss_iou = loss_iou * self.loss_weight + return loss_iou + + +@register +@serializable +class GIoULoss(object): + """ + Generalized Intersection over Union, see https://arxiv.org/abs/1902.09630 + Args: + loss_weight (float): giou loss weight, default as 1 + eps (float): epsilon to avoid divide by zero, default as 1e-10 + reduction (string): Options are "none", "mean" and "sum". default as none + """ + + def __init__(self, loss_weight=1., eps=1e-10, reduction='none'): + self.loss_weight = loss_weight + self.eps = eps + assert reduction in ('none', 'mean', 'sum') + self.reduction = reduction + + def bbox_overlap(self, box1, box2, eps=1e-10): + """calculate the iou of box1 and box2 + Args: + box1 (Tensor): box1 with the shape (..., 4) + box2 (Tensor): box1 with the shape (..., 4) + eps (float): epsilon to avoid divide by zero + Return: + iou (Tensor): iou of box1 and box2 + overlap (Tensor): overlap of box1 and box2 + union (Tensor): union of box1 and box2 + """ + x1, y1, x2, y2 = box1 + x1g, y1g, x2g, y2g = box2 + + xkis1 = paddle.maximum(x1, x1g) + ykis1 = paddle.maximum(y1, y1g) + xkis2 = paddle.minimum(x2, x2g) + ykis2 = paddle.minimum(y2, y2g) + w_inter = (xkis2 - xkis1).clip(0) + h_inter = (ykis2 - ykis1).clip(0) + overlap = w_inter * h_inter + + area1 = (x2 - x1) * (y2 - y1) + area2 = (x2g - x1g) * (y2g - y1g) + union = area1 + area2 - overlap + eps + iou = overlap / union + + return iou, overlap, union + + def __call__(self, pbox, gbox, iou_weight=1., loc_reweight=None): + x1, y1, x2, y2 = paddle.split(pbox, num_or_sections=4, axis=-1) + x1g, y1g, x2g, y2g = paddle.split(gbox, num_or_sections=4, axis=-1) + box1 = [x1, y1, x2, y2] + box2 = [x1g, y1g, x2g, y2g] + iou, overlap, union = self.bbox_overlap(box1, box2, self.eps) + xc1 = paddle.minimum(x1, x1g) + yc1 = paddle.minimum(y1, y1g) + xc2 = paddle.maximum(x2, x2g) + yc2 = paddle.maximum(y2, y2g) + + area_c = (xc2 - xc1) * (yc2 - yc1) + self.eps + miou = iou - ((area_c - union) / area_c) + if loc_reweight is not None: + loc_reweight = paddle.reshape(loc_reweight, shape=(-1, 1)) + loc_thresh = 0.9 + giou = 1 - (1 - loc_thresh + ) * miou - loc_thresh * miou * loc_reweight + else: + giou = 1 - miou + if self.reduction == 'none': + loss = giou + elif self.reduction == 'sum': + loss = paddle.sum(giou * iou_weight) + else: + loss = paddle.mean(giou * iou_weight) + return loss * self.loss_weight + + +@register +@serializable +class DIouLoss(GIoULoss): + """ + Distance-IoU Loss, see https://arxiv.org/abs/1911.08287 + Args: + loss_weight (float): giou loss weight, default as 1 + eps (float): epsilon to avoid divide by zero, default as 1e-10 + use_complete_iou_loss (bool): whether to use complete iou loss + """ + + def __init__(self, loss_weight=1., eps=1e-10, use_complete_iou_loss=True): + super(DIouLoss, self).__init__(loss_weight=loss_weight, eps=eps) + self.use_complete_iou_loss = use_complete_iou_loss + + def __call__(self, pbox, gbox, iou_weight=1.): + x1, y1, x2, y2 = paddle.split(pbox, num_or_sections=4, axis=-1) + x1g, y1g, x2g, y2g = paddle.split(gbox, num_or_sections=4, axis=-1) + cx = (x1 + x2) / 2 + cy = (y1 + y2) / 2 + w = x2 - x1 + h = y2 - y1 + + cxg = (x1g + x2g) / 2 + cyg = (y1g + y2g) / 2 + wg = x2g - x1g + hg = y2g - y1g + + x2 = paddle.maximum(x1, x2) + y2 = paddle.maximum(y1, y2) + + # A and B + xkis1 = paddle.maximum(x1, x1g) + ykis1 = paddle.maximum(y1, y1g) + xkis2 = paddle.minimum(x2, x2g) + ykis2 = paddle.minimum(y2, y2g) + + # A or B + xc1 = paddle.minimum(x1, x1g) + yc1 = paddle.minimum(y1, y1g) + xc2 = paddle.maximum(x2, x2g) + yc2 = paddle.maximum(y2, y2g) + + intsctk = (xkis2 - xkis1) * (ykis2 - ykis1) + intsctk = intsctk * paddle.greater_than( + xkis2, xkis1) * paddle.greater_than(ykis2, ykis1) + unionk = (x2 - x1) * (y2 - y1) + (x2g - x1g) * (y2g - y1g + ) - intsctk + self.eps + iouk = intsctk / unionk + + # DIOU term + dist_intersection = (cx - cxg) * (cx - cxg) + (cy - cyg) * (cy - cyg) + dist_union = (xc2 - xc1) * (xc2 - xc1) + (yc2 - yc1) * (yc2 - yc1) + diou_term = (dist_intersection + self.eps) / (dist_union + self.eps) + + # CIOU term + ciou_term = 0 + if self.use_complete_iou_loss: + ar_gt = wg / hg + ar_pred = w / h + arctan = paddle.atan(ar_gt) - paddle.atan(ar_pred) + ar_loss = 4. / np.pi / np.pi * arctan * arctan + alpha = ar_loss / (1 - iouk + ar_loss + self.eps) + alpha.stop_gradient = True + ciou_term = alpha * ar_loss + + diou = paddle.mean((1 - iouk + ciou_term + diou_term) * iou_weight) + + return diou * self.loss_weight diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/jde_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/jde_loss.py new file mode 100644 index 000000000..5c3b5a615 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/jde_loss.py @@ -0,0 +1,193 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register + +__all__ = ['JDEDetectionLoss', 'JDEEmbeddingLoss', 'JDELoss'] + + +@register +class JDEDetectionLoss(nn.Layer): + __shared__ = ['num_classes'] + + def __init__(self, num_classes=1, for_mot=True): + super(JDEDetectionLoss, self).__init__() + self.num_classes = num_classes + self.for_mot = for_mot + + def det_loss(self, p_det, anchor, t_conf, t_box): + pshape = paddle.shape(p_det) + pshape.stop_gradient = True + nB, nGh, nGw = pshape[0], pshape[-2], pshape[-1] + nA = len(anchor) + p_det = paddle.reshape( + p_det, [nB, nA, self.num_classes + 5, nGh, nGw]).transpose( + (0, 1, 3, 4, 2)) + + # 1. loss_conf: cross_entropy + p_conf = p_det[:, :, :, :, 4:6] + p_conf_flatten = paddle.reshape(p_conf, [-1, 2]) + t_conf_flatten = t_conf.flatten() + t_conf_flatten = paddle.cast(t_conf_flatten, dtype="int64") + t_conf_flatten.stop_gradient = True + loss_conf = F.cross_entropy( + p_conf_flatten, t_conf_flatten, ignore_index=-1, reduction='mean') + loss_conf.stop_gradient = False + + # 2. loss_box: smooth_l1_loss + p_box = p_det[:, :, :, :, :4] + p_box_flatten = paddle.reshape(p_box, [-1, 4]) + t_box_flatten = paddle.reshape(t_box, [-1, 4]) + fg_inds = paddle.nonzero(t_conf_flatten > 0).flatten() + if fg_inds.numel() > 0: + reg_delta = paddle.gather(p_box_flatten, fg_inds) + reg_target = paddle.gather(t_box_flatten, fg_inds) + else: + reg_delta = paddle.to_tensor([0, 0, 0, 0], dtype='float32') + reg_delta.stop_gradient = False + reg_target = paddle.to_tensor([0, 0, 0, 0], dtype='float32') + reg_target.stop_gradient = True + loss_box = F.smooth_l1_loss( + reg_delta, reg_target, reduction='mean', delta=1.0) + loss_box.stop_gradient = False + + return loss_conf, loss_box + + def forward(self, det_outs, targets, anchors): + """ + Args: + det_outs (list[Tensor]): output from detection head, each one + is a 4-D Tensor with shape [N, C, H, W]. + targets (dict): contains 'im_id', 'gt_bbox', 'gt_ide', 'image', + 'im_shape', 'scale_factor' and 'tbox', 'tconf', 'tide' of + each FPN level. + anchors (list[list]): anchor setting of JDE model, N row M col, N is + the anchor levels(FPN levels), M is the anchor scales each + level. + """ + assert len(det_outs) == len(anchors) + loss_confs = [] + loss_boxes = [] + for i, (p_det, anchor) in enumerate(zip(det_outs, anchors)): + t_conf = targets['tconf{}'.format(i)] + t_box = targets['tbox{}'.format(i)] + + loss_conf, loss_box = self.det_loss(p_det, anchor, t_conf, t_box) + loss_confs.append(loss_conf) + loss_boxes.append(loss_box) + if self.for_mot: + return {'loss_confs': loss_confs, 'loss_boxes': loss_boxes} + else: + jde_conf_losses = sum(loss_confs) + jde_box_losses = sum(loss_boxes) + jde_det_losses = { + "loss_conf": jde_conf_losses, + "loss_box": jde_box_losses, + "loss": jde_conf_losses + jde_box_losses, + } + return jde_det_losses + + +@register +class JDEEmbeddingLoss(nn.Layer): + def __init__(self, ): + super(JDEEmbeddingLoss, self).__init__() + self.phony = self.create_parameter(shape=[1], dtype="float32") + + def emb_loss(self, p_ide, t_conf, t_ide, emb_scale, classifier): + emb_dim = p_ide.shape[1] + p_ide = p_ide.transpose((0, 2, 3, 1)) + p_ide_flatten = paddle.reshape(p_ide, [-1, emb_dim]) + mask = t_conf > 0 + mask = paddle.cast(mask, dtype="int64") + mask.stop_gradient = True + emb_mask = mask.max(1).flatten() + emb_mask_inds = paddle.nonzero(emb_mask > 0).flatten() + emb_mask_inds.stop_gradient = True + # use max(1) to decide the id, TODO: more reseanable strategy + t_ide_flatten = t_ide.max(1).flatten() + t_ide_flatten = paddle.cast(t_ide_flatten, dtype="int64") + valid_inds = paddle.nonzero(t_ide_flatten != -1).flatten() + + if emb_mask_inds.numel() == 0 or valid_inds.numel() == 0: + # loss_ide = paddle.to_tensor([0]) # will be error in gradient backward + loss_ide = self.phony * 0 # todo + else: + embedding = paddle.gather(p_ide_flatten, emb_mask_inds) + embedding = emb_scale * F.normalize(embedding) + logits = classifier(embedding) + + ide_target = paddle.gather(t_ide_flatten, emb_mask_inds) + + loss_ide = F.cross_entropy( + logits, ide_target, ignore_index=-1, reduction='mean') + loss_ide.stop_gradient = False + + return loss_ide + + def forward(self, ide_outs, targets, emb_scale, classifier): + loss_ides = [] + for i, p_ide in enumerate(ide_outs): + t_conf = targets['tconf{}'.format(i)] + t_ide = targets['tide{}'.format(i)] + + loss_ide = self.emb_loss(p_ide, t_conf, t_ide, emb_scale, + classifier) + loss_ides.append(loss_ide) + return loss_ides + + +@register +class JDELoss(nn.Layer): + def __init__(self): + super(JDELoss, self).__init__() + + def forward(self, loss_confs, loss_boxes, loss_ides, loss_params_cls, + loss_params_reg, loss_params_ide, targets): + assert len(loss_confs) == len(loss_boxes) == len(loss_ides) + assert len(loss_params_cls) == len(loss_params_reg) == len( + loss_params_ide) + assert len(loss_confs) == len(loss_params_cls) + + batchsize = targets['gt_bbox'].shape[0] + nTargets = paddle.nonzero(paddle.sum(targets['gt_bbox'], axis=2)).shape[ + 0] / batchsize + nTargets = paddle.to_tensor(nTargets, dtype='float32') + nTargets.stop_gradient = True + + jde_losses = [] + for i, (loss_conf, loss_box, loss_ide, l_conf_p, l_box_p, + l_ide_p) in enumerate( + zip(loss_confs, loss_boxes, loss_ides, loss_params_cls, + loss_params_reg, loss_params_ide)): + + jde_loss = l_conf_p(loss_conf) + l_box_p(loss_box) + l_ide_p( + loss_ide) + jde_losses.append(jde_loss) + + loss_all = { + "loss_conf": sum(loss_confs), + "loss_box": sum(loss_boxes), + "loss_ide": sum(loss_ides), + "loss": sum(jde_losses), + "nTargets": nTargets, + } + return loss_all diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/keypoint_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/keypoint_loss.py new file mode 100644 index 000000000..9c3c113db --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/keypoint_loss.py @@ -0,0 +1,228 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from itertools import cycle, islice +from collections import abc +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register, serializable + +__all__ = ['HrHRNetLoss', 'KeyPointMSELoss'] + + +@register +@serializable +class KeyPointMSELoss(nn.Layer): + def __init__(self, use_target_weight=True, loss_scale=0.5): + """ + KeyPointMSELoss layer + + Args: + use_target_weight (bool): whether to use target weight + """ + super(KeyPointMSELoss, self).__init__() + self.criterion = nn.MSELoss(reduction='mean') + self.use_target_weight = use_target_weight + self.loss_scale = loss_scale + + def forward(self, output, records): + target = records['target'] + target_weight = records['target_weight'] + batch_size = output.shape[0] + num_joints = output.shape[1] + heatmaps_pred = output.reshape( + (batch_size, num_joints, -1)).split(num_joints, 1) + heatmaps_gt = target.reshape( + (batch_size, num_joints, -1)).split(num_joints, 1) + loss = 0 + for idx in range(num_joints): + heatmap_pred = heatmaps_pred[idx].squeeze() + heatmap_gt = heatmaps_gt[idx].squeeze() + if self.use_target_weight: + loss += self.loss_scale * self.criterion( + heatmap_pred.multiply(target_weight[:, idx]), + heatmap_gt.multiply(target_weight[:, idx])) + else: + loss += self.loss_scale * self.criterion(heatmap_pred, + heatmap_gt) + keypoint_losses = dict() + keypoint_losses['loss'] = loss / num_joints + return keypoint_losses + + +@register +@serializable +class HrHRNetLoss(nn.Layer): + def __init__(self, num_joints, swahr): + """ + HrHRNetLoss layer + + Args: + num_joints (int): number of keypoints + """ + super(HrHRNetLoss, self).__init__() + if swahr: + self.heatmaploss = HeatMapSWAHRLoss(num_joints) + else: + self.heatmaploss = HeatMapLoss() + self.aeloss = AELoss() + self.ziploss = ZipLoss( + [self.heatmaploss, self.heatmaploss, self.aeloss]) + + def forward(self, inputs, records): + targets = [] + targets.append([records['heatmap_gt1x'], records['mask_1x']]) + targets.append([records['heatmap_gt2x'], records['mask_2x']]) + targets.append(records['tagmap']) + keypoint_losses = dict() + loss = self.ziploss(inputs, targets) + keypoint_losses['heatmap_loss'] = loss[0] + loss[1] + keypoint_losses['pull_loss'] = loss[2][0] + keypoint_losses['push_loss'] = loss[2][1] + keypoint_losses['loss'] = recursive_sum(loss) + return keypoint_losses + + +class HeatMapLoss(object): + def __init__(self, loss_factor=1.0): + super(HeatMapLoss, self).__init__() + self.loss_factor = loss_factor + + def __call__(self, preds, targets): + heatmap, mask = targets + loss = ((preds - heatmap)**2 * mask.cast('float').unsqueeze(1)) + loss = paddle.clip(loss, min=0, max=2).mean() + loss *= self.loss_factor + return loss + + +class HeatMapSWAHRLoss(object): + def __init__(self, num_joints, loss_factor=1.0): + super(HeatMapSWAHRLoss, self).__init__() + self.loss_factor = loss_factor + self.num_joints = num_joints + + def __call__(self, preds, targets): + heatmaps_gt, mask = targets + heatmaps_pred = preds[0] + scalemaps_pred = preds[1] + + heatmaps_scaled_gt = paddle.where(heatmaps_gt > 0, 0.5 * heatmaps_gt * ( + 1 + (1 + + (scalemaps_pred - 1.) * paddle.log(heatmaps_gt + 1e-10))**2), + heatmaps_gt) + + regularizer_loss = paddle.mean( + paddle.pow((scalemaps_pred - 1.) * (heatmaps_gt > 0).astype(float), + 2)) + omiga = 0.01 + # thres = 2**(-1/omiga), threshold for positive weight + hm_weight = heatmaps_scaled_gt**( + omiga + ) * paddle.abs(1 - heatmaps_pred) + paddle.abs(heatmaps_pred) * ( + 1 - heatmaps_scaled_gt**(omiga)) + + loss = (((heatmaps_pred - heatmaps_scaled_gt)**2) * + mask.cast('float').unsqueeze(1)) * hm_weight + loss = loss.mean() + loss = self.loss_factor * (loss + 1.0 * regularizer_loss) + return loss + + +class AELoss(object): + def __init__(self, pull_factor=0.001, push_factor=0.001): + super(AELoss, self).__init__() + self.pull_factor = pull_factor + self.push_factor = push_factor + + def apply_single(self, pred, tagmap): + if tagmap.numpy()[:, :, 3].sum() == 0: + return (paddle.zeros([1]), paddle.zeros([1])) + nonzero = paddle.nonzero(tagmap[:, :, 3] > 0) + if nonzero.shape[0] == 0: + return (paddle.zeros([1]), paddle.zeros([1])) + p_inds = paddle.unique(nonzero[:, 0]) + num_person = p_inds.shape[0] + if num_person == 0: + return (paddle.zeros([1]), paddle.zeros([1])) + + pull = 0 + tagpull_num = 0 + embs_all = [] + person_unvalid = 0 + for person_idx in p_inds.numpy(): + valid_single = tagmap[person_idx.item()] + validkpts = paddle.nonzero(valid_single[:, 3] > 0) + valid_single = paddle.index_select(valid_single, validkpts) + emb = paddle.gather_nd(pred, valid_single[:, :3]) + if emb.shape[0] == 1: + person_unvalid += 1 + mean = paddle.mean(emb, axis=0) + embs_all.append(mean) + pull += paddle.mean(paddle.pow(emb - mean, 2), axis=0) + tagpull_num += emb.shape[0] + pull /= max(num_person - person_unvalid, 1) + if num_person < 2: + return pull, paddle.zeros([1]) + + embs_all = paddle.stack(embs_all) + A = embs_all.expand([num_person, num_person]) + B = A.transpose([1, 0]) + diff = A - B + + diff = paddle.pow(diff, 2) + push = paddle.exp(-diff) + push = paddle.sum(push) - num_person + + push /= 2 * num_person * (num_person - 1) + return pull, push + + def __call__(self, preds, tagmaps): + bs = preds.shape[0] + losses = [ + self.apply_single(preds[i:i + 1].squeeze(), + tagmaps[i:i + 1].squeeze()) for i in range(bs) + ] + pull = self.pull_factor * sum(loss[0] for loss in losses) / len(losses) + push = self.push_factor * sum(loss[1] for loss in losses) / len(losses) + return pull, push + + +class ZipLoss(object): + def __init__(self, loss_funcs): + super(ZipLoss, self).__init__() + self.loss_funcs = loss_funcs + + def __call__(self, inputs, targets): + assert len(self.loss_funcs) == len(targets) >= len(inputs) + + def zip_repeat(*args): + longest = max(map(len, args)) + filled = [islice(cycle(x), longest) for x in args] + return zip(*filled) + + return tuple( + fn(x, y) + for x, y, fn in zip_repeat(inputs, targets, self.loss_funcs)) + + +def recursive_sum(inputs): + if isinstance(inputs, abc.Sequence): + return sum([recursive_sum(x) for x in inputs]) + return inputs diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/solov2_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/solov2_loss.py new file mode 100644 index 000000000..ef97a7707 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/solov2_loss.py @@ -0,0 +1,101 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable + +__all__ = ['SOLOv2Loss'] + + +@register +@serializable +class SOLOv2Loss(object): + """ + SOLOv2Loss + Args: + ins_loss_weight (float): Weight of instance loss. + focal_loss_gamma (float): Gamma parameter for focal loss. + focal_loss_alpha (float): Alpha parameter for focal loss. + """ + + def __init__(self, + ins_loss_weight=3.0, + focal_loss_gamma=2.0, + focal_loss_alpha=0.25): + self.ins_loss_weight = ins_loss_weight + self.focal_loss_gamma = focal_loss_gamma + self.focal_loss_alpha = focal_loss_alpha + + def _dice_loss(self, input, target): + input = paddle.reshape(input, shape=(paddle.shape(input)[0], -1)) + target = paddle.reshape(target, shape=(paddle.shape(target)[0], -1)) + a = paddle.sum(input * target, axis=1) + b = paddle.sum(input * input, axis=1) + 0.001 + c = paddle.sum(target * target, axis=1) + 0.001 + d = (2 * a) / (b + c) + return 1 - d + + def __call__(self, ins_pred_list, ins_label_list, cate_preds, cate_labels, + num_ins): + """ + Get loss of network of SOLOv2. + Args: + ins_pred_list (list): Variable list of instance branch output. + ins_label_list (list): List of instance labels pre batch. + cate_preds (list): Concat Variable list of categroy branch output. + cate_labels (list): Concat list of categroy labels pre batch. + num_ins (int): Number of positive samples in a mini-batch. + Returns: + loss_ins (Variable): The instance loss Variable of SOLOv2 network. + loss_cate (Variable): The category loss Variable of SOLOv2 network. + """ + + #1. Ues dice_loss to calculate instance loss + loss_ins = [] + total_weights = paddle.zeros(shape=[1], dtype='float32') + for input, target in zip(ins_pred_list, ins_label_list): + if input is None: + continue + target = paddle.cast(target, 'float32') + target = paddle.reshape( + target, + shape=[-1, paddle.shape(input)[-2], paddle.shape(input)[-1]]) + weights = paddle.cast( + paddle.sum(target, axis=[1, 2]) > 0, 'float32') + input = F.sigmoid(input) + dice_out = paddle.multiply(self._dice_loss(input, target), weights) + total_weights += paddle.sum(weights) + loss_ins.append(dice_out) + loss_ins = paddle.sum(paddle.concat(loss_ins)) / total_weights + loss_ins = loss_ins * self.ins_loss_weight + + #2. Ues sigmoid_focal_loss to calculate category loss + # expand onehot labels + num_classes = cate_preds.shape[-1] + cate_labels_bin = F.one_hot(cate_labels, num_classes=num_classes + 1) + cate_labels_bin = cate_labels_bin[:, 1:] + + loss_cate = F.sigmoid_focal_loss( + cate_preds, + label=cate_labels_bin, + normalizer=num_ins + 1., + gamma=self.focal_loss_gamma, + alpha=self.focal_loss_alpha) + + return loss_ins, loss_cate diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/sparsercnn_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/sparsercnn_loss.py new file mode 100644 index 000000000..2d36b21a2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/sparsercnn_loss.py @@ -0,0 +1,425 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/PeizeSun/SparseR-CNN/blob/main/projects/SparseRCNN/sparsercnn/loss.py +Ths copyright of PeizeSun/SparseR-CNN is as follows: +MIT License [see LICENSE for details] +""" + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from scipy.optimize import linear_sum_assignment +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.metric import accuracy +from ppdet.core.workspace import register +from ppdet.modeling.losses.iou_loss import GIoULoss + +__all__ = ["SparseRCNNLoss"] + + +@register +class SparseRCNNLoss(nn.Layer): + """ This class computes the loss for SparseRCNN. + The process happens in two steps: + 1) we compute hungarian assignment between ground truth boxes and the outputs of the model + 2) we supervise each pair of matched ground-truth / prediction (supervise class and box) + """ + __shared__ = ['num_classes'] + + def __init__(self, + losses, + focal_loss_alpha, + focal_loss_gamma, + num_classes=80, + class_weight=2., + l1_weight=5., + giou_weight=2.): + """ Create the criterion. + Parameters: + num_classes: number of object categories, omitting the special no-object category + weight_dict: dict containing as key the names of the losses and as values their relative weight. + losses: list of all the losses to be applied. See get_loss for list of available losses. + matcher: module able to compute a matching between targets and proposals + """ + super().__init__() + self.num_classes = num_classes + weight_dict = { + "loss_ce": class_weight, + "loss_bbox": l1_weight, + "loss_giou": giou_weight + } + self.weight_dict = weight_dict + self.losses = losses + self.giou_loss = GIoULoss(reduction="sum") + + self.focal_loss_alpha = focal_loss_alpha + self.focal_loss_gamma = focal_loss_gamma + + self.matcher = HungarianMatcher(focal_loss_alpha, focal_loss_gamma, + class_weight, l1_weight, giou_weight) + + def loss_labels(self, outputs, targets, indices, num_boxes, log=True): + """Classification loss (NLL) + targets dicts must contain the key "labels" containing a tensor of dim [nb_target_boxes] + """ + assert 'pred_logits' in outputs + src_logits = outputs['pred_logits'] + + idx = self._get_src_permutation_idx(indices) + target_classes_o = paddle.concat([ + paddle.gather( + t["labels"], J, axis=0) for t, (_, J) in zip(targets, indices) + ]) + target_classes = paddle.full( + src_logits.shape[:2], self.num_classes, dtype="int32") + for i, ind in enumerate(zip(idx[0], idx[1])): + target_classes[int(ind[0]), int(ind[1])] = target_classes_o[i] + target_classes.stop_gradient = True + + src_logits = src_logits.flatten(start_axis=0, stop_axis=1) + + # prepare one_hot target. + target_classes = target_classes.flatten(start_axis=0, stop_axis=1) + class_ids = paddle.arange(0, self.num_classes) + labels = (target_classes.unsqueeze(-1) == class_ids).astype("float32") + labels.stop_gradient = True + + # comp focal loss. + class_loss = sigmoid_focal_loss( + src_logits, + labels, + alpha=self.focal_loss_alpha, + gamma=self.focal_loss_gamma, + reduction="sum", ) / num_boxes + losses = {'loss_ce': class_loss} + + if log: + label_acc = target_classes_o.unsqueeze(-1) + src_idx = [src for (src, _) in indices] + + pred_list = [] + for i in range(outputs["pred_logits"].shape[0]): + pred_list.append( + paddle.gather( + outputs["pred_logits"][i], src_idx[i], axis=0)) + + pred = F.sigmoid(paddle.concat(pred_list, axis=0)) + acc = accuracy(pred, label_acc.astype("int64")) + losses["acc"] = acc + + return losses + + def loss_boxes(self, outputs, targets, indices, num_boxes): + """Compute the losses related to the bounding boxes, the L1 regression loss and the GIoU loss + targets dicts must contain the key "boxes" containing a tensor of dim [nb_target_boxes, 4] + The target boxes are expected in format (center_x, center_y, w, h), normalized by the image size. + """ + assert 'pred_boxes' in outputs # [batch_size, num_proposals, 4] + src_idx = [src for (src, _) in indices] + src_boxes_list = [] + + for i in range(outputs["pred_boxes"].shape[0]): + src_boxes_list.append( + paddle.gather( + outputs["pred_boxes"][i], src_idx[i], axis=0)) + + src_boxes = paddle.concat(src_boxes_list, axis=0) + + target_boxes = paddle.concat( + [ + paddle.gather( + t['boxes'], I, axis=0) + for t, (_, I) in zip(targets, indices) + ], + axis=0) + target_boxes.stop_gradient = True + losses = {} + + losses['loss_giou'] = self.giou_loss(src_boxes, + target_boxes) / num_boxes + + image_size = paddle.concat([v["img_whwh_tgt"] for v in targets]) + src_boxes_ = src_boxes / image_size + target_boxes_ = target_boxes / image_size + + loss_bbox = F.l1_loss(src_boxes_, target_boxes_, reduction='sum') + losses['loss_bbox'] = loss_bbox / num_boxes + + return losses + + def _get_src_permutation_idx(self, indices): + # permute predictions following indices + batch_idx = paddle.concat( + [paddle.full_like(src, i) for i, (src, _) in enumerate(indices)]) + src_idx = paddle.concat([src for (src, _) in indices]) + return batch_idx, src_idx + + def _get_tgt_permutation_idx(self, indices): + # permute targets following indices + batch_idx = paddle.concat( + [paddle.full_like(tgt, i) for i, (_, tgt) in enumerate(indices)]) + tgt_idx = paddle.concat([tgt for (_, tgt) in indices]) + return batch_idx, tgt_idx + + def get_loss(self, loss, outputs, targets, indices, num_boxes, **kwargs): + loss_map = { + 'labels': self.loss_labels, + 'boxes': self.loss_boxes, + } + assert loss in loss_map, f'do you really want to compute {loss} loss?' + return loss_map[loss](outputs, targets, indices, num_boxes, **kwargs) + + def forward(self, outputs, targets): + """ This performs the loss computation. + Parameters: + outputs: dict of tensors, see the output specification of the model for the format + targets: list of dicts, such that len(targets) == batch_size. + The expected keys in each dict depends on the losses applied, see each loss' doc + """ + outputs_without_aux = { + k: v + for k, v in outputs.items() if k != 'aux_outputs' + } + + # Retrieve the matching between the outputs of the last layer and the targets + indices = self.matcher(outputs_without_aux, targets) + + # Compute the average number of target boxes accross all nodes, for normalization purposes + num_boxes = sum(len(t["labels"]) for t in targets) + num_boxes = paddle.to_tensor( + [num_boxes], + dtype="float32", + place=next(iter(outputs.values())).place) + + # Compute all the requested losses + losses = {} + for loss in self.losses: + losses.update( + self.get_loss(loss, outputs, targets, indices, num_boxes)) + + # In case of auxiliary losses, we repeat this process with the output of each intermediate layer. + if 'aux_outputs' in outputs: + for i, aux_outputs in enumerate(outputs['aux_outputs']): + indices = self.matcher(aux_outputs, targets) + for loss in self.losses: + kwargs = {} + if loss == 'labels': + # Logging is enabled only for the last layer + kwargs = {'log': False} + l_dict = self.get_loss(loss, aux_outputs, targets, indices, + num_boxes, **kwargs) + + w_dict = {} + for k in l_dict.keys(): + if k in self.weight_dict: + w_dict[k + f'_{i}'] = l_dict[k] * self.weight_dict[ + k] + else: + w_dict[k + f'_{i}'] = l_dict[k] + losses.update(w_dict) + + return losses + + +class HungarianMatcher(nn.Layer): + """This class computes an assignment between the targets and the predictions of the network + For efficiency reasons, the targets don't include the no_object. Because of this, in general, + there are more predictions than targets. In this case, we do a 1-to-1 matching of the best predictions, + while the others are un-matched (and thus treated as non-objects). + """ + + def __init__(self, + focal_loss_alpha, + focal_loss_gamma, + cost_class: float=1, + cost_bbox: float=1, + cost_giou: float=1): + """Creates the matcher + Params: + cost_class: This is the relative weight of the classification error in the matching cost + cost_bbox: This is the relative weight of the L1 error of the bounding box coordinates in the matching cost + cost_giou: This is the relative weight of the giou loss of the bounding box in the matching cost + """ + super().__init__() + self.cost_class = cost_class + self.cost_bbox = cost_bbox + self.cost_giou = cost_giou + self.focal_loss_alpha = focal_loss_alpha + self.focal_loss_gamma = focal_loss_gamma + assert cost_class != 0 or cost_bbox != 0 or cost_giou != 0, "all costs cant be 0" + + @paddle.no_grad() + def forward(self, outputs, targets): + """ Performs the matching + Args: + outputs: This is a dict that contains at least these entries: + "pred_logits": Tensor of dim [batch_size, num_queries, num_classes] with the classification logits + "pred_boxes": Tensor of dim [batch_size, num_queries, 4] with the predicted box coordinates + eg. outputs = {"pred_logits": pred_logits, "pred_boxes": pred_boxes} + targets: This is a list of targets (len(targets) = batch_size), where each target is a dict containing: + "labels": Tensor of dim [num_target_boxes] (where num_target_boxes is the number of ground-truth + objects in the target) containing the class labels + "boxes": Tensor of dim [num_target_boxes, 4] containing the target box coordinates + eg. targets = [{"labels":labels, "boxes": boxes}, ...,{"labels":labels, "boxes": boxes}] + Returns: + A list of size batch_size, containing tuples of (index_i, index_j) where: + - index_i is the indices of the selected predictions (in order) + - index_j is the indices of the corresponding selected targets (in order) + For each batch element, it holds: + len(index_i) = len(index_j) = min(num_queries, num_target_boxes) + """ + bs, num_queries = outputs["pred_logits"].shape[:2] + + # We flatten to compute the cost matrices in a batch + out_prob = F.sigmoid(outputs["pred_logits"].flatten( + start_axis=0, stop_axis=1)) + out_bbox = outputs["pred_boxes"].flatten(start_axis=0, stop_axis=1) + + # Also concat the target labels and boxes + tgt_ids = paddle.concat([v["labels"] for v in targets]) + assert (tgt_ids > -1).all() + tgt_bbox = paddle.concat([v["boxes"] for v in targets]) + + # Compute the classification cost. Contrary to the loss, we don't use the NLL, + # but approximate it in 1 - proba[target class]. + # The 1 is a constant that doesn't change the matching, it can be ommitted. + + # Compute the classification cost. + alpha = self.focal_loss_alpha + gamma = self.focal_loss_gamma + + neg_cost_class = (1 - alpha) * (out_prob**gamma) * (-( + 1 - out_prob + 1e-8).log()) + pos_cost_class = alpha * ((1 - out_prob) + **gamma) * (-(out_prob + 1e-8).log()) + + cost_class = paddle.gather( + pos_cost_class, tgt_ids, axis=1) - paddle.gather( + neg_cost_class, tgt_ids, axis=1) + + # Compute the L1 cost between boxes + image_size_out = paddle.concat( + [v["img_whwh"].unsqueeze(0) for v in targets]) + image_size_out = image_size_out.unsqueeze(1).tile( + [1, num_queries, 1]).flatten( + start_axis=0, stop_axis=1) + image_size_tgt = paddle.concat([v["img_whwh_tgt"] for v in targets]) + + out_bbox_ = out_bbox / image_size_out + tgt_bbox_ = tgt_bbox / image_size_tgt + cost_bbox = F.l1_loss( + out_bbox_.unsqueeze(-2), tgt_bbox_, + reduction='none').sum(-1) # [batch_size * num_queries, num_tgts] + + # Compute the giou cost betwen boxes + cost_giou = -get_bboxes_giou(out_bbox, tgt_bbox) + + # Final cost matrix + C = self.cost_bbox * cost_bbox + self.cost_class * cost_class + self.cost_giou * cost_giou + C = C.reshape([bs, num_queries, -1]) + + sizes = [len(v["boxes"]) for v in targets] + + indices = [ + linear_sum_assignment(c[i].numpy()) + for i, c in enumerate(C.split(sizes, -1)) + ] + return [(paddle.to_tensor( + i, dtype="int32"), paddle.to_tensor( + j, dtype="int32")) for i, j in indices] + + +def box_area(boxes): + assert (boxes[:, 2:] >= boxes[:, :2]).all() + wh = boxes[:, 2:] - boxes[:, :2] + return wh[:, 0] * wh[:, 1] + + +def boxes_iou(boxes1, boxes2): + ''' + Compute iou + + Args: + boxes1 (paddle.tensor) shape (N, 4) + boxes2 (paddle.tensor) shape (M, 4) + + Return: + (paddle.tensor) shape (N, M) + ''' + area1 = box_area(boxes1) + area2 = box_area(boxes2) + + lt = paddle.maximum(boxes1.unsqueeze(-2)[:, :, :2], boxes2[:, :2]) + rb = paddle.minimum(boxes1.unsqueeze(-2)[:, :, 2:], boxes2[:, 2:]) + + wh = (rb - lt).astype("float32").clip(min=1e-9) + inter = wh[:, :, 0] * wh[:, :, 1] + + union = area1.unsqueeze(-1) + area2 - inter + 1e-9 + + iou = inter / union + return iou, union + + +def get_bboxes_giou(boxes1, boxes2, eps=1e-9): + """calculate the ious of boxes1 and boxes2 + + Args: + boxes1 (Tensor): shape [N, 4] + boxes2 (Tensor): shape [M, 4] + eps (float): epsilon to avoid divide by zero + + Return: + ious (Tensor): ious of boxes1 and boxes2, with the shape [N, M] + """ + assert (boxes1[:, 2:] >= boxes1[:, :2]).all() + assert (boxes2[:, 2:] >= boxes2[:, :2]).all() + + iou, union = boxes_iou(boxes1, boxes2) + + lt = paddle.minimum(boxes1.unsqueeze(-2)[:, :, :2], boxes2[:, :2]) + rb = paddle.maximum(boxes1.unsqueeze(-2)[:, :, 2:], boxes2[:, 2:]) + + wh = (rb - lt).astype("float32").clip(min=eps) + enclose_area = wh[:, :, 0] * wh[:, :, 1] + + giou = iou - (enclose_area - union) / enclose_area + + return giou + + +def sigmoid_focal_loss(inputs, targets, alpha, gamma, reduction="sum"): + + assert reduction in ["sum", "mean" + ], f'do not support this {reduction} reduction?' + + p = F.sigmoid(inputs) + ce_loss = F.binary_cross_entropy_with_logits( + inputs, targets, reduction="none") + p_t = p * targets + (1 - p) * (1 - targets) + loss = ce_loss * ((1 - p_t)**gamma) + + if alpha >= 0: + alpha_t = alpha * targets + (1 - alpha) * (1 - targets) + loss = alpha_t * loss + + if reduction == "mean": + loss = loss.mean() + elif reduction == "sum": + loss = loss.sum() + + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ssd_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ssd_loss.py new file mode 100644 index 000000000..62aecc1f3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/ssd_loss.py @@ -0,0 +1,169 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +from ..ops import iou_similarity +from ..bbox_utils import bbox2delta + +__all__ = ['SSDLoss'] + + +@register +class SSDLoss(nn.Layer): + """ + SSDLoss + + Args: + overlap_threshold (float32, optional): IoU threshold for negative bboxes + and positive bboxes, 0.5 by default. + neg_pos_ratio (float): The ratio of negative samples / positive samples. + loc_loss_weight (float): The weight of loc_loss. + conf_loss_weight (float): The weight of conf_loss. + prior_box_var (list): Variances corresponding to prior box coord, [0.1, + 0.1, 0.2, 0.2] by default. + """ + + def __init__(self, + overlap_threshold=0.5, + neg_pos_ratio=3.0, + loc_loss_weight=1.0, + conf_loss_weight=1.0, + prior_box_var=[0.1, 0.1, 0.2, 0.2]): + super(SSDLoss, self).__init__() + self.overlap_threshold = overlap_threshold + self.neg_pos_ratio = neg_pos_ratio + self.loc_loss_weight = loc_loss_weight + self.conf_loss_weight = conf_loss_weight + self.prior_box_var = [1. / a for a in prior_box_var] + + def _bipartite_match_for_batch(self, gt_bbox, gt_label, prior_boxes, + bg_index): + """ + Args: + gt_bbox (Tensor): [B, N, 4] + gt_label (Tensor): [B, N, 1] + prior_boxes (Tensor): [A, 4] + bg_index (int): Background class index + """ + batch_size, num_priors = gt_bbox.shape[0], prior_boxes.shape[0] + ious = iou_similarity(gt_bbox.reshape((-1, 4)), prior_boxes).reshape( + (batch_size, -1, num_priors)) + + # For each prior box, get the max IoU of all GTs. + prior_max_iou, prior_argmax_iou = ious.max(axis=1), ious.argmax(axis=1) + # For each GT, get the max IoU of all prior boxes. + gt_max_iou, gt_argmax_iou = ious.max(axis=2), ious.argmax(axis=2) + + # Gather target bbox and label according to 'prior_argmax_iou' index. + batch_ind = paddle.arange(end=batch_size, dtype='int64').unsqueeze(-1) + prior_argmax_iou = paddle.stack( + [batch_ind.tile([1, num_priors]), prior_argmax_iou], axis=-1) + targets_bbox = paddle.gather_nd(gt_bbox, prior_argmax_iou) + targets_label = paddle.gather_nd(gt_label, prior_argmax_iou) + # Assign negative + bg_index_tensor = paddle.full([batch_size, num_priors, 1], bg_index, + 'int64') + targets_label = paddle.where( + prior_max_iou.unsqueeze(-1) < self.overlap_threshold, + bg_index_tensor, targets_label) + + # Ensure each GT can match the max IoU prior box. + batch_ind = (batch_ind * num_priors + gt_argmax_iou).flatten() + targets_bbox = paddle.scatter( + targets_bbox.reshape([-1, 4]), batch_ind, + gt_bbox.reshape([-1, 4])).reshape([batch_size, -1, 4]) + targets_label = paddle.scatter( + targets_label.reshape([-1, 1]), batch_ind, + gt_label.reshape([-1, 1])).reshape([batch_size, -1, 1]) + targets_label[:, :1] = bg_index + + # Encode box + prior_boxes = prior_boxes.unsqueeze(0).tile([batch_size, 1, 1]) + targets_bbox = bbox2delta( + prior_boxes.reshape([-1, 4]), + targets_bbox.reshape([-1, 4]), self.prior_box_var) + targets_bbox = targets_bbox.reshape([batch_size, -1, 4]) + + return targets_bbox, targets_label + + def _mine_hard_example(self, + conf_loss, + targets_label, + bg_index, + mine_neg_ratio=0.01): + pos = (targets_label != bg_index).astype(conf_loss.dtype) + num_pos = pos.sum(axis=1, keepdim=True) + neg = (targets_label == bg_index).astype(conf_loss.dtype) + + conf_loss = conf_loss.detach() * neg + loss_idx = conf_loss.argsort(axis=1, descending=True) + idx_rank = loss_idx.argsort(axis=1) + num_negs = [] + for i in range(conf_loss.shape[0]): + cur_num_pos = num_pos[i] + num_neg = paddle.clip( + cur_num_pos * self.neg_pos_ratio, max=pos.shape[1]) + num_neg = num_neg if num_neg > 0 else paddle.to_tensor( + [pos.shape[1] * mine_neg_ratio]) + num_negs.append(num_neg) + num_negs = paddle.stack(num_negs).expand_as(idx_rank) + neg_mask = (idx_rank < num_negs).astype(conf_loss.dtype) + + return (neg_mask + pos).astype('bool') + + def forward(self, boxes, scores, gt_bbox, gt_label, prior_boxes): + boxes = paddle.concat(boxes, axis=1) + scores = paddle.concat(scores, axis=1) + gt_label = gt_label.unsqueeze(-1).astype('int64') + prior_boxes = paddle.concat(prior_boxes, axis=0) + bg_index = scores.shape[-1] - 1 + + # Match bbox and get targets. + targets_bbox, targets_label = \ + self._bipartite_match_for_batch(gt_bbox, gt_label, prior_boxes, bg_index) + targets_bbox.stop_gradient = True + targets_label.stop_gradient = True + + # Compute regression loss. + # Select positive samples. + bbox_mask = paddle.tile(targets_label != bg_index, [1, 1, 4]) + if bbox_mask.astype(boxes.dtype).sum() > 0: + location = paddle.masked_select(boxes, bbox_mask) + targets_bbox = paddle.masked_select(targets_bbox, bbox_mask) + loc_loss = F.smooth_l1_loss(location, targets_bbox, reduction='sum') + loc_loss = loc_loss * self.loc_loss_weight + else: + loc_loss = paddle.zeros([1]) + + # Compute confidence loss. + conf_loss = F.cross_entropy(scores, targets_label, reduction="none") + # Mining hard examples. + label_mask = self._mine_hard_example( + conf_loss.squeeze(-1), targets_label.squeeze(-1), bg_index) + conf_loss = paddle.masked_select(conf_loss, label_mask.unsqueeze(-1)) + conf_loss = conf_loss.sum() * self.conf_loss_weight + + # Compute overall weighted loss. + normalizer = (targets_label != bg_index).astype('float32').sum().clip( + min=1) + loss = (conf_loss + loc_loss) / normalizer + + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/varifocal_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/varifocal_loss.py new file mode 100644 index 000000000..42d18a659 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/varifocal_loss.py @@ -0,0 +1,152 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/losses/varifocal_loss.py + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ppdet.modeling import ops + +__all__ = ['VarifocalLoss'] + + +def varifocal_loss(pred, + target, + alpha=0.75, + gamma=2.0, + iou_weighted=True, + use_sigmoid=True): + """`Varifocal Loss `_ + + Args: + pred (Tensor): The prediction with shape (N, C), C is the + number of classes + target (Tensor): The learning target of the iou-aware + classification score with shape (N, C), C is the number of classes. + alpha (float, optional): A balance factor for the negative part of + Varifocal Loss, which is different from the alpha of Focal Loss. + Defaults to 0.75. + gamma (float, optional): The gamma for calculating the modulating + factor. Defaults to 2.0. + iou_weighted (bool, optional): Whether to weight the loss of the + positive example with the iou target. Defaults to True. + """ + # pred and target should be of the same size + assert pred.shape == target.shape + if use_sigmoid: + pred_new = F.sigmoid(pred) + else: + pred_new = pred + target = target.cast(pred.dtype) + if iou_weighted: + focal_weight = target * (target > 0.0).cast('float32') + \ + alpha * (pred_new - target).abs().pow(gamma) * \ + (target <= 0.0).cast('float32') + else: + focal_weight = (target > 0.0).cast('float32') + \ + alpha * (pred_new - target).abs().pow(gamma) * \ + (target <= 0.0).cast('float32') + + if use_sigmoid: + loss = F.binary_cross_entropy_with_logits( + pred, target, reduction='none') * focal_weight + else: + loss = F.binary_cross_entropy( + pred, target, reduction='none') * focal_weight + loss = loss.sum(axis=1) + return loss + + +@register +@serializable +class VarifocalLoss(nn.Layer): + def __init__(self, + use_sigmoid=True, + alpha=0.75, + gamma=2.0, + iou_weighted=True, + reduction='mean', + loss_weight=1.0): + """`Varifocal Loss `_ + + Args: + use_sigmoid (bool, optional): Whether the prediction is + used for sigmoid or softmax. Defaults to True. + alpha (float, optional): A balance factor for the negative part of + Varifocal Loss, which is different from the alpha of Focal + Loss. Defaults to 0.75. + gamma (float, optional): The gamma for calculating the modulating + factor. Defaults to 2.0. + iou_weighted (bool, optional): Whether to weight the loss of the + positive examples with the iou target. Defaults to True. + reduction (str, optional): The method used to reduce the loss into + a scalar. Defaults to 'mean'. Options are "none", "mean" and + "sum". + loss_weight (float, optional): Weight of loss. Defaults to 1.0. + """ + super(VarifocalLoss, self).__init__() + assert alpha >= 0.0 + self.use_sigmoid = use_sigmoid + self.alpha = alpha + self.gamma = gamma + self.iou_weighted = iou_weighted + self.reduction = reduction + self.loss_weight = loss_weight + + def forward(self, pred, target, weight=None, avg_factor=None): + """Forward function. + + Args: + pred (Tensor): The prediction. + target (Tensor): The learning target of the prediction. + weight (Tensor, optional): The weight of loss for each + prediction. Defaults to None. + avg_factor (int, optional): Average factor that is used to average + the loss. Defaults to None. + Returns: + Tensor: The calculated loss + """ + loss = self.loss_weight * varifocal_loss( + pred, + target, + alpha=self.alpha, + gamma=self.gamma, + iou_weighted=self.iou_weighted, + use_sigmoid=self.use_sigmoid) + + if weight is not None: + loss = loss * weight + if avg_factor is None: + if self.reduction == 'none': + return loss + elif self.reduction == 'mean': + return loss.mean() + elif self.reduction == 'sum': + return loss.sum() + else: + # if reduction is mean, then average the loss by avg_factor + if self.reduction == 'mean': + loss = loss.sum() / avg_factor + # if reduction is 'none', then do nothing, otherwise raise an error + elif self.reduction != 'none': + raise ValueError( + 'avg_factor can not be used with reduction="sum"') + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/yolo_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/yolo_loss.py new file mode 100644 index 000000000..657959cd7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/losses/yolo_loss.py @@ -0,0 +1,206 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register + +from ..bbox_utils import decode_yolo, xywh2xyxy, iou_similarity + +__all__ = ['YOLOv3Loss'] + + +def bbox_transform(pbox, anchor, downsample): + pbox = decode_yolo(pbox, anchor, downsample) + pbox = xywh2xyxy(pbox) + return pbox + + +@register +class YOLOv3Loss(nn.Layer): + + __inject__ = ['iou_loss', 'iou_aware_loss'] + __shared__ = ['num_classes'] + + def __init__(self, + num_classes=80, + ignore_thresh=0.7, + label_smooth=False, + downsample=[32, 16, 8], + scale_x_y=1., + iou_loss=None, + iou_aware_loss=None): + """ + YOLOv3Loss layer + + Args: + num_calsses (int): number of foreground classes + ignore_thresh (float): threshold to ignore confidence loss + label_smooth (bool): whether to use label smoothing + downsample (list): downsample ratio for each detection block + scale_x_y (float): scale_x_y factor + iou_loss (object): IoULoss instance + iou_aware_loss (object): IouAwareLoss instance + """ + super(YOLOv3Loss, self).__init__() + self.num_classes = num_classes + self.ignore_thresh = ignore_thresh + self.label_smooth = label_smooth + self.downsample = downsample + self.scale_x_y = scale_x_y + self.iou_loss = iou_loss + self.iou_aware_loss = iou_aware_loss + self.distill_pairs = [] + + def obj_loss(self, pbox, gbox, pobj, tobj, anchor, downsample): + # pbox + pbox = decode_yolo(pbox, anchor, downsample) + pbox = xywh2xyxy(pbox) + pbox = paddle.concat(pbox, axis=-1) + b = pbox.shape[0] + pbox = pbox.reshape((b, -1, 4)) + # gbox + gxy = gbox[:, :, 0:2] - gbox[:, :, 2:4] * 0.5 + gwh = gbox[:, :, 0:2] + gbox[:, :, 2:4] * 0.5 + gbox = paddle.concat([gxy, gwh], axis=-1) + + iou = iou_similarity(pbox, gbox) + iou.stop_gradient = True + iou_max = iou.max(2) # [N, M1] + iou_mask = paddle.cast(iou_max <= self.ignore_thresh, dtype=pbox.dtype) + iou_mask.stop_gradient = True + + pobj = pobj.reshape((b, -1)) + tobj = tobj.reshape((b, -1)) + obj_mask = paddle.cast(tobj > 0, dtype=pbox.dtype) + obj_mask.stop_gradient = True + + loss_obj = F.binary_cross_entropy_with_logits( + pobj, obj_mask, reduction='none') + loss_obj_pos = (loss_obj * tobj) + loss_obj_neg = (loss_obj * (1 - obj_mask) * iou_mask) + return loss_obj_pos + loss_obj_neg + + def cls_loss(self, pcls, tcls): + if self.label_smooth: + delta = min(1. / self.num_classes, 1. / 40) + pos, neg = 1 - delta, delta + # 1 for positive, 0 for negative + tcls = pos * paddle.cast( + tcls > 0., dtype=tcls.dtype) + neg * paddle.cast( + tcls <= 0., dtype=tcls.dtype) + + loss_cls = F.binary_cross_entropy_with_logits( + pcls, tcls, reduction='none') + return loss_cls + + def yolov3_loss(self, p, t, gt_box, anchor, downsample, scale=1., + eps=1e-10): + na = len(anchor) + b, c, h, w = p.shape + if self.iou_aware_loss: + ioup, p = p[:, 0:na, :, :], p[:, na:, :, :] + ioup = ioup.unsqueeze(-1) + p = p.reshape((b, na, -1, h, w)).transpose((0, 1, 3, 4, 2)) + x, y = p[:, :, :, :, 0:1], p[:, :, :, :, 1:2] + w, h = p[:, :, :, :, 2:3], p[:, :, :, :, 3:4] + obj, pcls = p[:, :, :, :, 4:5], p[:, :, :, :, 5:] + self.distill_pairs.append([x, y, w, h, obj, pcls]) + + t = t.transpose((0, 1, 3, 4, 2)) + tx, ty = t[:, :, :, :, 0:1], t[:, :, :, :, 1:2] + tw, th = t[:, :, :, :, 2:3], t[:, :, :, :, 3:4] + tscale = t[:, :, :, :, 4:5] + tobj, tcls = t[:, :, :, :, 5:6], t[:, :, :, :, 6:] + + tscale_obj = tscale * tobj + loss = dict() + + x = scale * F.sigmoid(x) - 0.5 * (scale - 1.) + y = scale * F.sigmoid(y) - 0.5 * (scale - 1.) + + if abs(scale - 1.) < eps: + loss_x = F.binary_cross_entropy(x, tx, reduction='none') + loss_y = F.binary_cross_entropy(y, ty, reduction='none') + loss_xy = tscale_obj * (loss_x + loss_y) + else: + loss_x = paddle.abs(x - tx) + loss_y = paddle.abs(y - ty) + loss_xy = tscale_obj * (loss_x + loss_y) + + loss_xy = loss_xy.sum([1, 2, 3, 4]).mean() + + loss_w = paddle.abs(w - tw) + loss_h = paddle.abs(h - th) + loss_wh = tscale_obj * (loss_w + loss_h) + loss_wh = loss_wh.sum([1, 2, 3, 4]).mean() + + loss['loss_xy'] = loss_xy + loss['loss_wh'] = loss_wh + + if self.iou_loss is not None: + # warn: do not modify x, y, w, h in place + box, tbox = [x, y, w, h], [tx, ty, tw, th] + pbox = bbox_transform(box, anchor, downsample) + gbox = bbox_transform(tbox, anchor, downsample) + loss_iou = self.iou_loss(pbox, gbox) + loss_iou = loss_iou * tscale_obj + loss_iou = loss_iou.sum([1, 2, 3, 4]).mean() + loss['loss_iou'] = loss_iou + + if self.iou_aware_loss is not None: + box, tbox = [x, y, w, h], [tx, ty, tw, th] + pbox = bbox_transform(box, anchor, downsample) + gbox = bbox_transform(tbox, anchor, downsample) + loss_iou_aware = self.iou_aware_loss(ioup, pbox, gbox) + loss_iou_aware = loss_iou_aware * tobj + loss_iou_aware = loss_iou_aware.sum([1, 2, 3, 4]).mean() + loss['loss_iou_aware'] = loss_iou_aware + + box = [x, y, w, h] + loss_obj = self.obj_loss(box, gt_box, obj, tobj, anchor, downsample) + loss_obj = loss_obj.sum(-1).mean() + loss['loss_obj'] = loss_obj + loss_cls = self.cls_loss(pcls, tcls) * tobj + loss_cls = loss_cls.sum([1, 2, 3, 4]).mean() + loss['loss_cls'] = loss_cls + return loss + + def forward(self, inputs, targets, anchors): + np = len(inputs) + gt_targets = [targets['target{}'.format(i)] for i in range(np)] + gt_box = targets['gt_bbox'] + yolo_losses = dict() + self.distill_pairs.clear() + for x, t, anchor, downsample in zip(inputs, gt_targets, anchors, + self.downsample): + yolo_loss = self.yolov3_loss(x, t, gt_box, anchor, downsample, + self.scale_x_y) + for k, v in yolo_loss.items(): + if k in yolo_losses: + yolo_losses[k] += v + else: + yolo_losses[k] = v + + loss = 0 + for k, v in yolo_losses.items(): + loss += v + + yolo_losses['loss'] = loss + return yolo_losses diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__init__.py new file mode 100644 index 000000000..258e4c901 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__init__.py @@ -0,0 +1,25 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import matching +from . import tracker +from . import motion +from . import visualization +from . import utils + +from .matching import * +from .tracker import * +from .motion import * +from .visualization import * +from .utils import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..2ee33f2d9 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/utils.cpython-37.pyc new file mode 100644 index 000000000..e00b5655a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/visualization.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/visualization.cpython-37.pyc new file mode 100644 index 000000000..5b2c5807a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/__pycache__/visualization.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__init__.py new file mode 100644 index 000000000..54c6680f7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__init__.py @@ -0,0 +1,19 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import jde_matching +from . import deepsort_matching + +from .jde_matching import * +from .deepsort_matching import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..2ea1093b4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/deepsort_matching.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/deepsort_matching.cpython-37.pyc new file mode 100644 index 000000000..c5cb999da Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/deepsort_matching.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/jde_matching.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/jde_matching.cpython-37.pyc new file mode 100644 index 000000000..37ac0e28c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/__pycache__/jde_matching.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/deepsort_matching.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/deepsort_matching.py new file mode 100644 index 000000000..3859ccfbd --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/deepsort_matching.py @@ -0,0 +1,379 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/nwojke/deep_sort/tree/master/deep_sort +""" + +import numpy as np +from scipy.optimize import linear_sum_assignment +from ..motion import kalman_filter + +INFTY_COST = 1e+5 + +__all__ = [ + 'iou_1toN', + 'iou_cost', + '_nn_euclidean_distance', + '_nn_cosine_distance', + 'NearestNeighborDistanceMetric', + 'min_cost_matching', + 'matching_cascade', + 'gate_cost_matrix', +] + + +def iou_1toN(bbox, candidates): + """ + Computer intersection over union (IoU) by one box to N candidates. + + Args: + bbox (ndarray): A bounding box in format `(top left x, top left y, width, height)`. + candidates (ndarray): A matrix of candidate bounding boxes (one per row) in the + same format as `bbox`. + + Returns: + ious (ndarray): The intersection over union in [0, 1] between the `bbox` + and each candidate. A higher score means a larger fraction of the + `bbox` is occluded by the candidate. + """ + bbox_tl = bbox[:2] + bbox_br = bbox[:2] + bbox[2:] + candidates_tl = candidates[:, :2] + candidates_br = candidates[:, :2] + candidates[:, 2:] + + tl = np.c_[np.maximum(bbox_tl[0], candidates_tl[:, 0])[:, np.newaxis], + np.maximum(bbox_tl[1], candidates_tl[:, 1])[:, np.newaxis]] + br = np.c_[np.minimum(bbox_br[0], candidates_br[:, 0])[:, np.newaxis], + np.minimum(bbox_br[1], candidates_br[:, 1])[:, np.newaxis]] + wh = np.maximum(0., br - tl) + + area_intersection = wh.prod(axis=1) + area_bbox = bbox[2:].prod() + area_candidates = candidates[:, 2:].prod(axis=1) + ious = area_intersection / (area_bbox + area_candidates - area_intersection) + return ious + + +def iou_cost(tracks, detections, track_indices=None, detection_indices=None): + """ + IoU distance metric. + + Args: + tracks (list[Track]): A list of tracks. + detections (list[Detection]): A list of detections. + track_indices (Optional[list[int]]): A list of indices to tracks that + should be matched. Defaults to all `tracks`. + detection_indices (Optional[list[int]]): A list of indices to detections + that should be matched. Defaults to all `detections`. + + Returns: + cost_matrix (ndarray): A cost matrix of shape len(track_indices), + len(detection_indices) where entry (i, j) is + `1 - iou(tracks[track_indices[i]], detections[detection_indices[j]])`. + """ + if track_indices is None: + track_indices = np.arange(len(tracks)) + if detection_indices is None: + detection_indices = np.arange(len(detections)) + + cost_matrix = np.zeros((len(track_indices), len(detection_indices))) + for row, track_idx in enumerate(track_indices): + if tracks[track_idx].time_since_update > 1: + cost_matrix[row, :] = 1e+5 + continue + + bbox = tracks[track_idx].to_tlwh() + candidates = np.asarray([detections[i].tlwh for i in detection_indices]) + cost_matrix[row, :] = 1. - iou_1toN(bbox, candidates) + return cost_matrix + + +def _nn_euclidean_distance(s, q): + """ + Compute pair-wise squared (Euclidean) distance between points in `s` and `q`. + + Args: + s (ndarray): Sample points: an NxM matrix of N samples of dimensionality M. + q (ndarray): Query points: an LxM matrix of L samples of dimensionality M. + + Returns: + distances (ndarray): A vector of length M that contains for each entry in `q` the + smallest Euclidean distance to a sample in `s`. + """ + s, q = np.asarray(s), np.asarray(q) + if len(s) == 0 or len(q) == 0: + return np.zeros((len(s), len(q))) + s2, q2 = np.square(s).sum(axis=1), np.square(q).sum(axis=1) + distances = -2. * np.dot(s, q.T) + s2[:, None] + q2[None, :] + distances = np.clip(distances, 0., float(np.inf)) + + return np.maximum(0.0, distances.min(axis=0)) + + +def _nn_cosine_distance(s, q): + """ + Compute pair-wise cosine distance between points in `s` and `q`. + + Args: + s (ndarray): Sample points: an NxM matrix of N samples of dimensionality M. + q (ndarray): Query points: an LxM matrix of L samples of dimensionality M. + + Returns: + distances (ndarray): A vector of length M that contains for each entry in `q` the + smallest Euclidean distance to a sample in `s`. + """ + s = np.asarray(s) / np.linalg.norm(s, axis=1, keepdims=True) + q = np.asarray(q) / np.linalg.norm(q, axis=1, keepdims=True) + distances = 1. - np.dot(s, q.T) + + return distances.min(axis=0) + + +class NearestNeighborDistanceMetric(object): + """ + A nearest neighbor distance metric that, for each target, returns + the closest distance to any sample that has been observed so far. + + Args: + metric (str): Either "euclidean" or "cosine". + matching_threshold (float): The matching threshold. Samples with larger + distance are considered an invalid match. + budget (Optional[int]): If not None, fix samples per class to at most + this number. Removes the oldest samples when the budget is reached. + + Attributes: + samples (Dict[int -> List[ndarray]]): A dictionary that maps from target + identities to the list of samples that have been observed so far. + """ + + def __init__(self, metric, matching_threshold, budget=None): + if metric == "euclidean": + self._metric = _nn_euclidean_distance + elif metric == "cosine": + self._metric = _nn_cosine_distance + else: + raise ValueError( + "Invalid metric; must be either 'euclidean' or 'cosine'") + self.matching_threshold = matching_threshold + self.budget = budget + self.samples = {} + + def partial_fit(self, features, targets, active_targets): + """ + Update the distance metric with new data. + + Args: + features (ndarray): An NxM matrix of N features of dimensionality M. + targets (ndarray): An integer array of associated target identities. + active_targets (List[int]): A list of targets that are currently + present in the scene. + """ + for feature, target in zip(features, targets): + self.samples.setdefault(target, []).append(feature) + if self.budget is not None: + self.samples[target] = self.samples[target][-self.budget:] + self.samples = {k: self.samples[k] for k in active_targets} + + def distance(self, features, targets): + """ + Compute distance between features and targets. + + Args: + features (ndarray): An NxM matrix of N features of dimensionality M. + targets (list[int]): A list of targets to match the given `features` against. + + Returns: + cost_matrix (ndarray): a cost matrix of shape len(targets), len(features), + where element (i, j) contains the closest squared distance between + `targets[i]` and `features[j]`. + """ + cost_matrix = np.zeros((len(targets), len(features))) + for i, target in enumerate(targets): + cost_matrix[i, :] = self._metric(self.samples[target], features) + return cost_matrix + + +def min_cost_matching(distance_metric, + max_distance, + tracks, + detections, + track_indices=None, + detection_indices=None): + """ + Solve linear assignment problem. + + Args: + distance_metric : + Callable[List[Track], List[Detection], List[int], List[int]) -> ndarray + The distance metric is given a list of tracks and detections as + well as a list of N track indices and M detection indices. The + metric should return the NxM dimensional cost matrix, where element + (i, j) is the association cost between the i-th track in the given + track indices and the j-th detection in the given detection_indices. + max_distance (float): Gating threshold. Associations with cost larger + than this value are disregarded. + tracks (list[Track]): A list of predicted tracks at the current time + step. + detections (list[Detection]): A list of detections at the current time + step. + track_indices (list[int]): List of track indices that maps rows in + `cost_matrix` to tracks in `tracks`. + detection_indices (List[int]): List of detection indices that maps + columns in `cost_matrix` to detections in `detections`. + + Returns: + A tuple (List[(int, int)], List[int], List[int]) with the following + three entries: + * A list of matched track and detection indices. + * A list of unmatched track indices. + * A list of unmatched detection indices. + """ + if track_indices is None: + track_indices = np.arange(len(tracks)) + if detection_indices is None: + detection_indices = np.arange(len(detections)) + + if len(detection_indices) == 0 or len(track_indices) == 0: + return [], track_indices, detection_indices # Nothing to match. + + cost_matrix = distance_metric(tracks, detections, track_indices, + detection_indices) + + cost_matrix[cost_matrix > max_distance] = max_distance + 1e-5 + indices = linear_sum_assignment(cost_matrix) + + matches, unmatched_tracks, unmatched_detections = [], [], [] + for col, detection_idx in enumerate(detection_indices): + if col not in indices[1]: + unmatched_detections.append(detection_idx) + for row, track_idx in enumerate(track_indices): + if row not in indices[0]: + unmatched_tracks.append(track_idx) + for row, col in zip(indices[0], indices[1]): + track_idx = track_indices[row] + detection_idx = detection_indices[col] + if cost_matrix[row, col] > max_distance: + unmatched_tracks.append(track_idx) + unmatched_detections.append(detection_idx) + else: + matches.append((track_idx, detection_idx)) + return matches, unmatched_tracks, unmatched_detections + + +def matching_cascade(distance_metric, + max_distance, + cascade_depth, + tracks, + detections, + track_indices=None, + detection_indices=None): + """ + Run matching cascade. + + Args: + distance_metric : + Callable[List[Track], List[Detection], List[int], List[int]) -> ndarray + The distance metric is given a list of tracks and detections as + well as a list of N track indices and M detection indices. The + metric should return the NxM dimensional cost matrix, where element + (i, j) is the association cost between the i-th track in the given + track indices and the j-th detection in the given detection_indices. + max_distance (float): Gating threshold. Associations with cost larger + than this value are disregarded. + cascade_depth (int): The cascade depth, should be se to the maximum + track age. + tracks (list[Track]): A list of predicted tracks at the current time + step. + detections (list[Detection]): A list of detections at the current time + step. + track_indices (list[int]): List of track indices that maps rows in + `cost_matrix` to tracks in `tracks`. + detection_indices (List[int]): List of detection indices that maps + columns in `cost_matrix` to detections in `detections`. + + Returns: + A tuple (List[(int, int)], List[int], List[int]) with the following + three entries: + * A list of matched track and detection indices. + * A list of unmatched track indices. + * A list of unmatched detection indices. + """ + if track_indices is None: + track_indices = list(range(len(tracks))) + if detection_indices is None: + detection_indices = list(range(len(detections))) + + unmatched_detections = detection_indices + matches = [] + for level in range(cascade_depth): + if len(unmatched_detections) == 0: # No detections left + break + + track_indices_l = [ + k for k in track_indices if tracks[k].time_since_update == 1 + level + ] + if len(track_indices_l) == 0: # Nothing to match at this level + continue + + matches_l, _, unmatched_detections = \ + min_cost_matching( + distance_metric, max_distance, tracks, detections, + track_indices_l, unmatched_detections) + matches += matches_l + unmatched_tracks = list(set(track_indices) - set(k for k, _ in matches)) + return matches, unmatched_tracks, unmatched_detections + + +def gate_cost_matrix(kf, + cost_matrix, + tracks, + detections, + track_indices, + detection_indices, + gated_cost=INFTY_COST, + only_position=False): + """ + Invalidate infeasible entries in cost matrix based on the state + distributions obtained by Kalman filtering. + + Args: + kf (object): The Kalman filter. + cost_matrix (ndarray): The NxM dimensional cost matrix, where N is the + number of track indices and M is the number of detection indices, + such that entry (i, j) is the association cost between + `tracks[track_indices[i]]` and `detections[detection_indices[j]]`. + tracks (list[Track]): A list of predicted tracks at the current time + step. + detections (list[Detection]): A list of detections at the current time + step. + track_indices (List[int]): List of track indices that maps rows in + `cost_matrix` to tracks in `tracks`. + detection_indices (List[int]): List of detection indices that maps + columns in `cost_matrix` to detections in `detections`. + gated_cost (Optional[float]): Entries in the cost matrix corresponding + to infeasible associations are set this value. Defaults to a very + large value. + only_position (Optional[bool]): If True, only the x, y position of the + state distribution is considered during gating. Default False. + """ + gating_dim = 2 if only_position else 4 + gating_threshold = kalman_filter.chi2inv95[gating_dim] + measurements = np.asarray( + [detections[i].to_xyah() for i in detection_indices]) + for row, track_idx in enumerate(track_indices): + track = tracks[track_idx] + gating_distance = kf.gating_distance(track.mean, track.covariance, + measurements, only_position) + cost_matrix[row, gating_distance > gating_threshold] = gated_cost + return cost_matrix diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/jde_matching.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/jde_matching.py new file mode 100644 index 000000000..e9c40dba4 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/matching/jde_matching.py @@ -0,0 +1,144 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/Zhongdao/Towards-Realtime-MOT/blob/master/tracker/matching.py +""" + +import lap +import scipy +import numpy as np +from scipy.spatial.distance import cdist +from ..motion import kalman_filter +import warnings +warnings.filterwarnings("ignore") + +__all__ = [ + 'merge_matches', + 'linear_assignment', + 'cython_bbox_ious', + 'iou_distance', + 'embedding_distance', + 'fuse_motion', +] + + +def merge_matches(m1, m2, shape): + O, P, Q = shape + m1 = np.asarray(m1) + m2 = np.asarray(m2) + + M1 = scipy.sparse.coo_matrix( + (np.ones(len(m1)), (m1[:, 0], m1[:, 1])), shape=(O, P)) + M2 = scipy.sparse.coo_matrix( + (np.ones(len(m2)), (m2[:, 0], m2[:, 1])), shape=(P, Q)) + + mask = M1 * M2 + match = mask.nonzero() + match = list(zip(match[0], match[1])) + unmatched_O = tuple(set(range(O)) - set([i for i, j in match])) + unmatched_Q = tuple(set(range(Q)) - set([j for i, j in match])) + + return match, unmatched_O, unmatched_Q + + +def linear_assignment(cost_matrix, thresh): + if cost_matrix.size == 0: + return np.empty( + (0, 2), dtype=int), tuple(range(cost_matrix.shape[0])), tuple( + range(cost_matrix.shape[1])) + matches, unmatched_a, unmatched_b = [], [], [] + cost, x, y = lap.lapjv(cost_matrix, extend_cost=True, cost_limit=thresh) + for ix, mx in enumerate(x): + if mx >= 0: + matches.append([ix, mx]) + unmatched_a = np.where(x < 0)[0] + unmatched_b = np.where(y < 0)[0] + matches = np.asarray(matches) + return matches, unmatched_a, unmatched_b + + +def cython_bbox_ious(atlbrs, btlbrs): + ious = np.zeros((len(atlbrs), len(btlbrs)), dtype=np.float) + if ious.size == 0: + return ious + try: + import cython_bbox + except Exception as e: + print('cython_bbox not found, please install cython_bbox.' + 'for example: `pip install cython_bbox`.') + raise e + + ious = cython_bbox.bbox_overlaps( + np.ascontiguousarray( + atlbrs, dtype=np.float), + np.ascontiguousarray( + btlbrs, dtype=np.float)) + return ious + + +def iou_distance(atracks, btracks): + """ + Compute cost based on IoU between two list[STrack]. + """ + if (len(atracks) > 0 and isinstance(atracks[0], np.ndarray)) or ( + len(btracks) > 0 and isinstance(btracks[0], np.ndarray)): + atlbrs = atracks + btlbrs = btracks + else: + atlbrs = [track.tlbr for track in atracks] + btlbrs = [track.tlbr for track in btracks] + _ious = cython_bbox_ious(atlbrs, btlbrs) + cost_matrix = 1 - _ious + + return cost_matrix + + +def embedding_distance(tracks, detections, metric='euclidean'): + """ + Compute cost based on features between two list[STrack]. + """ + cost_matrix = np.zeros((len(tracks), len(detections)), dtype=np.float) + if cost_matrix.size == 0: + return cost_matrix + det_features = np.asarray( + [track.curr_feat for track in detections], dtype=np.float) + track_features = np.asarray( + [track.smooth_feat for track in tracks], dtype=np.float) + cost_matrix = np.maximum(0.0, cdist(track_features, det_features, + metric)) # Nomalized features + return cost_matrix + + +def fuse_motion(kf, + cost_matrix, + tracks, + detections, + only_position=False, + lambda_=0.98): + if cost_matrix.size == 0: + return cost_matrix + gating_dim = 2 if only_position else 4 + gating_threshold = kalman_filter.chi2inv95[gating_dim] + measurements = np.asarray([det.to_xyah() for det in detections]) + for row, track in enumerate(tracks): + gating_distance = kf.gating_distance( + track.mean, + track.covariance, + measurements, + only_position, + metric='maha') + cost_matrix[row, gating_distance > gating_threshold] = np.inf + cost_matrix[row] = lambda_ * cost_matrix[row] + (1 - lambda_ + ) * gating_distance + return cost_matrix diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__init__.py new file mode 100644 index 000000000..e42dd0b01 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__init__.py @@ -0,0 +1,17 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import kalman_filter + +from .kalman_filter import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..710870d0f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/kalman_filter.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/kalman_filter.cpython-37.pyc new file mode 100644 index 000000000..6df67d811 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/__pycache__/kalman_filter.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/kalman_filter.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/kalman_filter.py new file mode 100644 index 000000000..e3d42ea14 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/motion/kalman_filter.py @@ -0,0 +1,270 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/nwojke/deep_sort/blob/master/deep_sort/kalman_filter.py +""" + +import numpy as np +import scipy.linalg +from ppdet.core.workspace import register, serializable + +__all__ = ['KalmanFilter'] +""" +Table for the 0.95 quantile of the chi-square distribution with N degrees of +freedom (contains values for N=1, ..., 9). Taken from MATLAB/Octave's chi2inv +function and used as Mahalanobis gating threshold. +""" + +chi2inv95 = { + 1: 3.8415, + 2: 5.9915, + 3: 7.8147, + 4: 9.4877, + 5: 11.070, + 6: 12.592, + 7: 14.067, + 8: 15.507, + 9: 16.919 +} + + +@register +@serializable +class KalmanFilter(object): + """ + A simple Kalman filter for tracking bounding boxes in image space. + + The 8-dimensional state space + + x, y, a, h, vx, vy, va, vh + + contains the bounding box center position (x, y), aspect ratio a, height h, + and their respective velocities. + + Object motion follows a constant velocity model. The bounding box location + (x, y, a, h) is taken as direct observation of the state space (linear + observation model). + + """ + + def __init__(self): + ndim, dt = 4, 1. + + # Create Kalman filter model matrices. + self._motion_mat = np.eye(2 * ndim, 2 * ndim) + for i in range(ndim): + self._motion_mat[i, ndim + i] = dt + self._update_mat = np.eye(ndim, 2 * ndim) + + # Motion and observation uncertainty are chosen relative to the current + # state estimate. These weights control the amount of uncertainty in + # the model. This is a bit hacky. + self._std_weight_position = 1. / 20 + self._std_weight_velocity = 1. / 160 + + def initiate(self, measurement): + """ + Create track from unassociated measurement. + + Args: + measurement (ndarray): Bounding box coordinates (x, y, a, h) with + center position (x, y), aspect ratio a, and height h. + + Returns: + The mean vector (8 dimensional) and covariance matrix (8x8 + dimensional) of the new track. Unobserved velocities are + initialized to 0 mean. + """ + mean_pos = measurement + mean_vel = np.zeros_like(mean_pos) + mean = np.r_[mean_pos, mean_vel] + + std = [ + 2 * self._std_weight_position * measurement[3], + 2 * self._std_weight_position * measurement[3], 1e-2, + 2 * self._std_weight_position * measurement[3], + 10 * self._std_weight_velocity * measurement[3], + 10 * self._std_weight_velocity * measurement[3], 1e-5, + 10 * self._std_weight_velocity * measurement[3] + ] + covariance = np.diag(np.square(std)) + return mean, covariance + + def predict(self, mean, covariance): + """ + Run Kalman filter prediction step. + + Args: + mean (ndarray): The 8 dimensional mean vector of the object state + at the previous time step. + covariance (ndarray): The 8x8 dimensional covariance matrix of the + object state at the previous time step. + + Returns: + The mean vector and covariance matrix of the predicted state. + Unobserved velocities are initialized to 0 mean. + """ + std_pos = [ + self._std_weight_position * mean[3], self._std_weight_position * + mean[3], 1e-2, self._std_weight_position * mean[3] + ] + std_vel = [ + self._std_weight_velocity * mean[3], self._std_weight_velocity * + mean[3], 1e-5, self._std_weight_velocity * mean[3] + ] + motion_cov = np.diag(np.square(np.r_[std_pos, std_vel])) + + #mean = np.dot(self._motion_mat, mean) + mean = np.dot(mean, self._motion_mat.T) + covariance = np.linalg.multi_dot( + (self._motion_mat, covariance, self._motion_mat.T)) + motion_cov + + return mean, covariance + + def project(self, mean, covariance): + """ + Project state distribution to measurement space. + + Args + mean (ndarray): The state's mean vector (8 dimensional array). + covariance (ndarray): The state's covariance matrix (8x8 dimensional). + + Returns: + The projected mean and covariance matrix of the given state estimate. + """ + std = [ + self._std_weight_position * mean[3], self._std_weight_position * + mean[3], 1e-1, self._std_weight_position * mean[3] + ] + innovation_cov = np.diag(np.square(std)) + + mean = np.dot(self._update_mat, mean) + covariance = np.linalg.multi_dot((self._update_mat, covariance, + self._update_mat.T)) + return mean, covariance + innovation_cov + + def multi_predict(self, mean, covariance): + """ + Run Kalman filter prediction step (Vectorized version). + + Args: + mean (ndarray): The Nx8 dimensional mean matrix of the object states + at the previous time step. + covariance (ndarray): The Nx8x8 dimensional covariance matrics of the + object states at the previous time step. + + Returns: + The mean vector and covariance matrix of the predicted state. + Unobserved velocities are initialized to 0 mean. + """ + std_pos = [ + self._std_weight_position * mean[:, 3], self._std_weight_position * + mean[:, 3], 1e-2 * np.ones_like(mean[:, 3]), + self._std_weight_position * mean[:, 3] + ] + std_vel = [ + self._std_weight_velocity * mean[:, 3], self._std_weight_velocity * + mean[:, 3], 1e-5 * np.ones_like(mean[:, 3]), + self._std_weight_velocity * mean[:, 3] + ] + sqr = np.square(np.r_[std_pos, std_vel]).T + + motion_cov = [] + for i in range(len(mean)): + motion_cov.append(np.diag(sqr[i])) + motion_cov = np.asarray(motion_cov) + + mean = np.dot(mean, self._motion_mat.T) + left = np.dot(self._motion_mat, covariance).transpose((1, 0, 2)) + covariance = np.dot(left, self._motion_mat.T) + motion_cov + + return mean, covariance + + def update(self, mean, covariance, measurement): + """ + Run Kalman filter correction step. + + Args: + mean (ndarray): The predicted state's mean vector (8 dimensional). + covariance (ndarray): The state's covariance matrix (8x8 dimensional). + measurement (ndarray): The 4 dimensional measurement vector + (x, y, a, h), where (x, y) is the center position, a the aspect + ratio, and h the height of the bounding box. + + Returns: + The measurement-corrected state distribution. + """ + projected_mean, projected_cov = self.project(mean, covariance) + + chol_factor, lower = scipy.linalg.cho_factor( + projected_cov, lower=True, check_finite=False) + kalman_gain = scipy.linalg.cho_solve( + (chol_factor, lower), + np.dot(covariance, self._update_mat.T).T, + check_finite=False).T + innovation = measurement - projected_mean + + new_mean = mean + np.dot(innovation, kalman_gain.T) + new_covariance = covariance - np.linalg.multi_dot( + (kalman_gain, projected_cov, kalman_gain.T)) + return new_mean, new_covariance + + def gating_distance(self, + mean, + covariance, + measurements, + only_position=False, + metric='maha'): + """ + Compute gating distance between state distribution and measurements. + A suitable distance threshold can be obtained from `chi2inv95`. If + `only_position` is False, the chi-square distribution has 4 degrees of + freedom, otherwise 2. + + Args: + mean (ndarray): Mean vector over the state distribution (8 + dimensional). + covariance (ndarray): Covariance of the state distribution (8x8 + dimensional). + measurements (ndarray): An Nx4 dimensional matrix of N measurements, + each in format (x, y, a, h) where (x, y) is the bounding box center + position, a the aspect ratio, and h the height. + only_position (Optional[bool]): If True, distance computation is + done with respect to the bounding box center position only. + metric (str): Metric type, 'gaussian' or 'maha'. + + Returns + An array of length N, where the i-th element contains the squared + Mahalanobis distance between (mean, covariance) and `measurements[i]`. + """ + mean, covariance = self.project(mean, covariance) + if only_position: + mean, covariance = mean[:2], covariance[:2, :2] + measurements = measurements[:, :2] + + d = measurements - mean + if metric == 'gaussian': + return np.sum(d * d, axis=1) + elif metric == 'maha': + cholesky_factor = np.linalg.cholesky(covariance) + z = scipy.linalg.solve_triangular( + cholesky_factor, + d.T, + lower=True, + check_finite=False, + overwrite_b=True) + squared_maha = np.sum(z * z, axis=0) + return squared_maha + else: + raise ValueError('invalid distance metric') diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__init__.py new file mode 100644 index 000000000..b74593b41 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__init__.py @@ -0,0 +1,23 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import base_jde_tracker +from . import base_sde_tracker +from . import jde_tracker +from . import deepsort_tracker + +from .base_jde_tracker import * +from .base_sde_tracker import * +from .jde_tracker import * +from .deepsort_tracker import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..23093785a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_jde_tracker.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_jde_tracker.cpython-37.pyc new file mode 100644 index 000000000..9faefa723 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_jde_tracker.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_sde_tracker.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_sde_tracker.cpython-37.pyc new file mode 100644 index 000000000..7a16be8f4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/base_sde_tracker.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/deepsort_tracker.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/deepsort_tracker.cpython-37.pyc new file mode 100644 index 000000000..ab2dffa68 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/deepsort_tracker.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/jde_tracker.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/jde_tracker.cpython-37.pyc new file mode 100644 index 000000000..8c3fe9ad6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/__pycache__/jde_tracker.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_jde_tracker.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_jde_tracker.py new file mode 100644 index 000000000..8e2ef38bc --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_jde_tracker.py @@ -0,0 +1,297 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/Zhongdao/Towards-Realtime-MOT/blob/master/tracker/multitracker.py +""" + +import numpy as np +from collections import defaultdict +from collections import deque, OrderedDict +from ..matching import jde_matching as matching +from ppdet.core.workspace import register, serializable +import warnings +warnings.filterwarnings("ignore") + +__all__ = [ + 'TrackState', + 'BaseTrack', + 'STrack', + 'joint_stracks', + 'sub_stracks', + 'remove_duplicate_stracks', +] + + +class TrackState(object): + New = 0 + Tracked = 1 + Lost = 2 + Removed = 3 + + +@register +@serializable +class BaseTrack(object): + _count_dict = defaultdict(int) # support single class and multi classes + + track_id = 0 + is_activated = False + state = TrackState.New + + history = OrderedDict() + features = [] + curr_feature = None + score = 0 + start_frame = 0 + frame_id = 0 + time_since_update = 0 + + # multi-camera + location = (np.inf, np.inf) + + @property + def end_frame(self): + return self.frame_id + + @staticmethod + def next_id(cls_id): + BaseTrack._count_dict[cls_id] += 1 + return BaseTrack._count_dict[cls_id] + + # @even: reset track id + @staticmethod + def init_count(num_classes): + """ + Initiate _count for all object classes + :param num_classes: + """ + for cls_id in range(num_classes): + BaseTrack._count_dict[cls_id] = 0 + + @staticmethod + def reset_track_count(cls_id): + BaseTrack._count_dict[cls_id] = 0 + + def activate(self, *args): + raise NotImplementedError + + def predict(self): + raise NotImplementedError + + def update(self, *args, **kwargs): + raise NotImplementedError + + def mark_lost(self): + self.state = TrackState.Lost + + def mark_removed(self): + self.state = TrackState.Removed + + +@register +@serializable +class STrack(BaseTrack): + def __init__(self, + tlwh, + score, + temp_feat, + num_classes, + cls_id, + buff_size=30): + # object class id + self.cls_id = cls_id + # wait activate + self._tlwh = np.asarray(tlwh, dtype=np.float) + self.kalman_filter = None + self.mean, self.covariance = None, None + self.is_activated = False + + self.score = score + self.track_len = 0 + + self.smooth_feat = None + self.update_features(temp_feat) + self.features = deque([], maxlen=buff_size) + self.alpha = 0.9 + + def update_features(self, feat): + # L2 normalizing + feat /= np.linalg.norm(feat) + self.curr_feat = feat + if self.smooth_feat is None: + self.smooth_feat = feat + else: + self.smooth_feat = self.alpha * self.smooth_feat + (1.0 - self.alpha + ) * feat + self.features.append(feat) + self.smooth_feat /= np.linalg.norm(self.smooth_feat) + + def predict(self): + mean_state = self.mean.copy() + if self.state != TrackState.Tracked: + mean_state[7] = 0 + self.mean, self.covariance = self.kalman_filter.predict(mean_state, + self.covariance) + + @staticmethod + def multi_predict(tracks, kalman_filter): + if len(tracks) > 0: + multi_mean = np.asarray([track.mean.copy() for track in tracks]) + multi_covariance = np.asarray( + [track.covariance for track in tracks]) + for i, st in enumerate(tracks): + if st.state != TrackState.Tracked: + multi_mean[i][7] = 0 + multi_mean, multi_covariance = kalman_filter.multi_predict( + multi_mean, multi_covariance) + for i, (mean, cov) in enumerate(zip(multi_mean, multi_covariance)): + tracks[i].mean = mean + tracks[i].covariance = cov + + def reset_track_id(self): + self.reset_track_count(self.cls_id) + + def activate(self, kalman_filter, frame_id): + """Start a new track""" + self.kalman_filter = kalman_filter + # update track id for the object class + self.track_id = self.next_id(self.cls_id) + self.mean, self.covariance = self.kalman_filter.initiate( + self.tlwh_to_xyah(self._tlwh)) + + self.track_len = 0 + self.state = TrackState.Tracked # set flag 'tracked' + + if frame_id == 1: # to record the first frame's detection result + self.is_activated = True + + self.frame_id = frame_id + self.start_frame = frame_id + + def re_activate(self, new_track, frame_id, new_id=False): + self.mean, self.covariance = self.kalman_filter.update( + self.mean, self.covariance, self.tlwh_to_xyah(new_track.tlwh)) + self.update_features(new_track.curr_feat) + self.track_len = 0 + self.state = TrackState.Tracked + self.is_activated = True + self.frame_id = frame_id + if new_id: # update track id for the object class + self.track_id = self.next_id(self.cls_id) + + def update(self, new_track, frame_id, update_feature=True): + self.frame_id = frame_id + self.track_len += 1 + + new_tlwh = new_track.tlwh + self.mean, self.covariance = self.kalman_filter.update( + self.mean, self.covariance, self.tlwh_to_xyah(new_tlwh)) + self.state = TrackState.Tracked # set flag 'tracked' + self.is_activated = True # set flag 'activated' + + self.score = new_track.score + if update_feature: + self.update_features(new_track.curr_feat) + + @property + def tlwh(self): + """Get current position in bounding box format `(top left x, top left y, + width, height)`. + """ + if self.mean is None: + return self._tlwh.copy() + + ret = self.mean[:4].copy() + ret[2] *= ret[3] + ret[:2] -= ret[2:] / 2 + return ret + + @property + def tlbr(self): + """Convert bounding box to format `(min x, min y, max x, max y)`, i.e., + `(top left, bottom right)`. + """ + ret = self.tlwh.copy() + ret[2:] += ret[:2] + return ret + + @staticmethod + def tlwh_to_xyah(tlwh): + """Convert bounding box to format `(center x, center y, aspect ratio, + height)`, where the aspect ratio is `width / height`. + """ + ret = np.asarray(tlwh).copy() + ret[:2] += ret[2:] / 2 + ret[2] /= ret[3] + return ret + + def to_xyah(self): + return self.tlwh_to_xyah(self.tlwh) + + @staticmethod + def tlbr_to_tlwh(tlbr): + ret = np.asarray(tlbr).copy() + ret[2:] -= ret[:2] + return ret + + @staticmethod + def tlwh_to_tlbr(tlwh): + ret = np.asarray(tlwh).copy() + ret[2:] += ret[:2] + return ret + + def __repr__(self): + return 'OT_({}-{})_({}-{})'.format(self.cls_id, self.track_id, + self.start_frame, self.end_frame) + + +def joint_stracks(tlista, tlistb): + exists = {} + res = [] + for t in tlista: + exists[t.track_id] = 1 + res.append(t) + for t in tlistb: + tid = t.track_id + if not exists.get(tid, 0): + exists[tid] = 1 + res.append(t) + return res + + +def sub_stracks(tlista, tlistb): + stracks = {} + for t in tlista: + stracks[t.track_id] = t + for t in tlistb: + tid = t.track_id + if stracks.get(tid, 0): + del stracks[tid] + return list(stracks.values()) + + +def remove_duplicate_stracks(stracksa, stracksb): + pdist = matching.iou_distance(stracksa, stracksb) + pairs = np.where(pdist < 0.15) + dupa, dupb = list(), list() + for p, q in zip(*pairs): + timep = stracksa[p].frame_id - stracksa[p].start_frame + timeq = stracksb[q].frame_id - stracksb[q].start_frame + if timep > timeq: + dupb.append(q) + else: + dupa.append(p) + resa = [t for i, t in enumerate(stracksa) if not i in dupa] + resb = [t for i, t in enumerate(stracksb) if not i in dupb] + return resa, resb diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_sde_tracker.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_sde_tracker.py new file mode 100644 index 000000000..accc2016f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/base_sde_tracker.py @@ -0,0 +1,156 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/nwojke/deep_sort/blob/master/deep_sort/track.py +""" + +import datetime +from ppdet.core.workspace import register, serializable + +__all__ = ['TrackState', 'Track'] + + +class TrackState(object): + """ + Enumeration type for the single target track state. Newly created tracks are + classified as `tentative` until enough evidence has been collected. Then, + the track state is changed to `confirmed`. Tracks that are no longer alive + are classified as `deleted` to mark them for removal from the set of active + tracks. + """ + Tentative = 1 + Confirmed = 2 + Deleted = 3 + + +@register +@serializable +class Track(object): + """ + A single target track with state space `(x, y, a, h)` and associated + velocities, where `(x, y)` is the center of the bounding box, `a` is the + aspect ratio and `h` is the height. + + Args: + mean (ndarray): Mean vector of the initial state distribution. + covariance (ndarray): Covariance matrix of the initial state distribution. + track_id (int): A unique track identifier. + n_init (int): Number of consecutive detections before the track is confirmed. + The track state is set to `Deleted` if a miss occurs within the first + `n_init` frames. + max_age (int): The maximum number of consecutive misses before the track + state is set to `Deleted`. + cls_id (int): The category id of the tracked box. + score (float): The confidence score of the tracked box. + feature (Optional[ndarray]): Feature vector of the detection this track + originates from. If not None, this feature is added to the `features` cache. + + Attributes: + hits (int): Total number of measurement updates. + age (int): Total number of frames since first occurance. + time_since_update (int): Total number of frames since last measurement + update. + state (TrackState): The current track state. + features (List[ndarray]): A cache of features. On each measurement update, + the associated feature vector is added to this list. + """ + + def __init__(self, + mean, + covariance, + track_id, + n_init, + max_age, + cls_id, + score, + feature=None): + self.mean = mean + self.covariance = covariance + self.track_id = track_id + self.hits = 1 + self.age = 1 + self.time_since_update = 0 + self.cls_id = cls_id + self.score = score + self.start_time = datetime.datetime.now() + + self.state = TrackState.Tentative + self.features = [] + self.feat = feature + if feature is not None: + self.features.append(feature) + + self._n_init = n_init + self._max_age = max_age + + def to_tlwh(self): + """Get position in format `(top left x, top left y, width, height)`.""" + ret = self.mean[:4].copy() + ret[2] *= ret[3] + ret[:2] -= ret[2:] / 2 + return ret + + def to_tlbr(self): + """Get position in bounding box format `(min x, miny, max x, max y)`.""" + ret = self.to_tlwh() + ret[2:] = ret[:2] + ret[2:] + return ret + + def predict(self, kalman_filter): + """ + Propagate the state distribution to the current time step using a Kalman + filter prediction step. + """ + self.mean, self.covariance = kalman_filter.predict(self.mean, + self.covariance) + self.age += 1 + self.time_since_update += 1 + + def update(self, kalman_filter, detection): + """ + Perform Kalman filter measurement update step and update the associated + detection feature cache. + """ + self.mean, self.covariance = kalman_filter.update(self.mean, + self.covariance, + detection.to_xyah()) + self.features.append(detection.feature) + self.feat = detection.feature + self.cls_id = detection.cls_id + self.score = detection.score + + self.hits += 1 + self.time_since_update = 0 + if self.state == TrackState.Tentative and self.hits >= self._n_init: + self.state = TrackState.Confirmed + + def mark_missed(self): + """Mark this track as missed (no association at the current time step). + """ + if self.state == TrackState.Tentative: + self.state = TrackState.Deleted + elif self.time_since_update > self._max_age: + self.state = TrackState.Deleted + + def is_tentative(self): + """Returns True if this track is tentative (unconfirmed).""" + return self.state == TrackState.Tentative + + def is_confirmed(self): + """Returns True if this track is confirmed.""" + return self.state == TrackState.Confirmed + + def is_deleted(self): + """Returns True if this track is dead and should be deleted.""" + return self.state == TrackState.Deleted diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/deepsort_tracker.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/deepsort_tracker.py new file mode 100644 index 000000000..ef38a67f9 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/deepsort_tracker.py @@ -0,0 +1,188 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/nwojke/deep_sort/blob/master/deep_sort/tracker.py +""" + +import numpy as np + +from ..motion import KalmanFilter +from ..matching.deepsort_matching import NearestNeighborDistanceMetric +from ..matching.deepsort_matching import iou_cost, min_cost_matching, matching_cascade, gate_cost_matrix +from .base_sde_tracker import Track +from ..utils import Detection + +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['DeepSORTTracker'] + + +@register +@serializable +class DeepSORTTracker(object): + """ + DeepSORT tracker + + Args: + input_size (list): input feature map size to reid model, [h, w] format, + [64, 192] as default. + min_box_area (int): min box area to filter out low quality boxes + vertical_ratio (float): w/h, the vertical ratio of the bbox to filter + bad results, set 1.6 default for pedestrian tracking. If set <=0 + means no need to filter bboxes. + budget (int): If not None, fix samples per class to at most this number. + Removes the oldest samples when the budget is reached. + max_age (int): maximum number of missed misses before a track is deleted + n_init (float): Number of frames that a track remains in initialization + phase. Number of consecutive detections before the track is confirmed. + The track state is set to `Deleted` if a miss occurs within the first + `n_init` frames. + metric_type (str): either "euclidean" or "cosine", the distance metric + used for measurement to track association. + matching_threshold (float): samples with larger distance are + considered an invalid match. + max_iou_distance (float): max iou distance threshold + motion (object): KalmanFilter instance + """ + + def __init__(self, + input_size=[64, 192], + min_box_area=0, + vertical_ratio=-1, + budget=100, + max_age=70, + n_init=3, + metric_type='cosine', + matching_threshold=0.2, + max_iou_distance=0.9, + motion='KalmanFilter'): + self.input_size = input_size + self.min_box_area = min_box_area + self.vertical_ratio = vertical_ratio + self.max_age = max_age + self.n_init = n_init + self.metric = NearestNeighborDistanceMetric(metric_type, + matching_threshold, budget) + self.max_iou_distance = max_iou_distance + if motion == 'KalmanFilter': + self.motion = KalmanFilter() + + self.tracks = [] + self._next_id = 1 + + def predict(self): + """ + Propagate track state distributions one time step forward. + This function should be called once every time step, before `update`. + """ + for track in self.tracks: + track.predict(self.motion) + + def update(self, pred_dets, pred_embs): + """ + Perform measurement update and track management. + Args: + pred_dets (np.array): Detection results of the image, the shape is + [N, 6], means 'x0, y0, x1, y1, score, cls_id'. + pred_embs (np.array): Embedding results of the image, the shape is + [N, 128], usually pred_embs.shape[1] is a multiple of 128. + """ + pred_tlwhs = pred_dets[:, :4] + pred_scores = pred_dets[:, 4:5] + pred_cls_ids = pred_dets[:, 5:] + + detections = [ + Detection(tlwh, score, feat, cls_id) + for tlwh, score, feat, cls_id in zip(pred_tlwhs, pred_scores, + pred_embs, pred_cls_ids) + ] + + # Run matching cascade. + matches, unmatched_tracks, unmatched_detections = \ + self._match(detections) + + # Update track set. + for track_idx, detection_idx in matches: + self.tracks[track_idx].update(self.motion, + detections[detection_idx]) + for track_idx in unmatched_tracks: + self.tracks[track_idx].mark_missed() + for detection_idx in unmatched_detections: + self._initiate_track(detections[detection_idx]) + self.tracks = [t for t in self.tracks if not t.is_deleted()] + + # Update distance metric. + active_targets = [t.track_id for t in self.tracks if t.is_confirmed()] + features, targets = [], [] + for track in self.tracks: + if not track.is_confirmed(): + continue + features += track.features + targets += [track.track_id for _ in track.features] + track.features = [] + self.metric.partial_fit( + np.asarray(features), np.asarray(targets), active_targets) + output_stracks = self.tracks + return output_stracks + + def _match(self, detections): + def gated_metric(tracks, dets, track_indices, detection_indices): + features = np.array([dets[i].feature for i in detection_indices]) + targets = np.array([tracks[i].track_id for i in track_indices]) + cost_matrix = self.metric.distance(features, targets) + cost_matrix = gate_cost_matrix(self.motion, cost_matrix, tracks, + dets, track_indices, + detection_indices) + return cost_matrix + + # Split track set into confirmed and unconfirmed tracks. + confirmed_tracks = [ + i for i, t in enumerate(self.tracks) if t.is_confirmed() + ] + unconfirmed_tracks = [ + i for i, t in enumerate(self.tracks) if not t.is_confirmed() + ] + + # Associate confirmed tracks using appearance features. + matches_a, unmatched_tracks_a, unmatched_detections = \ + matching_cascade( + gated_metric, self.metric.matching_threshold, self.max_age, + self.tracks, detections, confirmed_tracks) + + # Associate remaining tracks together with unconfirmed tracks using IOU. + iou_track_candidates = unconfirmed_tracks + [ + k for k in unmatched_tracks_a + if self.tracks[k].time_since_update == 1 + ] + unmatched_tracks_a = [ + k for k in unmatched_tracks_a + if self.tracks[k].time_since_update != 1 + ] + matches_b, unmatched_tracks_b, unmatched_detections = \ + min_cost_matching( + iou_cost, self.max_iou_distance, self.tracks, + detections, iou_track_candidates, unmatched_detections) + + matches = matches_a + matches_b + unmatched_tracks = list(set(unmatched_tracks_a + unmatched_tracks_b)) + return matches, unmatched_tracks, unmatched_detections + + def _initiate_track(self, detection): + mean, covariance = self.motion.initiate(detection.to_xyah()) + self.tracks.append( + Track(mean, covariance, self._next_id, self.n_init, self.max_age, + detection.cls_id, detection.score, detection.feature)) + self._next_id += 1 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/jde_tracker.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/jde_tracker.py new file mode 100644 index 000000000..af5411a26 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/tracker/jde_tracker.py @@ -0,0 +1,273 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" +This code is based on https://github.com/Zhongdao/Towards-Realtime-MOT/blob/master/tracker/multitracker.py +""" + +import numpy as np +from collections import defaultdict + +from ..matching import jde_matching as matching +from ..motion import KalmanFilter +from .base_jde_tracker import TrackState, STrack +from .base_jde_tracker import joint_stracks, sub_stracks, remove_duplicate_stracks + +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['JDETracker'] + + +@register +@serializable +class JDETracker(object): + __shared__ = ['num_classes'] + """ + JDE tracker, support single class and multi classes + + Args: + num_classes (int): the number of classes + det_thresh (float): threshold of detection score + track_buffer (int): buffer for tracker + min_box_area (int): min box area to filter out low quality boxes + vertical_ratio (float): w/h, the vertical ratio of the bbox to filter + bad results. If set <0 means no need to filter bboxes锛寀sually set + 1.6 for pedestrian tracking. + tracked_thresh (float): linear assignment threshold of tracked + stracks and detections + r_tracked_thresh (float): linear assignment threshold of + tracked stracks and unmatched detections + unconfirmed_thresh (float): linear assignment threshold of + unconfirmed stracks and unmatched detections + motion (str): motion model, KalmanFilter as default + conf_thres (float): confidence threshold for tracking + metric_type (str): either "euclidean" or "cosine", the distance metric + used for measurement to track association. + """ + + def __init__(self, + num_classes=1, + det_thresh=0.3, + track_buffer=30, + min_box_area=200, + vertical_ratio=1.6, + tracked_thresh=0.7, + r_tracked_thresh=0.5, + unconfirmed_thresh=0.7, + motion='KalmanFilter', + conf_thres=0, + metric_type='euclidean'): + self.num_classes = num_classes + self.det_thresh = det_thresh + self.track_buffer = track_buffer + self.min_box_area = min_box_area + self.vertical_ratio = vertical_ratio + + self.tracked_thresh = tracked_thresh + self.r_tracked_thresh = r_tracked_thresh + self.unconfirmed_thresh = unconfirmed_thresh + if motion == 'KalmanFilter': + self.motion = KalmanFilter() + self.conf_thres = conf_thres + self.metric_type = metric_type + + self.frame_id = 0 + self.tracked_tracks_dict = defaultdict(list) # dict(list[STrack]) + self.lost_tracks_dict = defaultdict(list) # dict(list[STrack]) + self.removed_tracks_dict = defaultdict(list) # dict(list[STrack]) + + self.max_time_lost = 0 + # max_time_lost will be calculated: int(frame_rate / 30.0 * track_buffer) + + def update(self, pred_dets, pred_embs): + """ + Processes the image frame and finds bounding box(detections). + Associates the detection with corresponding tracklets and also handles + lost, removed, refound and active tracklets. + + Args: + pred_dets (np.array): Detection results of the image, the shape is + [N, 6], means 'x0, y0, x1, y1, score, cls_id'. + pred_embs (np.array): Embedding results of the image, the shape is + [N, 128] or [N, 512]. + + Return: + output_stracks_dict (dict(list)): The list contains information + regarding the online_tracklets for the recieved image tensor. + """ + self.frame_id += 1 + if self.frame_id == 1: + STrack.init_count(self.num_classes) + activated_tracks_dict = defaultdict(list) + refined_tracks_dict = defaultdict(list) + lost_tracks_dict = defaultdict(list) + removed_tracks_dict = defaultdict(list) + output_tracks_dict = defaultdict(list) + + pred_dets_dict = defaultdict(list) + pred_embs_dict = defaultdict(list) + + # unify single and multi classes detection and embedding results + for cls_id in range(self.num_classes): + cls_idx = (pred_dets[:, 5:] == cls_id).squeeze(-1) + pred_dets_dict[cls_id] = pred_dets[cls_idx] + pred_embs_dict[cls_id] = pred_embs[cls_idx] + + for cls_id in range(self.num_classes): + """ Step 1: Get detections by class""" + pred_dets_cls = pred_dets_dict[cls_id] + pred_embs_cls = pred_embs_dict[cls_id] + remain_inds = (pred_dets_cls[:, 4:5] > self.conf_thres).squeeze(-1) + if remain_inds.sum() > 0: + pred_dets_cls = pred_dets_cls[remain_inds] + pred_embs_cls = pred_embs_cls[remain_inds] + detections = [ + STrack( + STrack.tlbr_to_tlwh(tlbrs[:4]), tlbrs[4], f, + self.num_classes, cls_id, 30) + for (tlbrs, f) in zip(pred_dets_cls, pred_embs_cls) + ] + else: + detections = [] + ''' Add newly detected tracklets to tracked_stracks''' + unconfirmed_dict = defaultdict(list) + tracked_tracks_dict = defaultdict(list) + for track in self.tracked_tracks_dict[cls_id]: + if not track.is_activated: + # previous tracks which are not active in the current frame are added in unconfirmed list + unconfirmed_dict[cls_id].append(track) + else: + # Active tracks are added to the local list 'tracked_stracks' + tracked_tracks_dict[cls_id].append(track) + """ Step 2: First association, with embedding""" + # building tracking pool for the current frame + track_pool_dict = defaultdict(list) + track_pool_dict[cls_id] = joint_stracks( + tracked_tracks_dict[cls_id], self.lost_tracks_dict[cls_id]) + + # Predict the current location with KalmanFilter + STrack.multi_predict(track_pool_dict[cls_id], self.motion) + + dists = matching.embedding_distance( + track_pool_dict[cls_id], detections, metric=self.metric_type) + dists = matching.fuse_motion(self.motion, dists, + track_pool_dict[cls_id], detections) + matches, u_track, u_detection = matching.linear_assignment( + dists, thresh=self.tracked_thresh) + + for i_tracked, idet in matches: + # i_tracked is the id of the track and idet is the detection + track = track_pool_dict[cls_id][i_tracked] + det = detections[idet] + if track.state == TrackState.Tracked: + # If the track is active, add the detection to the track + track.update(detections[idet], self.frame_id) + activated_tracks_dict[cls_id].append(track) + else: + # We have obtained a detection from a track which is not active, + # hence put the track in refind_stracks list + track.re_activate(det, self.frame_id, new_id=False) + refined_tracks_dict[cls_id].append(track) + + # None of the steps below happen if there are no undetected tracks. + """ Step 3: Second association, with IOU""" + detections = [detections[i] for i in u_detection] + r_tracked_stracks = [] + for i in u_track: + if track_pool_dict[cls_id][i].state == TrackState.Tracked: + r_tracked_stracks.append(track_pool_dict[cls_id][i]) + + dists = matching.iou_distance(r_tracked_stracks, detections) + matches, u_track, u_detection = matching.linear_assignment( + dists, thresh=self.r_tracked_thresh) + + for i_tracked, idet in matches: + track = r_tracked_stracks[i_tracked] + det = detections[idet] + if track.state == TrackState.Tracked: + track.update(det, self.frame_id) + activated_tracks_dict[cls_id].append(track) + else: + track.re_activate(det, self.frame_id, new_id=False) + refined_tracks_dict[cls_id].append(track) + + for it in u_track: + track = r_tracked_stracks[it] + if not track.state == TrackState.Lost: + track.mark_lost() + lost_tracks_dict[cls_id].append(track) + '''Deal with unconfirmed tracks, usually tracks with only one beginning frame''' + detections = [detections[i] for i in u_detection] + dists = matching.iou_distance(unconfirmed_dict[cls_id], detections) + matches, u_unconfirmed, u_detection = matching.linear_assignment( + dists, thresh=self.unconfirmed_thresh) + for i_tracked, idet in matches: + unconfirmed_dict[cls_id][i_tracked].update(detections[idet], + self.frame_id) + activated_tracks_dict[cls_id].append(unconfirmed_dict[cls_id][ + i_tracked]) + for it in u_unconfirmed: + track = unconfirmed_dict[cls_id][it] + track.mark_removed() + removed_tracks_dict[cls_id].append(track) + """ Step 4: Init new stracks""" + for inew in u_detection: + track = detections[inew] + if track.score < self.det_thresh: + continue + track.activate(self.motion, self.frame_id) + activated_tracks_dict[cls_id].append(track) + """ Step 5: Update state""" + for track in self.lost_tracks_dict[cls_id]: + if self.frame_id - track.end_frame > self.max_time_lost: + track.mark_removed() + removed_tracks_dict[cls_id].append(track) + + self.tracked_tracks_dict[cls_id] = [ + t for t in self.tracked_tracks_dict[cls_id] + if t.state == TrackState.Tracked + ] + self.tracked_tracks_dict[cls_id] = joint_stracks( + self.tracked_tracks_dict[cls_id], activated_tracks_dict[cls_id]) + self.tracked_tracks_dict[cls_id] = joint_stracks( + self.tracked_tracks_dict[cls_id], refined_tracks_dict[cls_id]) + self.lost_tracks_dict[cls_id] = sub_stracks( + self.lost_tracks_dict[cls_id], self.tracked_tracks_dict[cls_id]) + self.lost_tracks_dict[cls_id].extend(lost_tracks_dict[cls_id]) + self.lost_tracks_dict[cls_id] = sub_stracks( + self.lost_tracks_dict[cls_id], self.removed_tracks_dict[cls_id]) + self.removed_tracks_dict[cls_id].extend(removed_tracks_dict[cls_id]) + self.tracked_tracks_dict[cls_id], self.lost_tracks_dict[ + cls_id] = remove_duplicate_stracks( + self.tracked_tracks_dict[cls_id], + self.lost_tracks_dict[cls_id]) + + # get scores of lost tracks + output_tracks_dict[cls_id] = [ + track for track in self.tracked_tracks_dict[cls_id] + if track.is_activated + ] + + logger.debug('===========Frame {}=========='.format(self.frame_id)) + logger.debug('Activated: {}'.format( + [track.track_id for track in activated_tracks_dict[cls_id]])) + logger.debug('Refind: {}'.format( + [track.track_id for track in refined_tracks_dict[cls_id]])) + logger.debug('Lost: {}'.format( + [track.track_id for track in lost_tracks_dict[cls_id]])) + logger.debug('Removed: {}'.format( + [track.track_id for track in removed_tracks_dict[cls_id]])) + + return output_tracks_dict diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/utils.py new file mode 100644 index 000000000..b3657d257 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/utils.py @@ -0,0 +1,262 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import cv2 +import time +import numpy as np +from .visualization import plot_tracking_dict, plot_tracking + +__all__ = [ + 'MOTTimer', + 'Detection', + 'write_mot_results', + 'save_vis_results', + 'load_det_results', + 'preprocess_reid', + 'get_crops', + 'clip_box', + 'scale_coords', +] + + +class MOTTimer(object): + """ + This class used to compute and print the current FPS while evaling. + """ + + def __init__(self): + self.total_time = 0. + self.calls = 0 + self.start_time = 0. + self.diff = 0. + self.average_time = 0. + self.duration = 0. + + def tic(self): + # using time.time instead of time.clock because time time.clock + # does not normalize for multithreading + self.start_time = time.time() + + def toc(self, average=True): + self.diff = time.time() - self.start_time + self.total_time += self.diff + self.calls += 1 + self.average_time = self.total_time / self.calls + if average: + self.duration = self.average_time + else: + self.duration = self.diff + return self.duration + + def clear(self): + self.total_time = 0. + self.calls = 0 + self.start_time = 0. + self.diff = 0. + self.average_time = 0. + self.duration = 0. + + +class Detection(object): + """ + This class represents a bounding box detection in a single image. + + Args: + tlwh (Tensor): Bounding box in format `(top left x, top left y, + width, height)`. + score (Tensor): Bounding box confidence score. + feature (Tensor): A feature vector that describes the object + contained in this image. + cls_id (Tensor): Bounding box category id. + """ + + def __init__(self, tlwh, score, feature, cls_id): + self.tlwh = np.asarray(tlwh, dtype=np.float32) + self.score = float(score) + self.feature = np.asarray(feature, dtype=np.float32) + self.cls_id = int(cls_id) + + def to_tlbr(self): + """ + Convert bounding box to format `(min x, min y, max x, max y)`, i.e., + `(top left, bottom right)`. + """ + ret = self.tlwh.copy() + ret[2:] += ret[:2] + return ret + + def to_xyah(self): + """ + Convert bounding box to format `(center x, center y, aspect ratio, + height)`, where the aspect ratio is `width / height`. + """ + ret = self.tlwh.copy() + ret[:2] += ret[2:] / 2 + ret[2] /= ret[3] + return ret + + +def write_mot_results(filename, results, data_type='mot', num_classes=1): + # support single and multi classes + if data_type in ['mot', 'mcmot']: + save_format = '{frame},{id},{x1},{y1},{w},{h},{score},{cls_id},-1,-1\n' + elif data_type == 'kitti': + save_format = '{frame} {id} car 0 0 -10 {x1} {y1} {x2} {y2} -10 -10 -10 -1000 -1000 -1000 -10\n' + else: + raise ValueError(data_type) + + f = open(filename, 'w') + for cls_id in range(num_classes): + for frame_id, tlwhs, tscores, track_ids in results[cls_id]: + if data_type == 'kitti': + frame_id -= 1 + for tlwh, score, track_id in zip(tlwhs, tscores, track_ids): + if track_id < 0: continue + if data_type == 'mot': + cls_id = -1 + + x1, y1, w, h = tlwh + x2, y2 = x1 + w, y1 + h + line = save_format.format( + frame=frame_id, + id=track_id, + x1=x1, + y1=y1, + x2=x2, + y2=y2, + w=w, + h=h, + score=score, + cls_id=cls_id) + f.write(line) + print('MOT results save in {}'.format(filename)) + + +def save_vis_results(data, + frame_id, + online_ids, + online_tlwhs, + online_scores, + average_time, + show_image, + save_dir, + num_classes=1): + if show_image or save_dir is not None: + assert 'ori_image' in data + img0 = data['ori_image'].numpy()[0] + if online_ids is None: + online_im = img0 + else: + if isinstance(online_tlwhs, dict): + online_im = plot_tracking_dict( + img0, + num_classes, + online_tlwhs, + online_ids, + online_scores, + frame_id=frame_id, + fps=1. / average_time) + else: + online_im = plot_tracking( + img0, + online_tlwhs, + online_ids, + online_scores, + frame_id=frame_id, + fps=1. / average_time) + if show_image: + cv2.imshow('online_im', online_im) + if save_dir is not None: + cv2.imwrite( + os.path.join(save_dir, '{:05d}.jpg'.format(frame_id)), online_im) + + +def load_det_results(det_file, num_frames): + assert os.path.exists(det_file) and os.path.isfile(det_file), \ + '{} is not exist or not a file.'.format(det_file) + labels = np.loadtxt(det_file, dtype='float32', delimiter=',') + assert labels.shape[1] == 7, \ + "Each line of {} should have 7 items: '[frame_id],[x0],[y0],[w],[h],[score],[class_id]'.".format(det_file) + results_list = [] + for frame_i in range(num_frames): + results = {'bbox': [], 'score': [], 'cls_id': []} + lables_with_frame = labels[labels[:, 0] == frame_i + 1] + # each line of lables_with_frame: + # [frame_id],[x0],[y0],[w],[h],[score],[class_id] + for l in lables_with_frame: + results['bbox'].append(l[1:5]) + results['score'].append(l[5:6]) + results['cls_id'].append(l[6:7]) + results_list.append(results) + return results_list + + +def scale_coords(coords, input_shape, im_shape, scale_factor): + # Note: ratio has only one value, scale_factor[0] == scale_factor[1] + # + # This function only used for JDE YOLOv3 or other detectors with + # LetterBoxResize and JDEBBoxPostProcess, coords output from detector had + # not scaled back to the origin image. + + ratio = scale_factor[0] + pad_w = (input_shape[1] - int(im_shape[1])) / 2 + pad_h = (input_shape[0] - int(im_shape[0])) / 2 + coords[:, 0::2] -= pad_w + coords[:, 1::2] -= pad_h + coords[:, 0:4] /= ratio + coords[:, :4] = np.clip(coords[:, :4], a_min=0, a_max=coords[:, :4].max()) + return coords.round() + + +def clip_box(xyxy, ori_image_shape): + H, W = ori_image_shape + xyxy[:, 0::2] = np.clip(xyxy[:, 0::2], a_min=0, a_max=W) + xyxy[:, 1::2] = np.clip(xyxy[:, 1::2], a_min=0, a_max=H) + w = xyxy[:, 2:3] - xyxy[:, 0:1] + h = xyxy[:, 3:4] - xyxy[:, 1:2] + mask = np.logical_and(h > 0, w > 0) + keep_idx = np.nonzero(mask) + return xyxy[keep_idx[0]], keep_idx + + +def get_crops(xyxy, ori_img, w, h): + crops = [] + xyxy = xyxy.astype(np.int64) + ori_img = ori_img.numpy() + ori_img = np.squeeze(ori_img, axis=0).transpose(1, 0, 2) # [h,w,3]->[w,h,3] + for i, bbox in enumerate(xyxy): + crop = ori_img[bbox[0]:bbox[2], bbox[1]:bbox[3], :] + crops.append(crop) + crops = preprocess_reid(crops, w, h) + return crops + + +def preprocess_reid(imgs, + w=64, + h=192, + mean=[0.485, 0.456, 0.406], + std=[0.229, 0.224, 0.225]): + im_batch = [] + for img in imgs: + img = cv2.resize(img, (w, h)) + img = img[:, :, ::-1].astype('float32').transpose((2, 0, 1)) / 255 + img_mean = np.array(mean).reshape((3, 1, 1)) + img_std = np.array(std).reshape((3, 1, 1)) + img -= img_mean + img /= img_std + img = np.expand_dims(img, axis=0) + im_batch.append(img) + im_batch = np.concatenate(im_batch, 0) + return im_batch diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/visualization.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/visualization.py new file mode 100644 index 000000000..6d13a2877 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/mot/visualization.py @@ -0,0 +1,146 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import cv2 +import numpy as np + + +def get_color(idx): + idx = idx * 3 + color = ((37 * idx) % 255, (17 * idx) % 255, (29 * idx) % 255) + return color + + +def plot_tracking(image, + tlwhs, + obj_ids, + scores=None, + frame_id=0, + fps=0., + ids2names=[]): + im = np.ascontiguousarray(np.copy(image)) + im_h, im_w = im.shape[:2] + + top_view = np.zeros([im_w, im_w, 3], dtype=np.uint8) + 255 + + text_scale = max(1, image.shape[1] / 1600.) + text_thickness = 2 + line_thickness = max(1, int(image.shape[1] / 500.)) + + radius = max(5, int(im_w / 140.)) + cv2.putText( + im, + 'frame: %d fps: %.2f num: %d' % (frame_id, fps, len(tlwhs)), + (0, int(15 * text_scale)), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 0, 255), + thickness=2) + + for i, tlwh in enumerate(tlwhs): + x1, y1, w, h = tlwh + intbox = tuple(map(int, (x1, y1, x1 + w, y1 + h))) + obj_id = int(obj_ids[i]) + id_text = '{}'.format(int(obj_id)) + if ids2names != []: + assert len( + ids2names) == 1, "plot_tracking only supports single classes." + id_text = '{}_'.format(ids2names[0]) + id_text + _line_thickness = 1 if obj_id <= 0 else line_thickness + color = get_color(abs(obj_id)) + cv2.rectangle( + im, intbox[0:2], intbox[2:4], color=color, thickness=line_thickness) + cv2.putText( + im, + id_text, (intbox[0], intbox[1] - 10), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 0, 255), + thickness=text_thickness) + + if scores is not None: + text = '{:.2f}'.format(float(scores[i])) + cv2.putText( + im, + text, (intbox[0], intbox[1] + 10), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 255, 255), + thickness=text_thickness) + return im + + +def plot_tracking_dict(image, + num_classes, + tlwhs_dict, + obj_ids_dict, + scores_dict, + frame_id=0, + fps=0., + ids2names=[]): + im = np.ascontiguousarray(np.copy(image)) + im_h, im_w = im.shape[:2] + + top_view = np.zeros([im_w, im_w, 3], dtype=np.uint8) + 255 + + text_scale = max(1, image.shape[1] / 1600.) + text_thickness = 2 + line_thickness = max(1, int(image.shape[1] / 500.)) + + radius = max(5, int(im_w / 140.)) + + for cls_id in range(num_classes): + tlwhs = tlwhs_dict[cls_id] + obj_ids = obj_ids_dict[cls_id] + scores = scores_dict[cls_id] + cv2.putText( + im, + 'frame: %d fps: %.2f num: %d' % (frame_id, fps, len(tlwhs)), + (0, int(15 * text_scale)), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 0, 255), + thickness=2) + + for i, tlwh in enumerate(tlwhs): + x1, y1, w, h = tlwh + intbox = tuple(map(int, (x1, y1, x1 + w, y1 + h))) + obj_id = int(obj_ids[i]) + + id_text = '{}'.format(int(obj_id)) + if ids2names != []: + id_text = '{}_{}'.format(ids2names[cls_id], id_text) + else: + id_text = 'class{}_{}'.format(cls_id, id_text) + + _line_thickness = 1 if obj_id <= 0 else line_thickness + color = get_color(abs(obj_id)) + cv2.rectangle( + im, + intbox[0:2], + intbox[2:4], + color=color, + thickness=line_thickness) + cv2.putText( + im, + id_text, (intbox[0], intbox[1] - 10), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 0, 255), + thickness=text_thickness) + + if scores is not None: + text = '{:.2f}'.format(float(scores[i])) + cv2.putText( + im, + text, (intbox[0], intbox[1] + 10), + cv2.FONT_HERSHEY_PLAIN, + text_scale, (0, 255, 255), + thickness=text_thickness) + return im diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__init__.py new file mode 100644 index 000000000..d66697caf --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__init__.py @@ -0,0 +1,30 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import fpn +from . import yolo_fpn +from . import hrfpn +from . import ttf_fpn +from . import centernet_fpn +from . import bifpn +from . import csp_pan + +from .fpn import * +from .yolo_fpn import * +from .hrfpn import * +from .ttf_fpn import * +from .centernet_fpn import * +from .blazeface_fpn import * +from .bifpn import * +from .csp_pan import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..922d2c1d6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/bifpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/bifpn.cpython-37.pyc new file mode 100644 index 000000000..57af360d2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/bifpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/blazeface_fpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/blazeface_fpn.cpython-37.pyc new file mode 100644 index 000000000..256668459 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/blazeface_fpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/centernet_fpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/centernet_fpn.cpython-37.pyc new file mode 100644 index 000000000..a1c91e7a2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/centernet_fpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/csp_pan.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/csp_pan.cpython-37.pyc new file mode 100644 index 000000000..0b14b978b Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/csp_pan.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/fpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/fpn.cpython-37.pyc new file mode 100644 index 000000000..d8b5164d3 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/fpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/hrfpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/hrfpn.cpython-37.pyc new file mode 100644 index 000000000..a9738dac7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/hrfpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/ttf_fpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/ttf_fpn.cpython-37.pyc new file mode 100644 index 000000000..f36ec7708 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/ttf_fpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/yolo_fpn.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/yolo_fpn.cpython-37.pyc new file mode 100644 index 000000000..59830319f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/__pycache__/yolo_fpn.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/bifpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/bifpn.py new file mode 100644 index 000000000..c60760893 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/bifpn.py @@ -0,0 +1,302 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Constant + +from ppdet.core.workspace import register, serializable +from ppdet.modeling.layers import ConvNormLayer +from ..shape_spec import ShapeSpec + +__all__ = ['BiFPN'] + + +class SeparableConvLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels=None, + kernel_size=3, + norm_type='bn', + norm_groups=32, + act='swish'): + super(SeparableConvLayer, self).__init__() + assert norm_type in ['bn', 'sync_bn', 'gn', None] + assert act in ['swish', 'relu', None] + + self.in_channels = in_channels + if out_channels is None: + self.out_channels = self.in_channels + self.norm_type = norm_type + self.norm_groups = norm_groups + self.depthwise_conv = nn.Conv2D( + in_channels, + in_channels, + kernel_size, + padding=kernel_size // 2, + groups=in_channels, + bias_attr=False) + self.pointwise_conv = nn.Conv2D(in_channels, self.out_channels, 1) + + # norm type + if self.norm_type == 'bn': + self.norm = nn.BatchNorm2D(self.out_channels) + elif self.norm_type == 'sync_bn': + self.norm = nn.SyncBatchNorm(self.out_channels) + elif self.norm_type == 'gn': + self.norm = nn.GroupNorm( + num_groups=self.norm_groups, num_channels=self.out_channels) + + # activation + if act == 'swish': + self.act = nn.Swish() + elif act == 'relu': + self.act = nn.ReLU() + + def forward(self, x): + if self.act is not None: + x = self.act(x) + out = self.depthwise_conv(x) + out = self.pointwise_conv(out) + if self.norm_type is not None: + out = self.norm(out) + return out + + +class BiFPNCell(nn.Layer): + def __init__(self, + channels=256, + num_levels=5, + eps=1e-5, + use_weighted_fusion=True, + kernel_size=3, + norm_type='bn', + norm_groups=32, + act='swish'): + super(BiFPNCell, self).__init__() + self.channels = channels + self.num_levels = num_levels + self.eps = eps + self.use_weighted_fusion = use_weighted_fusion + + # up + self.conv_up = nn.LayerList([ + SeparableConvLayer( + self.channels, + kernel_size=kernel_size, + norm_type=norm_type, + norm_groups=norm_groups, + act=act) for _ in range(self.num_levels - 1) + ]) + # down + self.conv_down = nn.LayerList([ + SeparableConvLayer( + self.channels, + kernel_size=kernel_size, + norm_type=norm_type, + norm_groups=norm_groups, + act=act) for _ in range(self.num_levels - 1) + ]) + + if self.use_weighted_fusion: + self.up_weights = self.create_parameter( + shape=[self.num_levels - 1, 2], + attr=ParamAttr(initializer=Constant(1.))) + self.down_weights = self.create_parameter( + shape=[self.num_levels - 1, 3], + attr=ParamAttr(initializer=Constant(1.))) + + def _feature_fusion_cell(self, + conv_layer, + lateral_feat, + sampling_feat, + route_feat=None, + weights=None): + if self.use_weighted_fusion: + weights = F.relu(weights) + weights = weights / (weights.sum() + self.eps) + if route_feat is not None: + out_feat = weights[0] * lateral_feat + \ + weights[1] * sampling_feat + \ + weights[2] * route_feat + else: + out_feat = weights[0] * lateral_feat + \ + weights[1] * sampling_feat + else: + if route_feat is not None: + out_feat = lateral_feat + sampling_feat + route_feat + else: + out_feat = lateral_feat + sampling_feat + + out_feat = conv_layer(out_feat) + return out_feat + + def forward(self, feats): + # feats: [P3 - P7] + lateral_feats = [] + + # up + up_feature = feats[-1] + for i, feature in enumerate(feats[::-1]): + if i == 0: + lateral_feats.append(feature) + else: + shape = paddle.shape(feature) + up_feature = F.interpolate( + up_feature, size=[shape[2], shape[3]]) + lateral_feature = self._feature_fusion_cell( + self.conv_up[i - 1], + feature, + up_feature, + weights=self.up_weights[i - 1] + if self.use_weighted_fusion else None) + lateral_feats.append(lateral_feature) + up_feature = lateral_feature + + out_feats = [] + # down + down_feature = lateral_feats[-1] + for i, (lateral_feature, + route_feature) in enumerate(zip(lateral_feats[::-1], feats)): + if i == 0: + out_feats.append(lateral_feature) + else: + down_feature = F.max_pool2d(down_feature, 3, 2, 1) + if i == len(feats) - 1: + route_feature = None + weights = self.down_weights[ + i - 1][:2] if self.use_weighted_fusion else None + else: + weights = self.down_weights[ + i - 1] if self.use_weighted_fusion else None + out_feature = self._feature_fusion_cell( + self.conv_down[i - 1], + lateral_feature, + down_feature, + route_feature, + weights=weights) + out_feats.append(out_feature) + down_feature = out_feature + + return out_feats + + +@register +@serializable +class BiFPN(nn.Layer): + """ + Bidirectional Feature Pyramid Network, see https://arxiv.org/abs/1911.09070 + + Args: + in_channels (list[int]): input channels of each level which can be + derived from the output shape of backbone by from_config. + out_channel (int): output channel of each level. + num_extra_levels (int): the number of extra stages added to the last level. + default: 2 + fpn_strides (List): The stride of each level. + num_stacks (int): the number of stacks for BiFPN, default: 1. + use_weighted_fusion (bool): use weighted feature fusion in BiFPN, default: True. + norm_type (string|None): the normalization type in BiFPN module. If + norm_type is None, norm will not be used after conv and if + norm_type is string, bn, gn, sync_bn are available. default: bn. + norm_groups (int): if you use gn, set this param. + act (string|None): the activation function of BiFPN. + """ + + def __init__(self, + in_channels=(512, 1024, 2048), + out_channel=256, + num_extra_levels=2, + fpn_strides=[8, 16, 32, 64, 128], + num_stacks=1, + use_weighted_fusion=True, + norm_type='bn', + norm_groups=32, + act='swish'): + super(BiFPN, self).__init__() + assert num_stacks > 0, "The number of stacks of BiFPN is at least 1." + assert norm_type in ['bn', 'sync_bn', 'gn', None] + assert act in ['swish', 'relu', None] + assert num_extra_levels >= 0, \ + "The `num_extra_levels` must be non negative(>=0)." + + self.in_channels = in_channels + self.out_channel = out_channel + self.num_extra_levels = num_extra_levels + self.num_stacks = num_stacks + self.use_weighted_fusion = use_weighted_fusion + self.norm_type = norm_type + self.norm_groups = norm_groups + self.act = act + self.num_levels = len(self.in_channels) + self.num_extra_levels + if len(fpn_strides) != self.num_levels: + for i in range(self.num_extra_levels): + fpn_strides += [fpn_strides[-1] * 2] + self.fpn_strides = fpn_strides + + self.lateral_convs = nn.LayerList() + for in_c in in_channels: + self.lateral_convs.append( + ConvNormLayer(in_c, self.out_channel, 1, 1)) + if self.num_extra_levels > 0: + self.extra_convs = nn.LayerList() + for i in range(self.num_extra_levels): + if i == 0: + self.extra_convs.append( + ConvNormLayer(self.in_channels[-1], self.out_channel, 3, + 2)) + else: + self.extra_convs.append(nn.MaxPool2D(3, 2, 1)) + + self.bifpn_cells = nn.LayerList() + for i in range(self.num_stacks): + self.bifpn_cells.append( + BiFPNCell( + self.out_channel, + self.num_levels, + use_weighted_fusion=self.use_weighted_fusion, + norm_type=self.norm_type, + norm_groups=self.norm_groups, + act=self.act)) + + @classmethod + def from_config(cls, cfg, input_shape): + return { + 'in_channels': [i.channels for i in input_shape], + 'fpn_strides': [i.stride for i in input_shape] + } + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self.out_channel, stride=s) for s in self.fpn_strides + ] + + def forward(self, feats): + assert len(feats) == len(self.in_channels) + fpn_feats = [] + for conv_layer, feature in zip(self.lateral_convs, feats): + fpn_feats.append(conv_layer(feature)) + if self.num_extra_levels > 0: + feat = feats[-1] + for conv_layer in self.extra_convs: + feat = conv_layer(feat) + fpn_feats.append(feat) + + for bifpn_cell in self.bifpn_cells: + fpn_feats = bifpn_cell(fpn_feats) + return fpn_feats diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/blazeface_fpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/blazeface_fpn.py new file mode 100644 index 000000000..18d7f3cf1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/blazeface_fpn.py @@ -0,0 +1,216 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn.functional as F +from paddle import ParamAttr +import paddle.nn as nn +from paddle.nn.initializer import KaimingNormal +from ppdet.core.workspace import register, serializable +from ..shape_spec import ShapeSpec + +__all__ = ['BlazeNeck'] + + +def hard_swish(x): + return x * F.relu6(x + 3) / 6. + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_channels, + out_channels, + kernel_size, + stride, + padding, + num_groups=1, + act='relu', + conv_lr=0.1, + conv_decay=0., + norm_decay=0., + norm_type='bn', + name=None): + super(ConvBNLayer, self).__init__() + self.act = act + self._conv = nn.Conv2D( + in_channels, + out_channels, + kernel_size=kernel_size, + stride=stride, + padding=padding, + groups=num_groups, + weight_attr=ParamAttr( + learning_rate=conv_lr, initializer=KaimingNormal()), + bias_attr=False) + + if norm_type == 'sync_bn': + self._batch_norm = nn.SyncBatchNorm(out_channels) + else: + self._batch_norm = nn.BatchNorm( + out_channels, act=None, use_global_stats=False) + + def forward(self, x): + x = self._conv(x) + x = self._batch_norm(x) + if self.act == "relu": + x = F.relu(x) + elif self.act == "relu6": + x = F.relu6(x) + elif self.act == 'leaky': + x = F.leaky_relu(x) + elif self.act == 'hard_swish': + x = hard_swish(x) + return x + + +class FPN(nn.Layer): + def __init__(self, in_channels, out_channels, name=None): + super(FPN, self).__init__() + self.conv1_fpn = ConvBNLayer( + in_channels, + out_channels // 2, + kernel_size=1, + padding=0, + stride=1, + act='leaky', + name=name + '_output1') + self.conv2_fpn = ConvBNLayer( + in_channels, + out_channels // 2, + kernel_size=1, + padding=0, + stride=1, + act='leaky', + name=name + '_output2') + self.conv3_fpn = ConvBNLayer( + out_channels // 2, + out_channels // 2, + kernel_size=3, + padding=1, + stride=1, + act='leaky', + name=name + '_merge') + + def forward(self, input): + output1 = self.conv1_fpn(input[0]) + output2 = self.conv2_fpn(input[1]) + up2 = F.upsample( + output2, size=paddle.shape(output1)[-2:], mode='nearest') + output1 = paddle.add(output1, up2) + output1 = self.conv3_fpn(output1) + return output1, output2 + + +class SSH(nn.Layer): + def __init__(self, in_channels, out_channels, name=None): + super(SSH, self).__init__() + assert out_channels % 4 == 0 + self.conv0_ssh = ConvBNLayer( + in_channels, + out_channels // 2, + kernel_size=3, + padding=1, + stride=1, + act=None, + name=name + 'ssh_conv3') + self.conv1_ssh = ConvBNLayer( + out_channels // 2, + out_channels // 4, + kernel_size=3, + padding=1, + stride=1, + act='leaky', + name=name + 'ssh_conv5_1') + self.conv2_ssh = ConvBNLayer( + out_channels // 4, + out_channels // 4, + kernel_size=3, + padding=1, + stride=1, + act=None, + name=name + 'ssh_conv5_2') + self.conv3_ssh = ConvBNLayer( + out_channels // 4, + out_channels // 4, + kernel_size=3, + padding=1, + stride=1, + act='leaky', + name=name + 'ssh_conv7_1') + self.conv4_ssh = ConvBNLayer( + out_channels // 4, + out_channels // 4, + kernel_size=3, + padding=1, + stride=1, + act=None, + name=name + 'ssh_conv7_2') + + def forward(self, x): + conv0 = self.conv0_ssh(x) + conv1 = self.conv1_ssh(conv0) + conv2 = self.conv2_ssh(conv1) + conv3 = self.conv3_ssh(conv2) + conv4 = self.conv4_ssh(conv3) + concat = paddle.concat([conv0, conv2, conv4], axis=1) + return F.relu(concat) + + +@register +@serializable +class BlazeNeck(nn.Layer): + def __init__(self, in_channel, neck_type="None", data_format='NCHW'): + super(BlazeNeck, self).__init__() + self.neck_type = neck_type + self.reture_input = False + self._out_channels = in_channel + if self.neck_type == 'None': + self.reture_input = True + if "fpn" in self.neck_type: + self.fpn = FPN(self._out_channels[0], + self._out_channels[1], + name='fpn') + self._out_channels = [ + self._out_channels[0] // 2, self._out_channels[1] // 2 + ] + if "ssh" in self.neck_type: + self.ssh1 = SSH(self._out_channels[0], + self._out_channels[0], + name='ssh1') + self.ssh2 = SSH(self._out_channels[1], + self._out_channels[1], + name='ssh2') + self._out_channels = [self._out_channels[0], self._out_channels[1]] + + def forward(self, inputs): + if self.reture_input: + return inputs + output1, output2 = None, None + if "fpn" in self.neck_type: + backout_4, backout_1 = inputs + output1, output2 = self.fpn([backout_4, backout_1]) + if self.neck_type == "only_fpn": + return [output1, output2] + if self.neck_type == "only_ssh": + output1, output2 = inputs + feature1 = self.ssh1(output1) + feature2 = self.ssh2(output2) + return [feature1, feature2] + + @property + def out_shape(self): + return [ + ShapeSpec(channels=c) + for c in [self._out_channels[0], self._out_channels[1]] + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/centernet_fpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/centernet_fpn.py new file mode 100644 index 000000000..df5ced2e7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/centernet_fpn.py @@ -0,0 +1,420 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +import math +import paddle +import paddle.nn as nn +from paddle import ParamAttr +from paddle.nn.initializer import Uniform +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ppdet.modeling.layers import ConvNormLayer +from ppdet.modeling.backbones.hardnet import ConvLayer, HarDBlock +from ..shape_spec import ShapeSpec + +__all__ = ['CenterNetDLAFPN', 'CenterNetHarDNetFPN'] + + +# SGE attention +class BasicConv(nn.Layer): + def __init__(self, + in_planes, + out_planes, + kernel_size, + stride=1, + padding=0, + dilation=1, + groups=1, + relu=True, + bn=True, + bias_attr=False): + super(BasicConv, self).__init__() + self.out_channels = out_planes + self.conv = nn.Conv2D( + in_planes, + out_planes, + kernel_size=kernel_size, + stride=stride, + padding=padding, + dilation=dilation, + groups=groups, + bias_attr=bias_attr) + self.bn = nn.BatchNorm2D( + out_planes, + epsilon=1e-5, + momentum=0.01, + weight_attr=False, + bias_attr=False) if bn else None + self.relu = nn.ReLU() if relu else None + + def forward(self, x): + x = self.conv(x) + if self.bn is not None: + x = self.bn(x) + if self.relu is not None: + x = self.relu(x) + return x + + +class ChannelPool(nn.Layer): + def forward(self, x): + return paddle.concat( + (paddle.max(x, 1).unsqueeze(1), paddle.mean(x, 1).unsqueeze(1)), + axis=1) + + +class SpatialGate(nn.Layer): + def __init__(self): + super(SpatialGate, self).__init__() + kernel_size = 7 + self.compress = ChannelPool() + self.spatial = BasicConv( + 2, + 1, + kernel_size, + stride=1, + padding=(kernel_size - 1) // 2, + relu=False) + + def forward(self, x): + x_compress = self.compress(x) + x_out = self.spatial(x_compress) + scale = F.sigmoid(x_out) # broadcasting + return x * scale + + +def fill_up_weights(up): + weight = up.weight.numpy() + f = math.ceil(weight.shape[2] / 2) + c = (2 * f - 1 - f % 2) / (2. * f) + for i in range(weight.shape[2]): + for j in range(weight.shape[3]): + weight[0, 0, i, j] = \ + (1 - math.fabs(i / f - c)) * (1 - math.fabs(j / f - c)) + for c in range(1, weight.shape[0]): + weight[c, 0, :, :] = weight[0, 0, :, :] + up.weight.set_value(weight) + + +class IDAUp(nn.Layer): + def __init__(self, ch_ins, ch_out, up_strides, dcn_v2=True): + super(IDAUp, self).__init__() + for i in range(1, len(ch_ins)): + ch_in = ch_ins[i] + up_s = int(up_strides[i]) + fan_in = ch_in * 3 * 3 + stdv = 1. / math.sqrt(fan_in) + proj = nn.Sequential( + ConvNormLayer( + ch_in, + ch_out, + filter_size=3, + stride=1, + use_dcn=dcn_v2, + bias_on=dcn_v2, + norm_decay=None, + dcn_lr_scale=1., + dcn_regularizer=None, + initializer=Uniform(-stdv, stdv)), + nn.ReLU()) + node = nn.Sequential( + ConvNormLayer( + ch_out, + ch_out, + filter_size=3, + stride=1, + use_dcn=dcn_v2, + bias_on=dcn_v2, + norm_decay=None, + dcn_lr_scale=1., + dcn_regularizer=None, + initializer=Uniform(-stdv, stdv)), + nn.ReLU()) + + kernel_size = up_s * 2 + fan_in = ch_out * kernel_size * kernel_size + stdv = 1. / math.sqrt(fan_in) + up = nn.Conv2DTranspose( + ch_out, + ch_out, + kernel_size=up_s * 2, + stride=up_s, + padding=up_s // 2, + groups=ch_out, + weight_attr=ParamAttr(initializer=Uniform(-stdv, stdv)), + bias_attr=False) + fill_up_weights(up) + setattr(self, 'proj_' + str(i), proj) + setattr(self, 'up_' + str(i), up) + setattr(self, 'node_' + str(i), node) + + def forward(self, inputs, start_level, end_level): + for i in range(start_level + 1, end_level): + upsample = getattr(self, 'up_' + str(i - start_level)) + project = getattr(self, 'proj_' + str(i - start_level)) + + inputs[i] = project(inputs[i]) + inputs[i] = upsample(inputs[i]) + node = getattr(self, 'node_' + str(i - start_level)) + inputs[i] = node(paddle.add(inputs[i], inputs[i - 1])) + + +class DLAUp(nn.Layer): + def __init__(self, start_level, channels, scales, ch_in=None, dcn_v2=True): + super(DLAUp, self).__init__() + self.start_level = start_level + if ch_in is None: + ch_in = channels + self.channels = channels + channels = list(channels) + scales = np.array(scales, dtype=int) + for i in range(len(channels) - 1): + j = -i - 2 + setattr( + self, + 'ida_{}'.format(i), + IDAUp( + ch_in[j:], + channels[j], + scales[j:] // scales[j], + dcn_v2=dcn_v2)) + scales[j + 1:] = scales[j] + ch_in[j + 1:] = [channels[j] for _ in channels[j + 1:]] + + def forward(self, inputs): + out = [inputs[-1]] # start with 32 + for i in range(len(inputs) - self.start_level - 1): + ida = getattr(self, 'ida_{}'.format(i)) + ida(inputs, len(inputs) - i - 2, len(inputs)) + out.insert(0, inputs[-1]) + return out + + +@register +@serializable +class CenterNetDLAFPN(nn.Layer): + """ + Args: + in_channels (list): number of input feature channels from backbone. + [16, 32, 64, 128, 256, 512] by default, means the channels of DLA-34 + down_ratio (int): the down ratio from images to heatmap, 4 by default + last_level (int): the last level of input feature fed into the upsamplng block + out_channel (int): the channel of the output feature, 0 by default means + the channel of the input feature whose down ratio is `down_ratio` + first_level (None): the first level of input feature fed into the upsamplng block. + if None, the first level stands for logs(down_ratio) + dcn_v2 (bool): whether use the DCNv2, True by default + with_sge (bool): whether use SGE attention, False by default + """ + + def __init__(self, + in_channels, + down_ratio=4, + last_level=5, + out_channel=0, + first_level=None, + dcn_v2=True, + with_sge=False): + super(CenterNetDLAFPN, self).__init__() + self.first_level = int(np.log2( + down_ratio)) if first_level is None else first_level + assert self.first_level >= 0, "first level in CenterNetDLAFPN should be greater or equal to 0, but received {}".format( + self.first_level) + self.down_ratio = down_ratio + self.last_level = last_level + scales = [2**i for i in range(len(in_channels[self.first_level:]))] + self.dla_up = DLAUp( + self.first_level, + in_channels[self.first_level:], + scales, + dcn_v2=dcn_v2) + self.out_channel = out_channel + if out_channel == 0: + self.out_channel = in_channels[self.first_level] + self.ida_up = IDAUp( + in_channels[self.first_level:self.last_level], + self.out_channel, + [2**i for i in range(self.last_level - self.first_level)], + dcn_v2=dcn_v2) + + self.with_sge = with_sge + if self.with_sge: + self.sge_attention = SpatialGate() + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape]} + + def forward(self, body_feats): + + dla_up_feats = self.dla_up(body_feats) + + ida_up_feats = [] + for i in range(self.last_level - self.first_level): + ida_up_feats.append(dla_up_feats[i].clone()) + + self.ida_up(ida_up_feats, 0, len(ida_up_feats)) + + feat = ida_up_feats[-1] + if self.with_sge: + feat = self.sge_attention(feat) + if self.down_ratio != 4: + feat = F.interpolate(feat, scale_factor=self.down_ratio // 4, mode="bilinear", align_corners=True) + return feat + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, stride=self.down_ratio)] + + +class TransitionUp(nn.Layer): + def __init__(self, in_channels, out_channels): + super().__init__() + + def forward(self, x, skip): + w, h = skip.shape[2], skip.shape[3] + out = F.interpolate(x, size=(w, h), mode="bilinear", align_corners=True) + out = paddle.concat([out, skip], 1) + return out + + +@register +@serializable +class CenterNetHarDNetFPN(nn.Layer): + """ + Args: + in_channels (list): number of input feature channels from backbone. + [96, 214, 458, 784] by default, means the channels of HarDNet85 + num_layers (int): HarDNet laters, 85 by default + down_ratio (int): the down ratio from images to heatmap, 4 by default + first_level (int|None): the first level of input feature fed into the upsamplng block. + if None, the first level stands for logs(down_ratio) - 1 + + last_level (int): the last level of input feature fed into the upsamplng block + out_channel (int): the channel of the output feature, 0 by default means + the channel of the input feature whose down ratio is `down_ratio` + """ + + def __init__(self, + in_channels, + num_layers=85, + down_ratio=4, + first_level=None, + last_level=4, + out_channel=0): + super(CenterNetHarDNetFPN, self).__init__() + self.first_level = int(np.log2( + down_ratio)) - 1 if first_level is None else first_level + assert self.first_level >= 0, "first level in CenterNetDLAFPN should be greater or equal to 0, but received {}".format( + self.first_level) + self.down_ratio = down_ratio + self.last_level = last_level + self.last_pool = nn.AvgPool2D(kernel_size=2, stride=2) + + assert num_layers in [68, 85], "HarDNet-{} not support.".format( + num_layers) + if num_layers == 85: + self.last_proj = ConvLayer(784, 256, kernel_size=1) + self.last_blk = HarDBlock(768, 80, 1.7, 8) + self.skip_nodes = [1, 3, 8, 13] + self.SC = [32, 32, 0] + gr = [64, 48, 28] + layers = [8, 8, 4] + ch_list2 = [224 + self.SC[0], 160 + self.SC[1], 96 + self.SC[2]] + channels = [96, 214, 458, 784] + self.skip_lv = 3 + + elif num_layers == 68: + self.last_proj = ConvLayer(654, 192, kernel_size=1) + self.last_blk = HarDBlock(576, 72, 1.7, 8) + self.skip_nodes = [1, 3, 8, 11] + self.SC = [32, 32, 0] + gr = [48, 32, 20] + layers = [8, 8, 4] + ch_list2 = [224 + self.SC[0], 96 + self.SC[1], 64 + self.SC[2]] + channels = [64, 124, 328, 654] + self.skip_lv = 2 + + self.transUpBlocks = nn.LayerList([]) + self.denseBlocksUp = nn.LayerList([]) + self.conv1x1_up = nn.LayerList([]) + self.avg9x9 = nn.AvgPool2D(kernel_size=(9, 9), stride=1, padding=(4, 4)) + prev_ch = self.last_blk.get_out_ch() + + for i in range(3): + skip_ch = channels[3 - i] + self.transUpBlocks.append(TransitionUp(prev_ch, prev_ch)) + if i < self.skip_lv: + cur_ch = prev_ch + skip_ch + else: + cur_ch = prev_ch + self.conv1x1_up.append( + ConvLayer( + cur_ch, ch_list2[i], kernel_size=1)) + cur_ch = ch_list2[i] + cur_ch -= self.SC[i] + cur_ch *= 3 + + blk = HarDBlock(cur_ch, gr[i], 1.7, layers[i]) + self.denseBlocksUp.append(blk) + prev_ch = blk.get_out_ch() + + prev_ch += self.SC[0] + self.SC[1] + self.SC[2] + self.out_channel = prev_ch + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape]} + + def forward(self, body_feats): + x = body_feats[-1] + x_sc = [] + x = self.last_proj(x) + x = self.last_pool(x) + x2 = self.avg9x9(x) + x3 = x / (x.sum((2, 3), keepdim=True) + 0.1) + x = paddle.concat([x, x2, x3], 1) + x = self.last_blk(x) + + for i in range(3): + skip_x = body_feats[3 - i] + x_up = self.transUpBlocks[i](x, skip_x) + x_ch = self.conv1x1_up[i](x_up) + if self.SC[i] > 0: + end = x_ch.shape[1] + new_st = end - self.SC[i] + x_sc.append(x_ch[:, new_st:, :, :]) + x_ch = x_ch[:, :new_st, :, :] + x2 = self.avg9x9(x_ch) + x3 = x_ch / (x_ch.sum((2, 3), keepdim=True) + 0.1) + x_new = paddle.concat([x_ch, x2, x3], 1) + x = self.denseBlocksUp[i](x_new) + + scs = [x] + for i in range(3): + if self.SC[i] > 0: + scs.insert( + 0, + F.interpolate( + x_sc[i], + size=(x.shape[2], x.shape[3]), + mode="bilinear", + align_corners=True)) + neck_feat = paddle.concat(scs, 1) + return neck_feat + + @property + def out_shape(self): + return [ShapeSpec(channels=self.out_channel, stride=self.down_ratio)] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/csp_pan.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/csp_pan.py new file mode 100644 index 000000000..7417c46ab --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/csp_pan.py @@ -0,0 +1,364 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/open-mmlab/mmdetection/blob/master/mmdet/models/necks/yolox_pafpn.py + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from ppdet.core.workspace import register, serializable +from ..shape_spec import ShapeSpec + +__all__ = ['CSPPAN'] + + +class ConvBNLayer(nn.Layer): + def __init__(self, + in_channel=96, + out_channel=96, + kernel_size=3, + stride=1, + groups=1, + act='leaky_relu'): + super(ConvBNLayer, self).__init__() + initializer = nn.initializer.KaimingUniform() + self.act = act + assert self.act in ['leaky_relu', "hard_swish"] + self.conv = nn.Conv2D( + in_channels=in_channel, + out_channels=out_channel, + kernel_size=kernel_size, + groups=groups, + padding=(kernel_size - 1) // 2, + stride=stride, + weight_attr=ParamAttr(initializer=initializer), + bias_attr=False) + self.bn = nn.BatchNorm2D(out_channel) + + def forward(self, x): + x = self.bn(self.conv(x)) + if self.act == "leaky_relu": + x = F.leaky_relu(x) + elif self.act == "hard_swish": + x = F.hardswish(x) + return x + + +class DPModule(nn.Layer): + """ + Depth-wise and point-wise module. + Args: + in_channel (int): The input channels of this Module. + out_channel (int): The output channels of this Module. + kernel_size (int): The conv2d kernel size of this Module. + stride (int): The conv2d's stride of this Module. + act (str): The activation function of this Module, + Now support `leaky_relu` and `hard_swish`. + """ + + def __init__(self, + in_channel=96, + out_channel=96, + kernel_size=3, + stride=1, + act='leaky_relu'): + super(DPModule, self).__init__() + initializer = nn.initializer.KaimingUniform() + self.act = act + self.dwconv = nn.Conv2D( + in_channels=in_channel, + out_channels=out_channel, + kernel_size=kernel_size, + groups=out_channel, + padding=(kernel_size - 1) // 2, + stride=stride, + weight_attr=ParamAttr(initializer=initializer), + bias_attr=False) + self.bn1 = nn.BatchNorm2D(out_channel) + self.pwconv = nn.Conv2D( + in_channels=out_channel, + out_channels=out_channel, + kernel_size=1, + groups=1, + padding=0, + weight_attr=ParamAttr(initializer=initializer), + bias_attr=False) + self.bn2 = nn.BatchNorm2D(out_channel) + + def act_func(self, x): + if self.act == "leaky_relu": + x = F.leaky_relu(x) + elif self.act == "hard_swish": + x = F.hardswish(x) + return x + + def forward(self, x): + x = self.act_func(self.bn1(self.dwconv(x))) + x = self.act_func(self.bn2(self.pwconv(x))) + return x + + +class DarknetBottleneck(nn.Layer): + """The basic bottleneck block used in Darknet. + + Each Block consists of two ConvModules and the input is added to the + final output. Each ConvModule is composed of Conv, BN, and act. + The first convLayer has filter size of 1x1 and the second one has the + filter size of 3x3. + + Args: + in_channels (int): The input channels of this Module. + out_channels (int): The output channels of this Module. + expansion (int): The kernel size of the convolution. Default: 0.5 + add_identity (bool): Whether to add identity to the out. + Default: True + use_depthwise (bool): Whether to use depthwise separable convolution. + Default: False + """ + + def __init__(self, + in_channels, + out_channels, + kernel_size=3, + expansion=0.5, + add_identity=True, + use_depthwise=False, + act="leaky_relu"): + super(DarknetBottleneck, self).__init__() + hidden_channels = int(out_channels * expansion) + conv_func = DPModule if use_depthwise else ConvBNLayer + self.conv1 = ConvBNLayer( + in_channel=in_channels, + out_channel=hidden_channels, + kernel_size=1, + act=act) + self.conv2 = conv_func( + in_channel=hidden_channels, + out_channel=out_channels, + kernel_size=kernel_size, + stride=1, + act=act) + self.add_identity = \ + add_identity and in_channels == out_channels + + def forward(self, x): + identity = x + out = self.conv1(x) + out = self.conv2(out) + + if self.add_identity: + return out + identity + else: + return out + + +class CSPLayer(nn.Layer): + """Cross Stage Partial Layer. + + Args: + in_channels (int): The input channels of the CSP layer. + out_channels (int): The output channels of the CSP layer. + expand_ratio (float): Ratio to adjust the number of channels of the + hidden layer. Default: 0.5 + num_blocks (int): Number of blocks. Default: 1 + add_identity (bool): Whether to add identity in blocks. + Default: True + use_depthwise (bool): Whether to depthwise separable convolution in + blocks. Default: False + """ + + def __init__(self, + in_channels, + out_channels, + kernel_size=3, + expand_ratio=0.5, + num_blocks=1, + add_identity=True, + use_depthwise=False, + act="leaky_relu"): + super().__init__() + mid_channels = int(out_channels * expand_ratio) + self.main_conv = ConvBNLayer(in_channels, mid_channels, 1, act=act) + self.short_conv = ConvBNLayer(in_channels, mid_channels, 1, act=act) + self.final_conv = ConvBNLayer( + 2 * mid_channels, out_channels, 1, act=act) + + self.blocks = nn.Sequential(* [ + DarknetBottleneck( + mid_channels, + mid_channels, + kernel_size, + 1.0, + add_identity, + use_depthwise, + act=act) for _ in range(num_blocks) + ]) + + def forward(self, x): + x_short = self.short_conv(x) + + x_main = self.main_conv(x) + x_main = self.blocks(x_main) + + x_final = paddle.concat((x_main, x_short), axis=1) + return self.final_conv(x_final) + + +class Channel_T(nn.Layer): + def __init__(self, + in_channels=[116, 232, 464], + out_channels=96, + act="leaky_relu"): + super(Channel_T, self).__init__() + self.convs = nn.LayerList() + for i in range(len(in_channels)): + self.convs.append( + ConvBNLayer( + in_channels[i], out_channels, 1, act=act)) + + def forward(self, x): + outs = [self.convs[i](x[i]) for i in range(len(x))] + return outs + + +@register +@serializable +class CSPPAN(nn.Layer): + """Path Aggregation Network with CSP module. + + Args: + in_channels (List[int]): Number of input channels per scale. + out_channels (int): Number of output channels (used at each scale) + kernel_size (int): The conv2d kernel size of this Module. + num_features (int): Number of output features of CSPPAN module. + num_csp_blocks (int): Number of bottlenecks in CSPLayer. Default: 1 + use_depthwise (bool): Whether to depthwise separable convolution in + blocks. Default: True + """ + + def __init__(self, + in_channels, + out_channels, + kernel_size=5, + num_features=3, + num_csp_blocks=1, + use_depthwise=True, + act='hard_swish', + spatial_scales=[0.125, 0.0625, 0.03125]): + super(CSPPAN, self).__init__() + self.conv_t = Channel_T(in_channels, out_channels, act=act) + in_channels = [out_channels] * len(spatial_scales) + self.in_channels = in_channels + self.out_channels = out_channels + self.spatial_scales = spatial_scales + self.num_features = num_features + conv_func = DPModule if use_depthwise else ConvBNLayer + + if self.num_features == 4: + self.first_top_conv = conv_func( + in_channels[0], in_channels[0], kernel_size, stride=2, act=act) + self.second_top_conv = conv_func( + in_channels[0], in_channels[0], kernel_size, stride=2, act=act) + self.spatial_scales.append(self.spatial_scales[-1] / 2) + + # build top-down blocks + self.upsample = nn.Upsample(scale_factor=2, mode='nearest') + self.top_down_blocks = nn.LayerList() + for idx in range(len(in_channels) - 1, 0, -1): + self.top_down_blocks.append( + CSPLayer( + in_channels[idx - 1] * 2, + in_channels[idx - 1], + kernel_size=kernel_size, + num_blocks=num_csp_blocks, + add_identity=False, + use_depthwise=use_depthwise, + act=act)) + + # build bottom-up blocks + self.downsamples = nn.LayerList() + self.bottom_up_blocks = nn.LayerList() + for idx in range(len(in_channels) - 1): + self.downsamples.append( + conv_func( + in_channels[idx], + in_channels[idx], + kernel_size=kernel_size, + stride=2, + act=act)) + self.bottom_up_blocks.append( + CSPLayer( + in_channels[idx] * 2, + in_channels[idx + 1], + kernel_size=kernel_size, + num_blocks=num_csp_blocks, + add_identity=False, + use_depthwise=use_depthwise, + act=act)) + + def forward(self, inputs): + """ + Args: + inputs (tuple[Tensor]): input features. + + Returns: + tuple[Tensor]: CSPPAN features. + """ + assert len(inputs) == len(self.in_channels) + inputs = self.conv_t(inputs) + + # top-down path + inner_outs = [inputs[-1]] + for idx in range(len(self.in_channels) - 1, 0, -1): + feat_heigh = inner_outs[0] + feat_low = inputs[idx - 1] + + upsample_feat = self.upsample(feat_heigh) + + inner_out = self.top_down_blocks[len(self.in_channels) - 1 - idx]( + paddle.concat([upsample_feat, feat_low], 1)) + inner_outs.insert(0, inner_out) + + # bottom-up path + outs = [inner_outs[0]] + for idx in range(len(self.in_channels) - 1): + feat_low = outs[-1] + feat_height = inner_outs[idx + 1] + downsample_feat = self.downsamples[idx](feat_low) + out = self.bottom_up_blocks[idx](paddle.concat( + [downsample_feat, feat_height], 1)) + outs.append(out) + + top_features = None + if self.num_features == 4: + top_features = self.first_top_conv(inputs[-1]) + top_features = top_features + self.second_top_conv(outs[-1]) + outs.append(top_features) + + return tuple(outs) + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self.out_channels, stride=1. / s) + for s in self.spatial_scales + ] + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/fpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/fpn.py new file mode 100644 index 000000000..0633fb5b2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/fpn.py @@ -0,0 +1,231 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import XavierUniform + +from ppdet.core.workspace import register, serializable +from ppdet.modeling.layers import ConvNormLayer +from ..shape_spec import ShapeSpec + +__all__ = ['FPN'] + + +@register +@serializable +class FPN(nn.Layer): + """ + Feature Pyramid Network, see https://arxiv.org/abs/1612.03144 + + Args: + in_channels (list[int]): input channels of each level which can be + derived from the output shape of backbone by from_config + out_channel (list[int]): output channel of each level + spatial_scales (list[float]): the spatial scales between input feature + maps and original input image which can be derived from the output + shape of backbone by from_config + has_extra_convs (bool): whether to add extra conv to the last level. + default False + extra_stage (int): the number of extra stages added to the last level. + default 1 + use_c5 (bool): Whether to use c5 as the input of extra stage, + otherwise p5 is used. default True + norm_type (string|None): The normalization type in FPN module. If + norm_type is None, norm will not be used after conv and if + norm_type is string, bn, gn, sync_bn are available. default None + norm_decay (float): weight decay for normalization layer weights. + default 0. + freeze_norm (bool): whether to freeze normalization layer. + default False + relu_before_extra_convs (bool): whether to add relu before extra convs. + default False + + """ + + def __init__(self, + in_channels, + out_channel, + spatial_scales=[0.25, 0.125, 0.0625, 0.03125], + has_extra_convs=False, + extra_stage=1, + use_c5=True, + norm_type=None, + norm_decay=0., + freeze_norm=False, + relu_before_extra_convs=True): + super(FPN, self).__init__() + self.out_channel = out_channel + for s in range(extra_stage): + spatial_scales = spatial_scales + [spatial_scales[-1] / 2.] + self.spatial_scales = spatial_scales + self.has_extra_convs = has_extra_convs + self.extra_stage = extra_stage + self.use_c5 = use_c5 + self.relu_before_extra_convs = relu_before_extra_convs + self.norm_type = norm_type + self.norm_decay = norm_decay + self.freeze_norm = freeze_norm + + self.lateral_convs = [] + self.fpn_convs = [] + fan = out_channel * 3 * 3 + + # stage index 0,1,2,3 stands for res2,res3,res4,res5 on ResNet Backbone + # 0 <= st_stage < ed_stage <= 3 + st_stage = 4 - len(in_channels) + ed_stage = st_stage + len(in_channels) - 1 + for i in range(st_stage, ed_stage + 1): + if i == 3: + lateral_name = 'fpn_inner_res5_sum' + else: + lateral_name = 'fpn_inner_res{}_sum_lateral'.format(i + 2) + in_c = in_channels[i - st_stage] + if self.norm_type is not None: + lateral = self.add_sublayer( + lateral_name, + ConvNormLayer( + ch_in=in_c, + ch_out=out_channel, + filter_size=1, + stride=1, + norm_type=self.norm_type, + norm_decay=self.norm_decay, + freeze_norm=self.freeze_norm, + initializer=XavierUniform(fan_out=in_c))) + else: + lateral = self.add_sublayer( + lateral_name, + nn.Conv2D( + in_channels=in_c, + out_channels=out_channel, + kernel_size=1, + weight_attr=ParamAttr( + initializer=XavierUniform(fan_out=in_c)))) + self.lateral_convs.append(lateral) + + fpn_name = 'fpn_res{}_sum'.format(i + 2) + if self.norm_type is not None: + fpn_conv = self.add_sublayer( + fpn_name, + ConvNormLayer( + ch_in=out_channel, + ch_out=out_channel, + filter_size=3, + stride=1, + norm_type=self.norm_type, + norm_decay=self.norm_decay, + freeze_norm=self.freeze_norm, + initializer=XavierUniform(fan_out=fan))) + else: + fpn_conv = self.add_sublayer( + fpn_name, + nn.Conv2D( + in_channels=out_channel, + out_channels=out_channel, + kernel_size=3, + padding=1, + weight_attr=ParamAttr( + initializer=XavierUniform(fan_out=fan)))) + self.fpn_convs.append(fpn_conv) + + # add extra conv levels for RetinaNet(use_c5)/FCOS(use_p5) + if self.has_extra_convs: + for i in range(self.extra_stage): + lvl = ed_stage + 1 + i + if i == 0 and self.use_c5: + in_c = in_channels[-1] + else: + in_c = out_channel + extra_fpn_name = 'fpn_{}'.format(lvl + 2) + if self.norm_type is not None: + extra_fpn_conv = self.add_sublayer( + extra_fpn_name, + ConvNormLayer( + ch_in=in_c, + ch_out=out_channel, + filter_size=3, + stride=2, + norm_type=self.norm_type, + norm_decay=self.norm_decay, + freeze_norm=self.freeze_norm, + initializer=XavierUniform(fan_out=fan))) + else: + extra_fpn_conv = self.add_sublayer( + extra_fpn_name, + nn.Conv2D( + in_channels=in_c, + out_channels=out_channel, + kernel_size=3, + stride=2, + padding=1, + weight_attr=ParamAttr( + initializer=XavierUniform(fan_out=fan)))) + self.fpn_convs.append(extra_fpn_conv) + + @classmethod + def from_config(cls, cfg, input_shape): + return { + 'in_channels': [i.channels for i in input_shape], + 'spatial_scales': [1.0 / i.stride for i in input_shape], + } + + def forward(self, body_feats): + laterals = [] + num_levels = len(body_feats) + for i in range(num_levels): + laterals.append(self.lateral_convs[i](body_feats[i])) + + for i in range(1, num_levels): + lvl = num_levels - i + upsample = F.interpolate( + laterals[lvl], + scale_factor=2., + mode='nearest', ) + laterals[lvl - 1] += upsample + + fpn_output = [] + for lvl in range(num_levels): + fpn_output.append(self.fpn_convs[lvl](laterals[lvl])) + + if self.extra_stage > 0: + # use max pool to get more levels on top of outputs (Faster R-CNN, Mask R-CNN) + if not self.has_extra_convs: + assert self.extra_stage == 1, 'extra_stage should be 1 if FPN has not extra convs' + fpn_output.append(F.max_pool2d(fpn_output[-1], 1, stride=2)) + # add extra conv levels for RetinaNet(use_c5)/FCOS(use_p5) + else: + if self.use_c5: + extra_source = body_feats[-1] + else: + extra_source = fpn_output[-1] + fpn_output.append(self.fpn_convs[num_levels](extra_source)) + + for i in range(1, self.extra_stage): + if self.relu_before_extra_convs: + fpn_output.append(self.fpn_convs[num_levels + i](F.relu( + fpn_output[-1]))) + else: + fpn_output.append(self.fpn_convs[num_levels + i]( + fpn_output[-1])) + return fpn_output + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self.out_channel, stride=1. / s) + for s in self.spatial_scales + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/hrfpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/hrfpn.py new file mode 100644 index 000000000..eb4768b8e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/hrfpn.py @@ -0,0 +1,126 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn.functional as F +import paddle.nn as nn +from ppdet.core.workspace import register +from ..shape_spec import ShapeSpec + +__all__ = ['HRFPN'] + + +@register +class HRFPN(nn.Layer): + """ + Args: + in_channels (list): number of input feature channels from backbone + out_channel (int): number of output feature channels + share_conv (bool): whether to share conv for different layers' reduction + extra_stage (int): add extra stage for returning HRFPN fpn_feats + spatial_scales (list): feature map scaling factor + """ + + def __init__(self, + in_channels=[18, 36, 72, 144], + out_channel=256, + share_conv=False, + extra_stage=1, + spatial_scales=[1. / 4, 1. / 8, 1. / 16, 1. / 32]): + super(HRFPN, self).__init__() + in_channel = sum(in_channels) + self.in_channel = in_channel + self.out_channel = out_channel + self.share_conv = share_conv + for i in range(extra_stage): + spatial_scales = spatial_scales + [spatial_scales[-1] / 2.] + self.spatial_scales = spatial_scales + self.num_out = len(self.spatial_scales) + + self.reduction = nn.Conv2D( + in_channels=in_channel, + out_channels=out_channel, + kernel_size=1, + bias_attr=False) + + if share_conv: + self.fpn_conv = nn.Conv2D( + in_channels=out_channel, + out_channels=out_channel, + kernel_size=3, + padding=1, + bias_attr=False) + else: + self.fpn_conv = [] + for i in range(self.num_out): + conv_name = "fpn_conv_" + str(i) + conv = self.add_sublayer( + conv_name, + nn.Conv2D( + in_channels=out_channel, + out_channels=out_channel, + kernel_size=3, + padding=1, + bias_attr=False)) + self.fpn_conv.append(conv) + + def forward(self, body_feats): + num_backbone_stages = len(body_feats) + + outs = [] + outs.append(body_feats[0]) + + # resize + for i in range(1, num_backbone_stages): + resized = F.interpolate( + body_feats[i], scale_factor=2**i, mode='bilinear') + outs.append(resized) + + # concat + out = paddle.concat(outs, axis=1) + assert out.shape[ + 1] == self.in_channel, 'in_channel should be {}, be received {}'.format( + out.shape[1], self.in_channel) + + # reduction + out = self.reduction(out) + + # conv + outs = [out] + for i in range(1, self.num_out): + outs.append(F.avg_pool2d(out, kernel_size=2**i, stride=2**i)) + outputs = [] + + for i in range(self.num_out): + conv_func = self.fpn_conv if self.share_conv else self.fpn_conv[i] + conv = conv_func(outs[i]) + outputs.append(conv) + + fpn_feats = [outputs[k] for k in range(self.num_out)] + return fpn_feats + + @classmethod + def from_config(cls, cfg, input_shape): + return { + 'in_channels': [i.channels for i in input_shape], + 'spatial_scales': [1.0 / i.stride for i in input_shape], + } + + @property + def out_shape(self): + return [ + ShapeSpec( + channels=self.out_channel, stride=1. / s) + for s in self.spatial_scales + ] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/ttf_fpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/ttf_fpn.py new file mode 100644 index 000000000..60cc69f80 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/ttf_fpn.py @@ -0,0 +1,242 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.nn.initializer import Constant, Uniform, Normal, XavierUniform +from ppdet.core.workspace import register, serializable +from paddle.regularizer import L2Decay +from ppdet.modeling.layers import DeformableConvV2, ConvNormLayer, LiteConv +import math +from ppdet.modeling.ops import batch_norm +from ..shape_spec import ShapeSpec + +__all__ = ['TTFFPN'] + + +class Upsample(nn.Layer): + def __init__(self, ch_in, ch_out, norm_type='bn'): + super(Upsample, self).__init__() + fan_in = ch_in * 3 * 3 + stdv = 1. / math.sqrt(fan_in) + self.dcn = DeformableConvV2( + ch_in, + ch_out, + kernel_size=3, + weight_attr=ParamAttr(initializer=Uniform(-stdv, stdv)), + bias_attr=ParamAttr( + initializer=Constant(0), + regularizer=L2Decay(0.), + learning_rate=2.), + lr_scale=2., + regularizer=L2Decay(0.)) + + self.bn = batch_norm( + ch_out, norm_type=norm_type, initializer=Constant(1.)) + + def forward(self, feat): + dcn = self.dcn(feat) + bn = self.bn(dcn) + relu = F.relu(bn) + out = F.interpolate(relu, scale_factor=2., mode='bilinear') + return out + + +class DeConv(nn.Layer): + def __init__(self, ch_in, ch_out, norm_type='bn'): + super(DeConv, self).__init__() + self.deconv = nn.Sequential() + conv1 = ConvNormLayer( + ch_in=ch_in, + ch_out=ch_out, + stride=1, + filter_size=1, + norm_type=norm_type, + initializer=XavierUniform()) + conv2 = nn.Conv2DTranspose( + in_channels=ch_out, + out_channels=ch_out, + kernel_size=4, + padding=1, + stride=2, + groups=ch_out, + weight_attr=ParamAttr(initializer=XavierUniform()), + bias_attr=False) + bn = batch_norm(ch_out, norm_type=norm_type, norm_decay=0.) + conv3 = ConvNormLayer( + ch_in=ch_out, + ch_out=ch_out, + stride=1, + filter_size=1, + norm_type=norm_type, + initializer=XavierUniform()) + + self.deconv.add_sublayer('conv1', conv1) + self.deconv.add_sublayer('relu6_1', nn.ReLU6()) + self.deconv.add_sublayer('conv2', conv2) + self.deconv.add_sublayer('bn', bn) + self.deconv.add_sublayer('relu6_2', nn.ReLU6()) + self.deconv.add_sublayer('conv3', conv3) + self.deconv.add_sublayer('relu6_3', nn.ReLU6()) + + def forward(self, inputs): + return self.deconv(inputs) + + +class LiteUpsample(nn.Layer): + def __init__(self, ch_in, ch_out, norm_type='bn'): + super(LiteUpsample, self).__init__() + self.deconv = DeConv(ch_in, ch_out, norm_type=norm_type) + self.conv = LiteConv(ch_in, ch_out, norm_type=norm_type) + + def forward(self, inputs): + deconv_up = self.deconv(inputs) + conv = self.conv(inputs) + interp_up = F.interpolate(conv, scale_factor=2., mode='bilinear') + return deconv_up + interp_up + + +class ShortCut(nn.Layer): + def __init__(self, + layer_num, + ch_in, + ch_out, + norm_type='bn', + lite_neck=False, + name=None): + super(ShortCut, self).__init__() + shortcut_conv = nn.Sequential() + for i in range(layer_num): + fan_out = 3 * 3 * ch_out + std = math.sqrt(2. / fan_out) + in_channels = ch_in if i == 0 else ch_out + shortcut_name = name + '.conv.{}'.format(i) + if lite_neck: + shortcut_conv.add_sublayer( + shortcut_name, + LiteConv( + in_channels=in_channels, + out_channels=ch_out, + with_act=i < layer_num - 1, + norm_type=norm_type)) + else: + shortcut_conv.add_sublayer( + shortcut_name, + nn.Conv2D( + in_channels=in_channels, + out_channels=ch_out, + kernel_size=3, + padding=1, + weight_attr=ParamAttr(initializer=Normal(0, std)), + bias_attr=ParamAttr( + learning_rate=2., regularizer=L2Decay(0.)))) + if i < layer_num - 1: + shortcut_conv.add_sublayer(shortcut_name + '.act', + nn.ReLU()) + self.shortcut = self.add_sublayer('shortcut', shortcut_conv) + + def forward(self, feat): + out = self.shortcut(feat) + return out + + +@register +@serializable +class TTFFPN(nn.Layer): + """ + Args: + in_channels (list): number of input feature channels from backbone. + [128,256,512,1024] by default, means the channels of DarkNet53 + backbone return_idx [1,2,3,4]. + planes (list): the number of output feature channels of FPN. + [256, 128, 64] by default + shortcut_num (list): the number of convolution layers in each shortcut. + [3,2,1] by default, means DarkNet53 backbone return_idx_1 has 3 convs + in its shortcut, return_idx_2 has 2 convs and return_idx_3 has 1 conv. + norm_type (string): norm type, 'sync_bn', 'bn', 'gn' are optional. + bn by default + lite_neck (bool): whether to use lite conv in TTFNet FPN, + False by default + fusion_method (string): the method to fusion upsample and lateral layer. + 'add' and 'concat' are optional, add by default + """ + + __shared__ = ['norm_type'] + + def __init__(self, + in_channels, + planes=[256, 128, 64], + shortcut_num=[3, 2, 1], + norm_type='bn', + lite_neck=False, + fusion_method='add'): + super(TTFFPN, self).__init__() + self.planes = planes + self.shortcut_num = shortcut_num[::-1] + self.shortcut_len = len(shortcut_num) + self.ch_in = in_channels[::-1] + self.fusion_method = fusion_method + + self.upsample_list = [] + self.shortcut_list = [] + self.upper_list = [] + for i, out_c in enumerate(self.planes): + in_c = self.ch_in[i] if i == 0 else self.upper_list[-1] + upsample_module = LiteUpsample if lite_neck else Upsample + upsample = self.add_sublayer( + 'upsample.' + str(i), + upsample_module( + in_c, out_c, norm_type=norm_type)) + self.upsample_list.append(upsample) + if i < self.shortcut_len: + shortcut = self.add_sublayer( + 'shortcut.' + str(i), + ShortCut( + self.shortcut_num[i], + self.ch_in[i + 1], + out_c, + norm_type=norm_type, + lite_neck=lite_neck, + name='shortcut.' + str(i))) + self.shortcut_list.append(shortcut) + if self.fusion_method == 'add': + upper_c = out_c + elif self.fusion_method == 'concat': + upper_c = out_c * 2 + else: + raise ValueError('Illegal fusion method. Expected add or\ + concat, but received {}'.format(self.fusion_method)) + self.upper_list.append(upper_c) + + def forward(self, inputs): + feat = inputs[-1] + for i, out_c in enumerate(self.planes): + feat = self.upsample_list[i](feat) + if i < self.shortcut_len: + shortcut = self.shortcut_list[i](inputs[-i - 2]) + if self.fusion_method == 'add': + feat = feat + shortcut + else: + feat = paddle.concat([feat, shortcut], axis=1) + return feat + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + @property + def out_shape(self): + return [ShapeSpec(channels=self.upper_list[-1], )] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/yolo_fpn.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/yolo_fpn.py new file mode 100644 index 000000000..4af0348d2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/necks/yolo_fpn.py @@ -0,0 +1,988 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register, serializable +from ppdet.modeling.layers import DropBlock +from ..backbones.darknet import ConvBNLayer +from ..shape_spec import ShapeSpec + +__all__ = ['YOLOv3FPN', 'PPYOLOFPN', 'PPYOLOTinyFPN', 'PPYOLOPAN'] + + +def add_coord(x, data_format): + b = paddle.shape(x)[0] + if data_format == 'NCHW': + h, w = x.shape[2], x.shape[3] + else: + h, w = x.shape[1], x.shape[2] + + gx = paddle.cast(paddle.arange(w) / ((w - 1.) * 2.0) - 1., x.dtype) + gy = paddle.cast(paddle.arange(h) / ((h - 1.) * 2.0) - 1., x.dtype) + + if data_format == 'NCHW': + gx = gx.reshape([1, 1, 1, w]).expand([b, 1, h, w]) + gy = gy.reshape([1, 1, h, 1]).expand([b, 1, h, w]) + else: + gx = gx.reshape([1, 1, w, 1]).expand([b, h, w, 1]) + gy = gy.reshape([1, h, 1, 1]).expand([b, h, w, 1]) + + gx.stop_gradient = True + gy.stop_gradient = True + return gx, gy + + +class YoloDetBlock(nn.Layer): + def __init__(self, + ch_in, + channel, + norm_type, + freeze_norm=False, + name='', + data_format='NCHW'): + """ + YOLODetBlock layer for yolov3, see https://arxiv.org/abs/1804.02767 + + Args: + ch_in (int): input channel + channel (int): base channel + norm_type (str): batch norm type + freeze_norm (bool): whether to freeze norm, default False + name (str): layer name + data_format (str): data format, NCHW or NHWC + """ + super(YoloDetBlock, self).__init__() + self.ch_in = ch_in + self.channel = channel + assert channel % 2 == 0, \ + "channel {} cannot be divided by 2".format(channel) + conv_def = [ + ['conv0', ch_in, channel, 1, '.0.0'], + ['conv1', channel, channel * 2, 3, '.0.1'], + ['conv2', channel * 2, channel, 1, '.1.0'], + ['conv3', channel, channel * 2, 3, '.1.1'], + ['route', channel * 2, channel, 1, '.2'], + ] + + self.conv_module = nn.Sequential() + for idx, (conv_name, ch_in, ch_out, filter_size, + post_name) in enumerate(conv_def): + self.conv_module.add_sublayer( + conv_name, + ConvBNLayer( + ch_in=ch_in, + ch_out=ch_out, + filter_size=filter_size, + padding=(filter_size - 1) // 2, + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name + post_name)) + + self.tip = ConvBNLayer( + ch_in=channel, + ch_out=channel * 2, + filter_size=3, + padding=1, + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name + '.tip') + + def forward(self, inputs): + route = self.conv_module(inputs) + tip = self.tip(route) + return route, tip + + +class SPP(nn.Layer): + def __init__(self, + ch_in, + ch_out, + k, + pool_size, + norm_type, + freeze_norm=False, + name='', + act='leaky', + data_format='NCHW'): + """ + SPP layer, which consist of four pooling layer follwed by conv layer + + Args: + ch_in (int): input channel of conv layer + ch_out (int): output channel of conv layer + k (int): kernel size of conv layer + norm_type (str): batch norm type + freeze_norm (bool): whether to freeze norm, default False + name (str): layer name + act (str): activation function + data_format (str): data format, NCHW or NHWC + """ + super(SPP, self).__init__() + self.pool = [] + self.data_format = data_format + for size in pool_size: + pool = self.add_sublayer( + '{}.pool1'.format(name), + nn.MaxPool2D( + kernel_size=size, + stride=1, + padding=size // 2, + data_format=data_format, + ceil_mode=False)) + self.pool.append(pool) + self.conv = ConvBNLayer( + ch_in, + ch_out, + k, + padding=k // 2, + norm_type=norm_type, + freeze_norm=freeze_norm, + name=name, + act=act, + data_format=data_format) + + def forward(self, x): + outs = [x] + for pool in self.pool: + outs.append(pool(x)) + if self.data_format == "NCHW": + y = paddle.concat(outs, axis=1) + else: + y = paddle.concat(outs, axis=-1) + + y = self.conv(y) + return y + + +class CoordConv(nn.Layer): + def __init__(self, + ch_in, + ch_out, + filter_size, + padding, + norm_type, + freeze_norm=False, + name='', + data_format='NCHW'): + """ + CoordConv layer, see https://arxiv.org/abs/1807.03247 + + Args: + ch_in (int): input channel + ch_out (int): output channel + filter_size (int): filter size, default 3 + padding (int): padding size, default 0 + norm_type (str): batch norm type, default bn + name (str): layer name + data_format (str): data format, NCHW or NHWC + + """ + super(CoordConv, self).__init__() + self.conv = ConvBNLayer( + ch_in + 2, + ch_out, + filter_size=filter_size, + padding=padding, + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name) + self.data_format = data_format + + def forward(self, x): + gx, gy = add_coord(x, self.data_format) + if self.data_format == 'NCHW': + y = paddle.concat([x, gx, gy], axis=1) + else: + y = paddle.concat([x, gx, gy], axis=-1) + y = self.conv(y) + return y + + +class PPYOLODetBlock(nn.Layer): + def __init__(self, cfg, name, data_format='NCHW'): + """ + PPYOLODetBlock layer + + Args: + cfg (list): layer configs for this block + name (str): block name + data_format (str): data format, NCHW or NHWC + """ + super(PPYOLODetBlock, self).__init__() + self.conv_module = nn.Sequential() + for idx, (conv_name, layer, args, kwargs) in enumerate(cfg[:-1]): + kwargs.update( + name='{}.{}'.format(name, conv_name), data_format=data_format) + self.conv_module.add_sublayer(conv_name, layer(*args, **kwargs)) + + conv_name, layer, args, kwargs = cfg[-1] + kwargs.update( + name='{}.{}'.format(name, conv_name), data_format=data_format) + self.tip = layer(*args, **kwargs) + + def forward(self, inputs): + route = self.conv_module(inputs) + tip = self.tip(route) + return route, tip + + +class PPYOLOTinyDetBlock(nn.Layer): + def __init__(self, + ch_in, + ch_out, + name, + drop_block=False, + block_size=3, + keep_prob=0.9, + data_format='NCHW'): + """ + PPYOLO Tiny DetBlock layer + Args: + ch_in (list): input channel number + ch_out (list): output channel number + name (str): block name + drop_block: whether user DropBlock + block_size: drop block size + keep_prob: probability to keep block in DropBlock + data_format (str): data format, NCHW or NHWC + """ + super(PPYOLOTinyDetBlock, self).__init__() + self.drop_block_ = drop_block + self.conv_module = nn.Sequential() + + cfgs = [ + # name, in channels, out channels, filter_size, + # stride, padding, groups + ['.0', ch_in, ch_out, 1, 1, 0, 1], + ['.1', ch_out, ch_out, 5, 1, 2, ch_out], + ['.2', ch_out, ch_out, 1, 1, 0, 1], + ['.route', ch_out, ch_out, 5, 1, 2, ch_out], + ] + for cfg in cfgs: + conv_name, conv_ch_in, conv_ch_out, filter_size, stride, padding, \ + groups = cfg + self.conv_module.add_sublayer( + name + conv_name, + ConvBNLayer( + ch_in=conv_ch_in, + ch_out=conv_ch_out, + filter_size=filter_size, + stride=stride, + padding=padding, + groups=groups, + name=name + conv_name)) + + self.tip = ConvBNLayer( + ch_in=ch_out, + ch_out=ch_out, + filter_size=1, + stride=1, + padding=0, + groups=1, + name=name + conv_name) + + if self.drop_block_: + self.drop_block = DropBlock( + block_size=block_size, + keep_prob=keep_prob, + data_format=data_format, + name=name + '.dropblock') + + def forward(self, inputs): + if self.drop_block_: + inputs = self.drop_block(inputs) + route = self.conv_module(inputs) + tip = self.tip(route) + return route, tip + + +class PPYOLODetBlockCSP(nn.Layer): + def __init__(self, + cfg, + ch_in, + ch_out, + act, + norm_type, + name, + data_format='NCHW'): + """ + PPYOLODetBlockCSP layer + + Args: + cfg (list): layer configs for this block + ch_in (int): input channel + ch_out (int): output channel + act (str): default mish + name (str): block name + data_format (str): data format, NCHW or NHWC + """ + super(PPYOLODetBlockCSP, self).__init__() + self.data_format = data_format + self.conv1 = ConvBNLayer( + ch_in, + ch_out, + 1, + padding=0, + act=act, + norm_type=norm_type, + name=name + '.left', + data_format=data_format) + self.conv2 = ConvBNLayer( + ch_in, + ch_out, + 1, + padding=0, + act=act, + norm_type=norm_type, + name=name + '.right', + data_format=data_format) + self.conv3 = ConvBNLayer( + ch_out * 2, + ch_out * 2, + 1, + padding=0, + act=act, + norm_type=norm_type, + name=name, + data_format=data_format) + self.conv_module = nn.Sequential() + for idx, (layer_name, layer, args, kwargs) in enumerate(cfg): + kwargs.update(name=name + layer_name, data_format=data_format) + self.conv_module.add_sublayer(layer_name, layer(*args, **kwargs)) + + def forward(self, inputs): + conv_left = self.conv1(inputs) + conv_right = self.conv2(inputs) + conv_left = self.conv_module(conv_left) + if self.data_format == 'NCHW': + conv = paddle.concat([conv_left, conv_right], axis=1) + else: + conv = paddle.concat([conv_left, conv_right], axis=-1) + + conv = self.conv3(conv) + return conv, conv + + +@register +@serializable +class YOLOv3FPN(nn.Layer): + __shared__ = ['norm_type', 'data_format'] + + def __init__(self, + in_channels=[256, 512, 1024], + norm_type='bn', + freeze_norm=False, + data_format='NCHW'): + """ + YOLOv3FPN layer + + Args: + in_channels (list): input channels for fpn + norm_type (str): batch norm type, default bn + data_format (str): data format, NCHW or NHWC + + """ + super(YOLOv3FPN, self).__init__() + assert len(in_channels) > 0, "in_channels length should > 0" + self.in_channels = in_channels + self.num_blocks = len(in_channels) + + self._out_channels = [] + self.yolo_blocks = [] + self.routes = [] + self.data_format = data_format + for i in range(self.num_blocks): + name = 'yolo_block.{}'.format(i) + in_channel = in_channels[-i - 1] + if i > 0: + in_channel += 512 // (2**i) + yolo_block = self.add_sublayer( + name, + YoloDetBlock( + in_channel, + channel=512 // (2**i), + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name)) + self.yolo_blocks.append(yolo_block) + # tip layer output channel doubled + self._out_channels.append(1024 // (2**i)) + + if i < self.num_blocks - 1: + name = 'yolo_transition.{}'.format(i) + route = self.add_sublayer( + name, + ConvBNLayer( + ch_in=512 // (2**i), + ch_out=256 // (2**i), + filter_size=1, + stride=1, + padding=0, + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name)) + self.routes.append(route) + + def forward(self, blocks, for_mot=False): + assert len(blocks) == self.num_blocks + blocks = blocks[::-1] + yolo_feats = [] + + # add embedding features output for multi-object tracking model + if for_mot: + emb_feats = [] + + for i, block in enumerate(blocks): + if i > 0: + if self.data_format == 'NCHW': + block = paddle.concat([route, block], axis=1) + else: + block = paddle.concat([route, block], axis=-1) + route, tip = self.yolo_blocks[i](block) + yolo_feats.append(tip) + + if for_mot: + # add embedding features output + emb_feats.append(route) + + if i < self.num_blocks - 1: + route = self.routes[i](route) + route = F.interpolate( + route, scale_factor=2., data_format=self.data_format) + + if for_mot: + return {'yolo_feats': yolo_feats, 'emb_feats': emb_feats} + else: + return yolo_feats + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] + + +@register +@serializable +class PPYOLOFPN(nn.Layer): + __shared__ = ['norm_type', 'data_format'] + + def __init__(self, + in_channels=[512, 1024, 2048], + norm_type='bn', + freeze_norm=False, + data_format='NCHW', + coord_conv=False, + conv_block_num=2, + drop_block=False, + block_size=3, + keep_prob=0.9, + spp=False): + """ + PPYOLOFPN layer + + Args: + in_channels (list): input channels for fpn + norm_type (str): batch norm type, default bn + data_format (str): data format, NCHW or NHWC + coord_conv (bool): whether use CoordConv or not + conv_block_num (int): conv block num of each pan block + drop_block (bool): whether use DropBlock or not + block_size (int): block size of DropBlock + keep_prob (float): keep probability of DropBlock + spp (bool): whether use spp or not + + """ + super(PPYOLOFPN, self).__init__() + assert len(in_channels) > 0, "in_channels length should > 0" + self.in_channels = in_channels + self.num_blocks = len(in_channels) + # parse kwargs + self.coord_conv = coord_conv + self.drop_block = drop_block + self.block_size = block_size + self.keep_prob = keep_prob + self.spp = spp + self.conv_block_num = conv_block_num + self.data_format = data_format + if self.coord_conv: + ConvLayer = CoordConv + else: + ConvLayer = ConvBNLayer + + if self.drop_block: + dropblock_cfg = [[ + 'dropblock', DropBlock, [self.block_size, self.keep_prob], + dict() + ]] + else: + dropblock_cfg = [] + + self._out_channels = [] + self.yolo_blocks = [] + self.routes = [] + for i, ch_in in enumerate(self.in_channels[::-1]): + if i > 0: + ch_in += 512 // (2**i) + channel = 64 * (2**self.num_blocks) // (2**i) + base_cfg = [] + c_in, c_out = ch_in, channel + for j in range(self.conv_block_num): + base_cfg += [ + [ + 'conv{}'.format(2 * j), ConvLayer, [c_in, c_out, 1], + dict( + padding=0, + norm_type=norm_type, + freeze_norm=freeze_norm) + ], + [ + 'conv{}'.format(2 * j + 1), ConvBNLayer, + [c_out, c_out * 2, 3], dict( + padding=1, + norm_type=norm_type, + freeze_norm=freeze_norm) + ], + ] + c_in, c_out = c_out * 2, c_out + + base_cfg += [[ + 'route', ConvLayer, [c_in, c_out, 1], dict( + padding=0, norm_type=norm_type, freeze_norm=freeze_norm) + ], [ + 'tip', ConvLayer, [c_out, c_out * 2, 3], dict( + padding=1, norm_type=norm_type, freeze_norm=freeze_norm) + ]] + + if self.conv_block_num == 2: + if i == 0: + if self.spp: + spp_cfg = [[ + 'spp', SPP, [channel * 4, channel, 1], dict( + pool_size=[5, 9, 13], + norm_type=norm_type, + freeze_norm=freeze_norm) + ]] + else: + spp_cfg = [] + cfg = base_cfg[0:3] + spp_cfg + base_cfg[ + 3:4] + dropblock_cfg + base_cfg[4:6] + else: + cfg = base_cfg[0:2] + dropblock_cfg + base_cfg[2:6] + elif self.conv_block_num == 0: + if self.spp and i == 0: + spp_cfg = [[ + 'spp', SPP, [c_in * 4, c_in, 1], dict( + pool_size=[5, 9, 13], + norm_type=norm_type, + freeze_norm=freeze_norm) + ]] + else: + spp_cfg = [] + cfg = spp_cfg + dropblock_cfg + base_cfg + name = 'yolo_block.{}'.format(i) + yolo_block = self.add_sublayer(name, PPYOLODetBlock(cfg, name)) + self.yolo_blocks.append(yolo_block) + self._out_channels.append(channel * 2) + if i < self.num_blocks - 1: + name = 'yolo_transition.{}'.format(i) + route = self.add_sublayer( + name, + ConvBNLayer( + ch_in=channel, + ch_out=256 // (2**i), + filter_size=1, + stride=1, + padding=0, + norm_type=norm_type, + freeze_norm=freeze_norm, + data_format=data_format, + name=name)) + self.routes.append(route) + + def forward(self, blocks, for_mot=False): + assert len(blocks) == self.num_blocks + blocks = blocks[::-1] + yolo_feats = [] + + # add embedding features output for multi-object tracking model + if for_mot: + emb_feats = [] + + for i, block in enumerate(blocks): + if i > 0: + if self.data_format == 'NCHW': + block = paddle.concat([route, block], axis=1) + else: + block = paddle.concat([route, block], axis=-1) + route, tip = self.yolo_blocks[i](block) + yolo_feats.append(tip) + + if for_mot: + # add embedding features output + emb_feats.append(route) + + if i < self.num_blocks - 1: + route = self.routes[i](route) + route = F.interpolate( + route, scale_factor=2., data_format=self.data_format) + + if for_mot: + return {'yolo_feats': yolo_feats, 'emb_feats': emb_feats} + else: + return yolo_feats + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] + + +@register +@serializable +class PPYOLOTinyFPN(nn.Layer): + __shared__ = ['norm_type', 'data_format'] + + def __init__(self, + in_channels=[80, 56, 34], + detection_block_channels=[160, 128, 96], + norm_type='bn', + data_format='NCHW', + **kwargs): + """ + PPYOLO Tiny FPN layer + Args: + in_channels (list): input channels for fpn + detection_block_channels (list): channels in fpn + norm_type (str): batch norm type, default bn + data_format (str): data format, NCHW or NHWC + kwargs: extra key-value pairs, such as parameter of DropBlock and spp + """ + super(PPYOLOTinyFPN, self).__init__() + assert len(in_channels) > 0, "in_channels length should > 0" + self.in_channels = in_channels[::-1] + assert len(detection_block_channels + ) > 0, "detection_block_channelslength should > 0" + self.detection_block_channels = detection_block_channels + self.data_format = data_format + self.num_blocks = len(in_channels) + # parse kwargs + self.drop_block = kwargs.get('drop_block', False) + self.block_size = kwargs.get('block_size', 3) + self.keep_prob = kwargs.get('keep_prob', 0.9) + + self.spp_ = kwargs.get('spp', False) + if self.spp_: + self.spp = SPP(self.in_channels[0] * 4, + self.in_channels[0], + k=1, + pool_size=[5, 9, 13], + norm_type=norm_type, + name='spp') + + self._out_channels = [] + self.yolo_blocks = [] + self.routes = [] + for i, ( + ch_in, ch_out + ) in enumerate(zip(self.in_channels, self.detection_block_channels)): + name = 'yolo_block.{}'.format(i) + if i > 0: + ch_in += self.detection_block_channels[i - 1] + yolo_block = self.add_sublayer( + name, + PPYOLOTinyDetBlock( + ch_in, + ch_out, + name, + drop_block=self.drop_block, + block_size=self.block_size, + keep_prob=self.keep_prob)) + self.yolo_blocks.append(yolo_block) + self._out_channels.append(ch_out) + + if i < self.num_blocks - 1: + name = 'yolo_transition.{}'.format(i) + route = self.add_sublayer( + name, + ConvBNLayer( + ch_in=ch_out, + ch_out=ch_out, + filter_size=1, + stride=1, + padding=0, + norm_type=norm_type, + data_format=data_format, + name=name)) + self.routes.append(route) + + def forward(self, blocks, for_mot=False): + assert len(blocks) == self.num_blocks + blocks = blocks[::-1] + yolo_feats = [] + + # add embedding features output for multi-object tracking model + if for_mot: + emb_feats = [] + + for i, block in enumerate(blocks): + if i == 0 and self.spp_: + block = self.spp(block) + + if i > 0: + if self.data_format == 'NCHW': + block = paddle.concat([route, block], axis=1) + else: + block = paddle.concat([route, block], axis=-1) + route, tip = self.yolo_blocks[i](block) + yolo_feats.append(tip) + + if for_mot: + # add embedding features output + emb_feats.append(route) + + if i < self.num_blocks - 1: + route = self.routes[i](route) + route = F.interpolate( + route, scale_factor=2., data_format=self.data_format) + + if for_mot: + return {'yolo_feats': yolo_feats, 'emb_feats': emb_feats} + else: + return yolo_feats + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] + + +@register +@serializable +class PPYOLOPAN(nn.Layer): + __shared__ = ['norm_type', 'data_format'] + + def __init__(self, + in_channels=[512, 1024, 2048], + norm_type='bn', + data_format='NCHW', + act='mish', + conv_block_num=3, + drop_block=False, + block_size=3, + keep_prob=0.9, + spp=False): + """ + PPYOLOPAN layer with SPP, DropBlock and CSP connection. + + Args: + in_channels (list): input channels for fpn + norm_type (str): batch norm type, default bn + data_format (str): data format, NCHW or NHWC + act (str): activation function, default mish + conv_block_num (int): conv block num of each pan block + drop_block (bool): whether use DropBlock or not + block_size (int): block size of DropBlock + keep_prob (float): keep probability of DropBlock + spp (bool): whether use spp or not + + """ + super(PPYOLOPAN, self).__init__() + assert len(in_channels) > 0, "in_channels length should > 0" + self.in_channels = in_channels + self.num_blocks = len(in_channels) + # parse kwargs + self.drop_block = drop_block + self.block_size = block_size + self.keep_prob = keep_prob + self.spp = spp + self.conv_block_num = conv_block_num + self.data_format = data_format + if self.drop_block: + dropblock_cfg = [[ + 'dropblock', DropBlock, [self.block_size, self.keep_prob], + dict() + ]] + else: + dropblock_cfg = [] + + # fpn + self.fpn_blocks = [] + self.fpn_routes = [] + fpn_channels = [] + for i, ch_in in enumerate(self.in_channels[::-1]): + if i > 0: + ch_in += 512 // (2**(i - 1)) + channel = 512 // (2**i) + base_cfg = [] + for j in range(self.conv_block_num): + base_cfg += [ + # name, layer, args + [ + '{}.0'.format(j), ConvBNLayer, [channel, channel, 1], + dict( + padding=0, act=act, norm_type=norm_type) + ], + [ + '{}.1'.format(j), ConvBNLayer, [channel, channel, 3], + dict( + padding=1, act=act, norm_type=norm_type) + ] + ] + + if i == 0 and self.spp: + base_cfg[3] = [ + 'spp', SPP, [channel * 4, channel, 1], dict( + pool_size=[5, 9, 13], act=act, norm_type=norm_type) + ] + + cfg = base_cfg[:4] + dropblock_cfg + base_cfg[4:] + name = 'fpn.{}'.format(i) + fpn_block = self.add_sublayer( + name, + PPYOLODetBlockCSP(cfg, ch_in, channel, act, norm_type, name, + data_format)) + self.fpn_blocks.append(fpn_block) + fpn_channels.append(channel * 2) + if i < self.num_blocks - 1: + name = 'fpn_transition.{}'.format(i) + route = self.add_sublayer( + name, + ConvBNLayer( + ch_in=channel * 2, + ch_out=channel, + filter_size=1, + stride=1, + padding=0, + act=act, + norm_type=norm_type, + data_format=data_format, + name=name)) + self.fpn_routes.append(route) + # pan + self.pan_blocks = [] + self.pan_routes = [] + self._out_channels = [512 // (2**(self.num_blocks - 2)), ] + for i in reversed(range(self.num_blocks - 1)): + name = 'pan_transition.{}'.format(i) + route = self.add_sublayer( + name, + ConvBNLayer( + ch_in=fpn_channels[i + 1], + ch_out=fpn_channels[i + 1], + filter_size=3, + stride=2, + padding=1, + act=act, + norm_type=norm_type, + data_format=data_format, + name=name)) + self.pan_routes = [route, ] + self.pan_routes + base_cfg = [] + ch_in = fpn_channels[i] + fpn_channels[i + 1] + channel = 512 // (2**i) + for j in range(self.conv_block_num): + base_cfg += [ + # name, layer, args + [ + '{}.0'.format(j), ConvBNLayer, [channel, channel, 1], + dict( + padding=0, act=act, norm_type=norm_type) + ], + [ + '{}.1'.format(j), ConvBNLayer, [channel, channel, 3], + dict( + padding=1, act=act, norm_type=norm_type) + ] + ] + + cfg = base_cfg[:4] + dropblock_cfg + base_cfg[4:] + name = 'pan.{}'.format(i) + pan_block = self.add_sublayer( + name, + PPYOLODetBlockCSP(cfg, ch_in, channel, act, norm_type, name, + data_format)) + + self.pan_blocks = [pan_block, ] + self.pan_blocks + self._out_channels.append(channel * 2) + + self._out_channels = self._out_channels[::-1] + + def forward(self, blocks, for_mot=False): + assert len(blocks) == self.num_blocks + blocks = blocks[::-1] + fpn_feats = [] + + # add embedding features output for multi-object tracking model + if for_mot: + emb_feats = [] + + for i, block in enumerate(blocks): + if i > 0: + if self.data_format == 'NCHW': + block = paddle.concat([route, block], axis=1) + else: + block = paddle.concat([route, block], axis=-1) + route, tip = self.fpn_blocks[i](block) + fpn_feats.append(tip) + + if for_mot: + # add embedding features output + emb_feats.append(route) + + if i < self.num_blocks - 1: + route = self.fpn_routes[i](route) + route = F.interpolate( + route, scale_factor=2., data_format=self.data_format) + + pan_feats = [fpn_feats[-1], ] + route = fpn_feats[self.num_blocks - 1] + for i in reversed(range(self.num_blocks - 1)): + block = fpn_feats[i] + route = self.pan_routes[i](route) + if self.data_format == 'NCHW': + block = paddle.concat([route, block], axis=1) + else: + block = paddle.concat([route, block], axis=-1) + + route, tip = self.pan_blocks[i](block) + pan_feats.append(tip) + + if for_mot: + return {'yolo_feats': pan_feats[::-1], 'emb_feats': emb_feats} + else: + return pan_feats[::-1] + + @classmethod + def from_config(cls, cfg, input_shape): + return {'in_channels': [i.channels for i in input_shape], } + + @property + def out_shape(self): + return [ShapeSpec(channels=c) for c in self._out_channels] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/ops.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/ops.py new file mode 100644 index 000000000..593d8dd37 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/ops.py @@ -0,0 +1,1601 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn.functional as F +import paddle.nn as nn +from paddle import ParamAttr +from paddle.regularizer import L2Decay + +from paddle.fluid.framework import Variable, in_dygraph_mode +from paddle.fluid import core +from paddle.fluid.layer_helper import LayerHelper +from paddle.fluid.data_feeder import check_variable_and_dtype, check_type, check_dtype + +__all__ = [ + 'roi_pool', + 'roi_align', + 'prior_box', + 'generate_proposals', + 'iou_similarity', + 'box_coder', + 'yolo_box', + 'multiclass_nms', + 'distribute_fpn_proposals', + 'collect_fpn_proposals', + 'matrix_nms', + 'batch_norm', + 'mish', +] + + +def mish(x): + return x * paddle.tanh(F.softplus(x)) + + +def batch_norm(ch, + norm_type='bn', + norm_decay=0., + freeze_norm=False, + initializer=None, + data_format='NCHW'): + if norm_type == 'sync_bn': + batch_norm = nn.SyncBatchNorm + else: + batch_norm = nn.BatchNorm2D + + norm_lr = 0. if freeze_norm else 1. + weight_attr = ParamAttr( + initializer=initializer, + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + bias_attr = ParamAttr( + learning_rate=norm_lr, + regularizer=L2Decay(norm_decay), + trainable=False if freeze_norm else True) + + norm_layer = batch_norm( + ch, + weight_attr=weight_attr, + bias_attr=bias_attr, + data_format=data_format) + + norm_params = norm_layer.parameters() + if freeze_norm: + for param in norm_params: + param.stop_gradient = True + + return norm_layer + + +@paddle.jit.not_to_static +def roi_pool(input, + rois, + output_size, + spatial_scale=1.0, + rois_num=None, + name=None): + """ + + This operator implements the roi_pooling layer. + Region of interest pooling (also known as RoI pooling) is to perform max pooling on inputs of nonuniform sizes to obtain fixed-size feature maps (e.g. 7*7). + + The operator has three steps: + + 1. Dividing each region proposal into equal-sized sections with output_size(h, w); + 2. Finding the largest value in each section; + 3. Copying these max values to the output buffer. + + For more information, please refer to https://stackoverflow.com/questions/43430056/what-is-roi-layer-in-fast-rcnn + + Args: + input (Tensor): Input feature, 4D-Tensor with the shape of [N,C,H,W], + where N is the batch size, C is the input channel, H is Height, W is weight. + The data type is float32 or float64. + rois (Tensor): ROIs (Regions of Interest) to pool over. + 2D-Tensor or 2D-LoDTensor with the shape of [num_rois,4], the lod level is 1. + Given as [[x1, y1, x2, y2], ...], (x1, y1) is the top left coordinates, + and (x2, y2) is the bottom right coordinates. + output_size (int or tuple[int, int]): The pooled output size(h, w), data type is int32. If int, h and w are both equal to output_size. + spatial_scale (float, optional): Multiplicative spatial scale factor to translate ROI coords from their input scale to the scale used when pooling. Default: 1.0 + rois_num (Tensor): The number of RoIs in each image. Default: None + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + + Returns: + Tensor: The pooled feature, 4D-Tensor with the shape of [num_rois, C, output_size[0], output_size[1]]. + + + Examples: + + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + + x = paddle.static.data( + name='data', shape=[None, 256, 32, 32], dtype='float32') + rois = paddle.static.data( + name='rois', shape=[None, 4], dtype='float32') + rois_num = paddle.static.data(name='rois_num', shape=[None], dtype='int32') + + pool_out = ops.roi_pool( + input=x, + rois=rois, + output_size=(1, 1), + spatial_scale=1.0, + rois_num=rois_num) + """ + check_type(output_size, 'output_size', (int, tuple), 'roi_pool') + if isinstance(output_size, int): + output_size = (output_size, output_size) + + pooled_height, pooled_width = output_size + if in_dygraph_mode(): + assert rois_num is not None, "rois_num should not be None in dygraph mode." + pool_out, argmaxes = core.ops.roi_pool( + input, rois, rois_num, "pooled_height", pooled_height, + "pooled_width", pooled_width, "spatial_scale", spatial_scale) + return pool_out, argmaxes + + else: + check_variable_and_dtype(input, 'input', ['float32'], 'roi_pool') + check_variable_and_dtype(rois, 'rois', ['float32'], 'roi_pool') + helper = LayerHelper('roi_pool', **locals()) + dtype = helper.input_dtype() + pool_out = helper.create_variable_for_type_inference(dtype) + argmaxes = helper.create_variable_for_type_inference(dtype='int32') + + inputs = { + "X": input, + "ROIs": rois, + } + if rois_num is not None: + inputs['RoisNum'] = rois_num + helper.append_op( + type="roi_pool", + inputs=inputs, + outputs={"Out": pool_out, + "Argmax": argmaxes}, + attrs={ + "pooled_height": pooled_height, + "pooled_width": pooled_width, + "spatial_scale": spatial_scale + }) + return pool_out, argmaxes + + +@paddle.jit.not_to_static +def roi_align(input, + rois, + output_size, + spatial_scale=1.0, + sampling_ratio=-1, + rois_num=None, + aligned=True, + name=None): + """ + + Region of interest align (also known as RoI align) is to perform + bilinear interpolation on inputs of nonuniform sizes to obtain + fixed-size feature maps (e.g. 7*7) + + Dividing each region proposal into equal-sized sections with + the pooled_width and pooled_height. Location remains the origin + result. + + In each ROI bin, the value of the four regularly sampled locations + are computed directly through bilinear interpolation. The output is + the mean of four locations. + Thus avoid the misaligned problem. + + Args: + input (Tensor): Input feature, 4D-Tensor with the shape of [N,C,H,W], + where N is the batch size, C is the input channel, H is Height, W is weight. + The data type is float32 or float64. + rois (Tensor): ROIs (Regions of Interest) to pool over.It should be + a 2-D Tensor or 2-D LoDTensor of shape (num_rois, 4), the lod level is 1. + The data type is float32 or float64. Given as [[x1, y1, x2, y2], ...], + (x1, y1) is the top left coordinates, and (x2, y2) is the bottom right coordinates. + output_size (int or tuple[int, int]): The pooled output size(h, w), data type is int32. If int, h and w are both equal to output_size. + spatial_scale (float32, optional): Multiplicative spatial scale factor to translate ROI coords + from their input scale to the scale used when pooling. Default: 1.0 + sampling_ratio(int32, optional): number of sampling points in the interpolation grid. + If <=0, then grid points are adaptive to roi_width and pooled_w, likewise for height. Default: -1 + rois_num (Tensor): The number of RoIs in each image. Default: None + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Tensor: + + Output: The output of ROIAlignOp is a 4-D tensor with shape (num_rois, channels, pooled_h, pooled_w). The data type is float32 or float64. + + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + + x = paddle.static.data( + name='data', shape=[None, 256, 32, 32], dtype='float32') + rois = paddle.static.data( + name='rois', shape=[None, 4], dtype='float32') + rois_num = paddle.static.data(name='rois_num', shape=[None], dtype='int32') + align_out = ops.roi_align(input=x, + rois=rois, + ouput_size=(7, 7), + spatial_scale=0.5, + sampling_ratio=-1, + rois_num=rois_num) + """ + check_type(output_size, 'output_size', (int, tuple), 'roi_align') + if isinstance(output_size, int): + output_size = (output_size, output_size) + + pooled_height, pooled_width = output_size + + if in_dygraph_mode(): + assert rois_num is not None, "rois_num should not be None in dygraph mode." + align_out = core.ops.roi_align( + input, rois, rois_num, "pooled_height", pooled_height, + "pooled_width", pooled_width, "spatial_scale", spatial_scale, + "sampling_ratio", sampling_ratio, "aligned", aligned) + return align_out + + else: + check_variable_and_dtype(input, 'input', ['float32', 'float64'], + 'roi_align') + check_variable_and_dtype(rois, 'rois', ['float32', 'float64'], + 'roi_align') + helper = LayerHelper('roi_align', **locals()) + dtype = helper.input_dtype() + align_out = helper.create_variable_for_type_inference(dtype) + inputs = { + "X": input, + "ROIs": rois, + } + if rois_num is not None: + inputs['RoisNum'] = rois_num + helper.append_op( + type="roi_align", + inputs=inputs, + outputs={"Out": align_out}, + attrs={ + "pooled_height": pooled_height, + "pooled_width": pooled_width, + "spatial_scale": spatial_scale, + "sampling_ratio": sampling_ratio, + "aligned": aligned, + }) + return align_out + + +@paddle.jit.not_to_static +def iou_similarity(x, y, box_normalized=True, name=None): + """ + Computes intersection-over-union (IOU) between two box lists. + Box list 'X' should be a LoDTensor and 'Y' is a common Tensor, + boxes in 'Y' are shared by all instance of the batched inputs of X. + Given two boxes A and B, the calculation of IOU is as follows: + + $$ + IOU(A, B) = + \\frac{area(A\\cap B)}{area(A)+area(B)-area(A\\cap B)} + $$ + + Args: + x (Tensor): Box list X is a 2-D Tensor with shape [N, 4] holds N + boxes, each box is represented as [xmin, ymin, xmax, ymax], + the shape of X is [N, 4]. [xmin, ymin] is the left top + coordinate of the box if the input is image feature map, they + are close to the origin of the coordinate system. + [xmax, ymax] is the right bottom coordinate of the box. + The data type is float32 or float64. + y (Tensor): Box list Y holds M boxes, each box is represented as + [xmin, ymin, xmax, ymax], the shape of X is [N, 4]. + [xmin, ymin] is the left top coordinate of the box if the + input is image feature map, and [xmax, ymax] is the right + bottom coordinate of the box. The data type is float32 or float64. + box_normalized(bool): Whether treat the priorbox as a normalized box. + Set true by default. + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Tensor: The output of iou_similarity op, a tensor with shape [N, M] + representing pairwise iou scores. The data type is same with x. + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + + x = paddle.static.data(name='x', shape=[None, 4], dtype='float32') + y = paddle.static.data(name='y', shape=[None, 4], dtype='float32') + iou = ops.iou_similarity(x=x, y=y) + """ + + if in_dygraph_mode(): + out = core.ops.iou_similarity(x, y, 'box_normalized', box_normalized) + return out + else: + helper = LayerHelper("iou_similarity", **locals()) + out = helper.create_variable_for_type_inference(dtype=x.dtype) + + helper.append_op( + type="iou_similarity", + inputs={"X": x, + "Y": y}, + attrs={"box_normalized": box_normalized}, + outputs={"Out": out}) + return out + + +@paddle.jit.not_to_static +def collect_fpn_proposals(multi_rois, + multi_scores, + min_level, + max_level, + post_nms_top_n, + rois_num_per_level=None, + name=None): + """ + + **This OP only supports LoDTensor as input**. Concat multi-level RoIs + (Region of Interest) and select N RoIs with respect to multi_scores. + This operation performs the following steps: + + 1. Choose num_level RoIs and scores as input: num_level = max_level - min_level + 2. Concat multi-level RoIs and scores + 3. Sort scores and select post_nms_top_n scores + 4. Gather RoIs by selected indices from scores + 5. Re-sort RoIs by corresponding batch_id + + Args: + multi_rois(list): List of RoIs to collect. Element in list is 2-D + LoDTensor with shape [N, 4] and data type is float32 or float64, + N is the number of RoIs. + multi_scores(list): List of scores of RoIs to collect. Element in list + is 2-D LoDTensor with shape [N, 1] and data type is float32 or + float64, N is the number of RoIs. + min_level(int): The lowest level of FPN layer to collect + max_level(int): The highest level of FPN layer to collect + post_nms_top_n(int): The number of selected RoIs + rois_num_per_level(list, optional): The List of RoIs' numbers. + Each element is 1-D Tensor which contains the RoIs' number of each + image on each level and the shape is [B] and data type is + int32, B is the number of images. If it is not None then return + a 1-D Tensor contains the output RoIs' number of each image and + the shape is [B]. Default: None + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Variable: + + fpn_rois(Variable): 2-D LoDTensor with shape [N, 4] and data type is + float32 or float64. Selected RoIs. + + rois_num(Tensor): 1-D Tensor contains the RoIs's number of each + image. The shape is [B] and data type is int32. B is the number of + images. + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + multi_rois = [] + multi_scores = [] + for i in range(4): + multi_rois.append(paddle.static.data( + name='roi_'+str(i), shape=[None, 4], dtype='float32', lod_level=1)) + for i in range(4): + multi_scores.append(paddle.static.data( + name='score_'+str(i), shape=[None, 1], dtype='float32', lod_level=1)) + + fpn_rois = ops.collect_fpn_proposals( + multi_rois=multi_rois, + multi_scores=multi_scores, + min_level=2, + max_level=5, + post_nms_top_n=2000) + """ + check_type(multi_rois, 'multi_rois', list, 'collect_fpn_proposals') + check_type(multi_scores, 'multi_scores', list, 'collect_fpn_proposals') + num_lvl = max_level - min_level + 1 + input_rois = multi_rois[:num_lvl] + input_scores = multi_scores[:num_lvl] + + if in_dygraph_mode(): + assert rois_num_per_level is not None, "rois_num_per_level should not be None in dygraph mode." + attrs = ('post_nms_topN', post_nms_top_n) + output_rois, rois_num = core.ops.collect_fpn_proposals( + input_rois, input_scores, rois_num_per_level, *attrs) + return output_rois, rois_num + + else: + helper = LayerHelper('collect_fpn_proposals', **locals()) + dtype = helper.input_dtype('multi_rois') + check_dtype(dtype, 'multi_rois', ['float32', 'float64'], + 'collect_fpn_proposals') + output_rois = helper.create_variable_for_type_inference(dtype) + output_rois.stop_gradient = True + + inputs = { + 'MultiLevelRois': input_rois, + 'MultiLevelScores': input_scores, + } + outputs = {'FpnRois': output_rois} + if rois_num_per_level is not None: + inputs['MultiLevelRoIsNum'] = rois_num_per_level + rois_num = helper.create_variable_for_type_inference(dtype='int32') + rois_num.stop_gradient = True + outputs['RoisNum'] = rois_num + helper.append_op( + type='collect_fpn_proposals', + inputs=inputs, + outputs=outputs, + attrs={'post_nms_topN': post_nms_top_n}) + return output_rois, rois_num + + +@paddle.jit.not_to_static +def distribute_fpn_proposals(fpn_rois, + min_level, + max_level, + refer_level, + refer_scale, + pixel_offset=False, + rois_num=None, + name=None): + """ + + **This op only takes LoDTensor as input.** In Feature Pyramid Networks + (FPN) models, it is needed to distribute all proposals into different FPN + level, with respect to scale of the proposals, the referring scale and the + referring level. Besides, to restore the order of proposals, we return an + array which indicates the original index of rois in current proposals. + To compute FPN level for each roi, the formula is given as follows: + + .. math:: + + roi\_scale &= \sqrt{BBoxArea(fpn\_roi)} + + level = floor(&\log(\\frac{roi\_scale}{refer\_scale}) + refer\_level) + + where BBoxArea is a function to compute the area of each roi. + + Args: + + fpn_rois(Variable): 2-D Tensor with shape [N, 4] and data type is + float32 or float64. The input fpn_rois. + min_level(int32): The lowest level of FPN layer where the proposals come + from. + max_level(int32): The highest level of FPN layer where the proposals + come from. + refer_level(int32): The referring level of FPN layer with specified scale. + refer_scale(int32): The referring scale of FPN layer with specified level. + rois_num(Tensor): 1-D Tensor contains the number of RoIs in each image. + The shape is [B] and data type is int32. B is the number of images. + If it is not None then return a list of 1-D Tensor. Each element + is the output RoIs' number of each image on the corresponding level + and the shape is [B]. None by default. + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Tuple: + + multi_rois(List) : A list of 2-D LoDTensor with shape [M, 4] + and data type of float32 and float64. The length is + max_level-min_level+1. The proposals in each FPN level. + + restore_ind(Variable): A 2-D Tensor with shape [N, 1], N is + the number of total rois. The data type is int32. It is + used to restore the order of fpn_rois. + + rois_num_per_level(List): A list of 1-D Tensor and each Tensor is + the RoIs' number in each image on the corresponding level. The shape + is [B] and data type of int32. B is the number of images + + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + fpn_rois = paddle.static.data( + name='data', shape=[None, 4], dtype='float32', lod_level=1) + multi_rois, restore_ind = ops.distribute_fpn_proposals( + fpn_rois=fpn_rois, + min_level=2, + max_level=5, + refer_level=4, + refer_scale=224) + """ + num_lvl = max_level - min_level + 1 + + if in_dygraph_mode(): + assert rois_num is not None, "rois_num should not be None in dygraph mode." + attrs = ('min_level', min_level, 'max_level', max_level, 'refer_level', + refer_level, 'refer_scale', refer_scale, 'pixel_offset', + pixel_offset) + multi_rois, restore_ind, rois_num_per_level = core.ops.distribute_fpn_proposals( + fpn_rois, rois_num, num_lvl, num_lvl, *attrs) + return multi_rois, restore_ind, rois_num_per_level + + else: + check_variable_and_dtype(fpn_rois, 'fpn_rois', ['float32', 'float64'], + 'distribute_fpn_proposals') + helper = LayerHelper('distribute_fpn_proposals', **locals()) + dtype = helper.input_dtype('fpn_rois') + multi_rois = [ + helper.create_variable_for_type_inference(dtype) + for i in range(num_lvl) + ] + + restore_ind = helper.create_variable_for_type_inference(dtype='int32') + + inputs = {'FpnRois': fpn_rois} + outputs = { + 'MultiFpnRois': multi_rois, + 'RestoreIndex': restore_ind, + } + + if rois_num is not None: + inputs['RoisNum'] = rois_num + rois_num_per_level = [ + helper.create_variable_for_type_inference(dtype='int32') + for i in range(num_lvl) + ] + outputs['MultiLevelRoIsNum'] = rois_num_per_level + + helper.append_op( + type='distribute_fpn_proposals', + inputs=inputs, + outputs=outputs, + attrs={ + 'min_level': min_level, + 'max_level': max_level, + 'refer_level': refer_level, + 'refer_scale': refer_scale, + 'pixel_offset': pixel_offset + }) + return multi_rois, restore_ind, rois_num_per_level + + +@paddle.jit.not_to_static +def yolo_box( + x, + origin_shape, + anchors, + class_num, + conf_thresh, + downsample_ratio, + clip_bbox=True, + scale_x_y=1., + name=None, ): + """ + + This operator generates YOLO detection boxes from output of YOLOv3 network. + + The output of previous network is in shape [N, C, H, W], while H and W + should be the same, H and W specify the grid size, each grid point predict + given number boxes, this given number, which following will be represented as S, + is specified by the number of anchors. In the second dimension(the channel + dimension), C should be equal to S * (5 + class_num), class_num is the object + category number of source dataset(such as 80 in coco dataset), so the + second(channel) dimension, apart from 4 box location coordinates x, y, w, h, + also includes confidence score of the box and class one-hot key of each anchor + box. + Assume the 4 location coordinates are :math:`t_x, t_y, t_w, t_h`, the box + predictions should be as follows: + $$ + b_x = \\sigma(t_x) + c_x + $$ + $$ + b_y = \\sigma(t_y) + c_y + $$ + $$ + b_w = p_w e^{t_w} + $$ + $$ + b_h = p_h e^{t_h} + $$ + in the equation above, :math:`c_x, c_y` is the left top corner of current grid + and :math:`p_w, p_h` is specified by anchors. + The logistic regression value of the 5th channel of each anchor prediction boxes + represents the confidence score of each prediction box, and the logistic + regression value of the last :attr:`class_num` channels of each anchor prediction + boxes represents the classifcation scores. Boxes with confidence scores less than + :attr:`conf_thresh` should be ignored, and box final scores is the product of + confidence scores and classification scores. + $$ + score_{pred} = score_{conf} * score_{class} + $$ + + Args: + x (Tensor): The input tensor of YoloBox operator is a 4-D tensor with shape of [N, C, H, W]. + The second dimension(C) stores box locations, confidence score and + classification one-hot keys of each anchor box. Generally, X should be the output of YOLOv3 network. + The data type is float32 or float64. + origin_shape (Tensor): The image size tensor of YoloBox operator, This is a 2-D tensor with shape of [N, 2]. + This tensor holds height and width of each input image used for resizing output box in input image + scale. The data type is int32. + anchors (list|tuple): The anchor width and height, it will be parsed pair by pair. + class_num (int): The number of classes to predict. + conf_thresh (float): The confidence scores threshold of detection boxes. Boxes with confidence scores + under threshold should be ignored. + downsample_ratio (int): The downsample ratio from network input to YoloBox operator input, + so 32, 16, 8 should be set for the first, second, and thrid YoloBox operators. + clip_bbox (bool): Whether clip output bonding box in Input(ImgSize) boundary. Default true. + scale_x_y (float): Scale the center point of decoded bounding box. Default 1.0. + name (string): The default value is None. Normally there is no need + for user to set this property. For more information, + please refer to :ref:`api_guide_Name` + + Returns: + boxes Tensor: A 3-D tensor with shape [N, M, 4], the coordinates of boxes, N is the batch num, + M is output box number, and the 3rd dimension stores [xmin, ymin, xmax, ymax] coordinates of boxes. + scores Tensor: A 3-D tensor with shape [N, M, :attr:`class_num`], the coordinates of boxes, N is the batch num, + M is output box number. + + Raises: + TypeError: Attr anchors of yolo box must be list or tuple + TypeError: Attr class_num of yolo box must be an integer + TypeError: Attr conf_thresh of yolo box must be a float number + + Examples: + + .. code-block:: python + + import paddle + from ppdet.modeling import ops + + paddle.enable_static() + x = paddle.static.data(name='x', shape=[None, 255, 13, 13], dtype='float32') + img_size = paddle.static.data(name='img_size',shape=[None, 2],dtype='int64') + anchors = [10, 13, 16, 30, 33, 23] + boxes,scores = ops.yolo_box(x=x, img_size=img_size, class_num=80, anchors=anchors, + conf_thresh=0.01, downsample_ratio=32) + """ + helper = LayerHelper('yolo_box', **locals()) + + if not isinstance(anchors, list) and not isinstance(anchors, tuple): + raise TypeError("Attr anchors of yolo_box must be list or tuple") + if not isinstance(class_num, int): + raise TypeError("Attr class_num of yolo_box must be an integer") + if not isinstance(conf_thresh, float): + raise TypeError("Attr ignore_thresh of yolo_box must be a float number") + + if in_dygraph_mode(): + attrs = ('anchors', anchors, 'class_num', class_num, 'conf_thresh', + conf_thresh, 'downsample_ratio', downsample_ratio, 'clip_bbox', + clip_bbox, 'scale_x_y', scale_x_y) + boxes, scores = core.ops.yolo_box(x, origin_shape, *attrs) + return boxes, scores + else: + boxes = helper.create_variable_for_type_inference(dtype=x.dtype) + scores = helper.create_variable_for_type_inference(dtype=x.dtype) + + attrs = { + "anchors": anchors, + "class_num": class_num, + "conf_thresh": conf_thresh, + "downsample_ratio": downsample_ratio, + "clip_bbox": clip_bbox, + "scale_x_y": scale_x_y, + } + + helper.append_op( + type='yolo_box', + inputs={ + "X": x, + "ImgSize": origin_shape, + }, + outputs={ + 'Boxes': boxes, + 'Scores': scores, + }, + attrs=attrs) + return boxes, scores + + +@paddle.jit.not_to_static +def prior_box(input, + image, + min_sizes, + max_sizes=None, + aspect_ratios=[1.], + variance=[0.1, 0.1, 0.2, 0.2], + flip=False, + clip=False, + steps=[0.0, 0.0], + offset=0.5, + min_max_aspect_ratios_order=False, + name=None): + """ + + This op generates prior boxes for SSD(Single Shot MultiBox Detector) algorithm. + Each position of the input produce N prior boxes, N is determined by + the count of min_sizes, max_sizes and aspect_ratios, The size of the + box is in range(min_size, max_size) interval, which is generated in + sequence according to the aspect_ratios. + + Parameters: + input(Tensor): 4-D tensor(NCHW), the data type should be float32 or float64. + image(Tensor): 4-D tensor(NCHW), the input image data of PriorBoxOp, + the data type should be float32 or float64. + min_sizes(list|tuple|float): the min sizes of generated prior boxes. + max_sizes(list|tuple|None): the max sizes of generated prior boxes. + Default: None. + aspect_ratios(list|tuple|float): the aspect ratios of generated + prior boxes. Default: [1.]. + variance(list|tuple): the variances to be encoded in prior boxes. + Default:[0.1, 0.1, 0.2, 0.2]. + flip(bool): Whether to flip aspect ratios. Default:False. + clip(bool): Whether to clip out-of-boundary boxes. Default: False. + step(list|tuple): Prior boxes step across width and height, If + step[0] equals to 0.0 or step[1] equals to 0.0, the prior boxes step across + height or weight of the input will be automatically calculated. + Default: [0., 0.] + offset(float): Prior boxes center offset. Default: 0.5 + min_max_aspect_ratios_order(bool): If set True, the output prior box is + in order of [min, max, aspect_ratios], which is consistent with + Caffe. Please note, this order affects the weights order of + convolution layer followed by and does not affect the final + detection results. Default: False. + name(str, optional): The default value is None. Normally there is no need for + user to set this property. For more information, please refer to :ref:`api_guide_Name` + + Returns: + Tuple: A tuple with two Variable (boxes, variances) + + boxes(Tensor): the output prior boxes of PriorBox. + 4-D tensor, the layout is [H, W, num_priors, 4]. + H is the height of input, W is the width of input, + num_priors is the total box count of each position of input. + + variances(Tensor): the expanded variances of PriorBox. + 4-D tensor, the layput is [H, W, num_priors, 4]. + H is the height of input, W is the width of input + num_priors is the total box count of each position of input + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + + paddle.enable_static() + input = paddle.static.data(name="input", shape=[None,3,6,9]) + image = paddle.static.data(name="image", shape=[None,3,9,12]) + box, var = ops.prior_box( + input=input, + image=image, + min_sizes=[100.], + clip=True, + flip=True) + """ + helper = LayerHelper("prior_box", **locals()) + dtype = helper.input_dtype() + check_variable_and_dtype( + input, 'input', ['uint8', 'int8', 'float32', 'float64'], 'prior_box') + + def _is_list_or_tuple_(data): + return (isinstance(data, list) or isinstance(data, tuple)) + + if not _is_list_or_tuple_(min_sizes): + min_sizes = [min_sizes] + if not _is_list_or_tuple_(aspect_ratios): + aspect_ratios = [aspect_ratios] + if not (_is_list_or_tuple_(steps) and len(steps) == 2): + raise ValueError('steps should be a list or tuple ', + 'with length 2, (step_width, step_height).') + + min_sizes = list(map(float, min_sizes)) + aspect_ratios = list(map(float, aspect_ratios)) + steps = list(map(float, steps)) + + cur_max_sizes = None + if max_sizes is not None and len(max_sizes) > 0 and max_sizes[0] > 0: + if not _is_list_or_tuple_(max_sizes): + max_sizes = [max_sizes] + cur_max_sizes = max_sizes + + if in_dygraph_mode(): + attrs = ('min_sizes', min_sizes, 'aspect_ratios', aspect_ratios, + 'variances', variance, 'flip', flip, 'clip', clip, 'step_w', + steps[0], 'step_h', steps[1], 'offset', offset, + 'min_max_aspect_ratios_order', min_max_aspect_ratios_order) + if cur_max_sizes is not None: + attrs += ('max_sizes', cur_max_sizes) + box, var = core.ops.prior_box(input, image, *attrs) + return box, var + else: + attrs = { + 'min_sizes': min_sizes, + 'aspect_ratios': aspect_ratios, + 'variances': variance, + 'flip': flip, + 'clip': clip, + 'step_w': steps[0], + 'step_h': steps[1], + 'offset': offset, + 'min_max_aspect_ratios_order': min_max_aspect_ratios_order + } + + if cur_max_sizes is not None: + attrs['max_sizes'] = cur_max_sizes + + box = helper.create_variable_for_type_inference(dtype) + var = helper.create_variable_for_type_inference(dtype) + helper.append_op( + type="prior_box", + inputs={"Input": input, + "Image": image}, + outputs={"Boxes": box, + "Variances": var}, + attrs=attrs, ) + box.stop_gradient = True + var.stop_gradient = True + return box, var + + +@paddle.jit.not_to_static +def multiclass_nms(bboxes, + scores, + score_threshold, + nms_top_k, + keep_top_k, + nms_threshold=0.3, + normalized=True, + nms_eta=1., + background_label=-1, + return_index=False, + return_rois_num=True, + rois_num=None, + name=None): + """ + This operator is to do multi-class non maximum suppression (NMS) on + boxes and scores. + In the NMS step, this operator greedily selects a subset of detection bounding + boxes that have high scores larger than score_threshold, if providing this + threshold, then selects the largest nms_top_k confidences scores if nms_top_k + is larger than -1. Then this operator pruns away boxes that have high IOU + (intersection over union) overlap with already selected boxes by adaptive + threshold NMS based on parameters of nms_threshold and nms_eta. + Aftern NMS step, at most keep_top_k number of total bboxes are to be kept + per image if keep_top_k is larger than -1. + Args: + bboxes (Tensor): Two types of bboxes are supported: + 1. (Tensor) A 3-D Tensor with shape + [N, M, 4 or 8 16 24 32] represents the + predicted locations of M bounding bboxes, + N is the batch size. Each bounding box has four + coordinate values and the layout is + [xmin, ymin, xmax, ymax], when box size equals to 4. + 2. (LoDTensor) A 3-D Tensor with shape [M, C, 4] + M is the number of bounding boxes, C is the + class number + scores (Tensor): Two types of scores are supported: + 1. (Tensor) A 3-D Tensor with shape [N, C, M] + represents the predicted confidence predictions. + N is the batch size, C is the class number, M is + number of bounding boxes. For each category there + are total M scores which corresponding M bounding + boxes. Please note, M is equal to the 2nd dimension + of BBoxes. + 2. (LoDTensor) A 2-D LoDTensor with shape [M, C]. + M is the number of bbox, C is the class number. + In this case, input BBoxes should be the second + case with shape [M, C, 4]. + background_label (int): The index of background label, the background + label will be ignored. If set to -1, then all + categories will be considered. Default: 0 + score_threshold (float): Threshold to filter out bounding boxes with + low confidence score. If not provided, + consider all boxes. + nms_top_k (int): Maximum number of detections to be kept according to + the confidences after the filtering detections based + on score_threshold. + nms_threshold (float): The threshold to be used in NMS. Default: 0.3 + nms_eta (float): The threshold to be used in NMS. Default: 1.0 + keep_top_k (int): Number of total bboxes to be kept per image after NMS + step. -1 means keeping all bboxes after NMS step. + normalized (bool): Whether detections are normalized. Default: True + return_index(bool): Whether return selected index. Default: False + rois_num(Tensor): 1-D Tensor contains the number of RoIs in each image. + The shape is [B] and data type is int32. B is the number of images. + If it is not None then return a list of 1-D Tensor. Each element + is the output RoIs' number of each image on the corresponding level + and the shape is [B]. None by default. + name(str): Name of the multiclass nms op. Default: None. + Returns: + A tuple with two Variables: (Out, Index) if return_index is True, + otherwise, a tuple with one Variable(Out) is returned. + Out: A 2-D LoDTensor with shape [No, 6] represents the detections. + Each row has 6 values: [label, confidence, xmin, ymin, xmax, ymax] + or A 2-D LoDTensor with shape [No, 10] represents the detections. + Each row has 10 values: [label, confidence, x1, y1, x2, y2, x3, y3, + x4, y4]. No is the total number of detections. + If all images have not detected results, all elements in LoD will be + 0, and output tensor is empty (None). + Index: Only return when return_index is True. A 2-D LoDTensor with + shape [No, 1] represents the selected index which type is Integer. + The index is the absolute value cross batches. No is the same number + as Out. If the index is used to gather other attribute such as age, + one needs to reshape the input(N, M, 1) to (N * M, 1) as first, where + N is the batch size and M is the number of boxes. + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + boxes = paddle.static.data(name='bboxes', shape=[81, 4], + dtype='float32', lod_level=1) + scores = paddle.static.data(name='scores', shape=[81], + dtype='float32', lod_level=1) + out, index = ops.multiclass_nms(bboxes=boxes, + scores=scores, + background_label=0, + score_threshold=0.5, + nms_top_k=400, + nms_threshold=0.3, + keep_top_k=200, + normalized=False, + return_index=True) + """ + helper = LayerHelper('multiclass_nms3', **locals()) + + if in_dygraph_mode(): + attrs = ('background_label', background_label, 'score_threshold', + score_threshold, 'nms_top_k', nms_top_k, 'nms_threshold', + nms_threshold, 'keep_top_k', keep_top_k, 'nms_eta', nms_eta, + 'normalized', normalized) + output, index, nms_rois_num = core.ops.multiclass_nms3(bboxes, scores, + rois_num, *attrs) + if not return_index: + index = None + return output, nms_rois_num, index + + else: + output = helper.create_variable_for_type_inference(dtype=bboxes.dtype) + index = helper.create_variable_for_type_inference(dtype='int32') + + inputs = {'BBoxes': bboxes, 'Scores': scores} + outputs = {'Out': output, 'Index': index} + + if rois_num is not None: + inputs['RoisNum'] = rois_num + + if return_rois_num: + nms_rois_num = helper.create_variable_for_type_inference( + dtype='int32') + outputs['NmsRoisNum'] = nms_rois_num + + helper.append_op( + type="multiclass_nms3", + inputs=inputs, + attrs={ + 'background_label': background_label, + 'score_threshold': score_threshold, + 'nms_top_k': nms_top_k, + 'nms_threshold': nms_threshold, + 'keep_top_k': keep_top_k, + 'nms_eta': nms_eta, + 'normalized': normalized + }, + outputs=outputs) + output.stop_gradient = True + index.stop_gradient = True + if not return_index: + index = None + if not return_rois_num: + nms_rois_num = None + + return output, nms_rois_num, index + + +@paddle.jit.not_to_static +def matrix_nms(bboxes, + scores, + score_threshold, + post_threshold, + nms_top_k, + keep_top_k, + use_gaussian=False, + gaussian_sigma=2., + background_label=0, + normalized=True, + return_index=False, + return_rois_num=True, + name=None): + """ + **Matrix NMS** + This operator does matrix non maximum suppression (NMS). + First selects a subset of candidate bounding boxes that have higher scores + than score_threshold (if provided), then the top k candidate is selected if + nms_top_k is larger than -1. Score of the remaining candidate are then + decayed according to the Matrix NMS scheme. + Aftern NMS step, at most keep_top_k number of total bboxes are to be kept + per image if keep_top_k is larger than -1. + Args: + bboxes (Tensor): A 3-D Tensor with shape [N, M, 4] represents the + predicted locations of M bounding bboxes, + N is the batch size. Each bounding box has four + coordinate values and the layout is + [xmin, ymin, xmax, ymax], when box size equals to 4. + The data type is float32 or float64. + scores (Tensor): A 3-D Tensor with shape [N, C, M] + represents the predicted confidence predictions. + N is the batch size, C is the class number, M is + number of bounding boxes. For each category there + are total M scores which corresponding M bounding + boxes. Please note, M is equal to the 2nd dimension + of BBoxes. The data type is float32 or float64. + score_threshold (float): Threshold to filter out bounding boxes with + low confidence score. + post_threshold (float): Threshold to filter out bounding boxes with + low confidence score AFTER decaying. + nms_top_k (int): Maximum number of detections to be kept according to + the confidences after the filtering detections based + on score_threshold. + keep_top_k (int): Number of total bboxes to be kept per image after NMS + step. -1 means keeping all bboxes after NMS step. + use_gaussian (bool): Use Gaussian as the decay function. Default: False + gaussian_sigma (float): Sigma for Gaussian decay function. Default: 2.0 + background_label (int): The index of background label, the background + label will be ignored. If set to -1, then all + categories will be considered. Default: 0 + normalized (bool): Whether detections are normalized. Default: True + return_index(bool): Whether return selected index. Default: False + return_rois_num(bool): whether return rois_num. Default: True + name(str): Name of the matrix nms op. Default: None. + Returns: + A tuple with three Tensor: (Out, Index, RoisNum) if return_index is True, + otherwise, a tuple with two Tensor (Out, RoisNum) is returned. + Out (Tensor): A 2-D Tensor with shape [No, 6] containing the + detection results. + Each row has 6 values: [label, confidence, xmin, ymin, xmax, ymax] + (After version 1.3, when no boxes detected, the lod is changed + from {0} to {1}) + Index (Tensor): A 2-D Tensor with shape [No, 1] containing the + selected indices, which are absolute values cross batches. + rois_num (Tensor): A 1-D Tensor with shape [N] containing + the number of detected boxes in each image. + Examples: + .. code-block:: python + import paddle + from ppdet.modeling import ops + boxes = paddle.static.data(name='bboxes', shape=[None,81, 4], + dtype='float32', lod_level=1) + scores = paddle.static.data(name='scores', shape=[None,81], + dtype='float32', lod_level=1) + out = ops.matrix_nms(bboxes=boxes, scores=scores, background_label=0, + score_threshold=0.5, post_threshold=0.1, + nms_top_k=400, keep_top_k=200, normalized=False) + """ + check_variable_and_dtype(bboxes, 'BBoxes', ['float32', 'float64'], + 'matrix_nms') + check_variable_and_dtype(scores, 'Scores', ['float32', 'float64'], + 'matrix_nms') + check_type(score_threshold, 'score_threshold', float, 'matrix_nms') + check_type(post_threshold, 'post_threshold', float, 'matrix_nms') + check_type(nms_top_k, 'nums_top_k', int, 'matrix_nms') + check_type(keep_top_k, 'keep_top_k', int, 'matrix_nms') + check_type(normalized, 'normalized', bool, 'matrix_nms') + check_type(use_gaussian, 'use_gaussian', bool, 'matrix_nms') + check_type(gaussian_sigma, 'gaussian_sigma', float, 'matrix_nms') + check_type(background_label, 'background_label', int, 'matrix_nms') + + if in_dygraph_mode(): + attrs = ('background_label', background_label, 'score_threshold', + score_threshold, 'post_threshold', post_threshold, 'nms_top_k', + nms_top_k, 'gaussian_sigma', gaussian_sigma, 'use_gaussian', + use_gaussian, 'keep_top_k', keep_top_k, 'normalized', + normalized) + out, index, rois_num = core.ops.matrix_nms(bboxes, scores, *attrs) + if not return_index: + index = None + if not return_rois_num: + rois_num = None + return out, rois_num, index + else: + helper = LayerHelper('matrix_nms', **locals()) + output = helper.create_variable_for_type_inference(dtype=bboxes.dtype) + index = helper.create_variable_for_type_inference(dtype='int32') + outputs = {'Out': output, 'Index': index} + if return_rois_num: + rois_num = helper.create_variable_for_type_inference(dtype='int32') + outputs['RoisNum'] = rois_num + + helper.append_op( + type="matrix_nms", + inputs={'BBoxes': bboxes, + 'Scores': scores}, + attrs={ + 'background_label': background_label, + 'score_threshold': score_threshold, + 'post_threshold': post_threshold, + 'nms_top_k': nms_top_k, + 'gaussian_sigma': gaussian_sigma, + 'use_gaussian': use_gaussian, + 'keep_top_k': keep_top_k, + 'normalized': normalized + }, + outputs=outputs) + output.stop_gradient = True + + if not return_index: + index = None + if not return_rois_num: + rois_num = None + return output, rois_num, index + + +def bipartite_match(dist_matrix, + match_type=None, + dist_threshold=None, + name=None): + """ + + This operator implements a greedy bipartite matching algorithm, which is + used to obtain the matching with the maximum distance based on the input + distance matrix. For input 2D matrix, the bipartite matching algorithm can + find the matched column for each row (matched means the largest distance), + also can find the matched row for each column. And this operator only + calculate matched indices from column to row. For each instance, + the number of matched indices is the column number of the input distance + matrix. **The OP only supports CPU**. + + There are two outputs, matched indices and distance. + A simple description, this algorithm matched the best (maximum distance) + row entity to the column entity and the matched indices are not duplicated + in each row of ColToRowMatchIndices. If the column entity is not matched + any row entity, set -1 in ColToRowMatchIndices. + + NOTE: the input DistMat can be LoDTensor (with LoD) or Tensor. + If LoDTensor with LoD, the height of ColToRowMatchIndices is batch size. + If Tensor, the height of ColToRowMatchIndices is 1. + + NOTE: This API is a very low level API. It is used by :code:`ssd_loss` + layer. Please consider to use :code:`ssd_loss` instead. + + Args: + dist_matrix(Tensor): This input is a 2-D LoDTensor with shape + [K, M]. The data type is float32 or float64. It is pair-wise + distance matrix between the entities represented by each row and + each column. For example, assumed one entity is A with shape [K], + another entity is B with shape [M]. The dist_matrix[i][j] is the + distance between A[i] and B[j]. The bigger the distance is, the + better matching the pairs are. NOTE: This tensor can contain LoD + information to represent a batch of inputs. One instance of this + batch can contain different numbers of entities. + match_type(str, optional): The type of matching method, should be + 'bipartite' or 'per_prediction'. None ('bipartite') by default. + dist_threshold(float32, optional): If `match_type` is 'per_prediction', + this threshold is to determine the extra matching bboxes based + on the maximum distance, 0.5 by default. + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Tuple: + + matched_indices(Tensor): A 2-D Tensor with shape [N, M]. The data + type is int32. N is the batch size. If match_indices[i][j] is -1, it + means B[j] does not match any entity in i-th instance. + Otherwise, it means B[j] is matched to row + match_indices[i][j] in i-th instance. The row number of + i-th instance is saved in match_indices[i][j]. + + matched_distance(Tensor): A 2-D Tensor with shape [N, M]. The data + type is float32. N is batch size. If match_indices[i][j] is -1, + match_distance[i][j] is also -1.0. Otherwise, assumed + match_distance[i][j] = d, and the row offsets of each instance + are called LoD. Then match_distance[i][j] = + dist_matrix[d+LoD[i]][j]. + + Examples: + + .. code-block:: python + import paddle + from ppdet.modeling import ops + from ppdet.modeling.utils import iou_similarity + + paddle.enable_static() + + x = paddle.static.data(name='x', shape=[None, 4], dtype='float32') + y = paddle.static.data(name='y', shape=[None, 4], dtype='float32') + iou = iou_similarity(x=x, y=y) + matched_indices, matched_dist = ops.bipartite_match(iou) + """ + check_variable_and_dtype(dist_matrix, 'dist_matrix', + ['float32', 'float64'], 'bipartite_match') + + if in_dygraph_mode(): + match_indices, match_distance = core.ops.bipartite_match( + dist_matrix, "match_type", match_type, "dist_threshold", + dist_threshold) + return match_indices, match_distance + + helper = LayerHelper('bipartite_match', **locals()) + match_indices = helper.create_variable_for_type_inference(dtype='int32') + match_distance = helper.create_variable_for_type_inference( + dtype=dist_matrix.dtype) + helper.append_op( + type='bipartite_match', + inputs={'DistMat': dist_matrix}, + attrs={ + 'match_type': match_type, + 'dist_threshold': dist_threshold, + }, + outputs={ + 'ColToRowMatchIndices': match_indices, + 'ColToRowMatchDist': match_distance + }) + return match_indices, match_distance + + +@paddle.jit.not_to_static +def box_coder(prior_box, + prior_box_var, + target_box, + code_type="encode_center_size", + box_normalized=True, + axis=0, + name=None): + """ + **Box Coder Layer** + Encode/Decode the target bounding box with the priorbox information. + + The Encoding schema described below: + .. math:: + ox = (tx - px) / pw / pxv + oy = (ty - py) / ph / pyv + ow = \log(\abs(tw / pw)) / pwv + oh = \log(\abs(th / ph)) / phv + The Decoding schema described below: + + .. math:: + + ox = (pw * pxv * tx * + px) - tw / 2 + oy = (ph * pyv * ty * + py) - th / 2 + ow = \exp(pwv * tw) * pw + tw / 2 + oh = \exp(phv * th) * ph + th / 2 + where `tx`, `ty`, `tw`, `th` denote the target box's center coordinates, + width and height respectively. Similarly, `px`, `py`, `pw`, `ph` denote + the priorbox's (anchor) center coordinates, width and height. `pxv`, + `pyv`, `pwv`, `phv` denote the variance of the priorbox and `ox`, `oy`, + `ow`, `oh` denote the encoded/decoded coordinates, width and height. + During Box Decoding, two modes for broadcast are supported. Say target + box has shape [N, M, 4], and the shape of prior box can be [N, 4] or + [M, 4]. Then prior box will broadcast to target box along the + assigned axis. + + Args: + prior_box(Tensor): Box list prior_box is a 2-D Tensor with shape + [M, 4] holds M boxes and data type is float32 or float64. Each box + is represented as [xmin, ymin, xmax, ymax], [xmin, ymin] is the + left top coordinate of the anchor box, if the input is image feature + map, they are close to the origin of the coordinate system. + [xmax, ymax] is the right bottom coordinate of the anchor box. + prior_box_var(List|Tensor|None): prior_box_var supports three types + of input. One is Tensor with shape [M, 4] which holds M group and + data type is float32 or float64. The second is list consist of + 4 elements shared by all boxes and data type is float32 or float64. + Other is None and not involved in calculation. + target_box(Tensor): This input can be a 2-D LoDTensor with shape + [N, 4] when code_type is 'encode_center_size'. This input also can + be a 3-D Tensor with shape [N, M, 4] when code_type is + 'decode_center_size'. Each box is represented as + [xmin, ymin, xmax, ymax]. The data type is float32 or float64. + code_type(str): The code type used with the target box. It can be + `encode_center_size` or `decode_center_size`. `encode_center_size` + by default. + box_normalized(bool): Whether treat the priorbox as a normalized box. + Set true by default. + axis(int): Which axis in PriorBox to broadcast for box decode, + for example, if axis is 0 and TargetBox has shape [N, M, 4] and + PriorBox has shape [M, 4], then PriorBox will broadcast to [N, M, 4] + for decoding. It is only valid when code type is + `decode_center_size`. Set 0 by default. + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + Tensor: + output_box(Tensor): When code_type is 'encode_center_size', the + output tensor of box_coder_op with shape [N, M, 4] representing the + result of N target boxes encoded with M Prior boxes and variances. + When code_type is 'decode_center_size', N represents the batch size + and M represents the number of decoded boxes. + + Examples: + + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + # For encode + prior_box_encode = paddle.static.data(name='prior_box_encode', + shape=[512, 4], + dtype='float32') + target_box_encode = paddle.static.data(name='target_box_encode', + shape=[81, 4], + dtype='float32') + output_encode = ops.box_coder(prior_box=prior_box_encode, + prior_box_var=[0.1,0.1,0.2,0.2], + target_box=target_box_encode, + code_type="encode_center_size") + # For decode + prior_box_decode = paddle.static.data(name='prior_box_decode', + shape=[512, 4], + dtype='float32') + target_box_decode = paddle.static.data(name='target_box_decode', + shape=[512, 81, 4], + dtype='float32') + output_decode = ops.box_coder(prior_box=prior_box_decode, + prior_box_var=[0.1,0.1,0.2,0.2], + target_box=target_box_decode, + code_type="decode_center_size", + box_normalized=False, + axis=1) + """ + check_variable_and_dtype(prior_box, 'prior_box', ['float32', 'float64'], + 'box_coder') + check_variable_and_dtype(target_box, 'target_box', ['float32', 'float64'], + 'box_coder') + + if in_dygraph_mode(): + if isinstance(prior_box_var, Variable): + output_box = core.ops.box_coder( + prior_box, prior_box_var, target_box, "code_type", code_type, + "box_normalized", box_normalized, "axis", axis) + + elif isinstance(prior_box_var, list): + output_box = core.ops.box_coder( + prior_box, None, target_box, "code_type", code_type, + "box_normalized", box_normalized, "axis", axis, "variance", + prior_box_var) + else: + raise TypeError( + "Input variance of box_coder must be Variable or list") + return output_box + else: + helper = LayerHelper("box_coder", **locals()) + + output_box = helper.create_variable_for_type_inference( + dtype=prior_box.dtype) + + inputs = {"PriorBox": prior_box, "TargetBox": target_box} + attrs = { + "code_type": code_type, + "box_normalized": box_normalized, + "axis": axis + } + if isinstance(prior_box_var, Variable): + inputs['PriorBoxVar'] = prior_box_var + elif isinstance(prior_box_var, list): + attrs['variance'] = prior_box_var + else: + raise TypeError( + "Input variance of box_coder must be Variable or list") + helper.append_op( + type="box_coder", + inputs=inputs, + attrs=attrs, + outputs={"OutputBox": output_box}) + return output_box + + +@paddle.jit.not_to_static +def generate_proposals(scores, + bbox_deltas, + im_shape, + anchors, + variances, + pre_nms_top_n=6000, + post_nms_top_n=1000, + nms_thresh=0.5, + min_size=0.1, + eta=1.0, + pixel_offset=False, + return_rois_num=False, + name=None): + """ + **Generate proposal Faster-RCNN** + This operation proposes RoIs according to each box with their + probability to be a foreground object and + the box can be calculated by anchors. Bbox_deltais and scores + to be an object are the output of RPN. Final proposals + could be used to train detection net. + For generating proposals, this operation performs following steps: + 1. Transposes and resizes scores and bbox_deltas in size of + (H*W*A, 1) and (H*W*A, 4) + 2. Calculate box locations as proposals candidates. + 3. Clip boxes to image + 4. Remove predicted boxes with small area. + 5. Apply NMS to get final proposals as output. + Args: + scores(Tensor): A 4-D Tensor with shape [N, A, H, W] represents + the probability for each box to be an object. + N is batch size, A is number of anchors, H and W are height and + width of the feature map. The data type must be float32. + bbox_deltas(Tensor): A 4-D Tensor with shape [N, 4*A, H, W] + represents the difference between predicted box location and + anchor location. The data type must be float32. + im_shape(Tensor): A 2-D Tensor with shape [N, 2] represents H, W, the + origin image size or input size. The data type can be float32 or + float64. + anchors(Tensor): A 4-D Tensor represents the anchors with a layout + of [H, W, A, 4]. H and W are height and width of the feature map, + num_anchors is the box count of each position. Each anchor is + in (xmin, ymin, xmax, ymax) format an unnormalized. The data type must be float32. + variances(Tensor): A 4-D Tensor. The expanded variances of anchors with a layout of + [H, W, num_priors, 4]. Each variance is in + (xcenter, ycenter, w, h) format. The data type must be float32. + pre_nms_top_n(float): Number of total bboxes to be kept per + image before NMS. The data type must be float32. `6000` by default. + post_nms_top_n(float): Number of total bboxes to be kept per + image after NMS. The data type must be float32. `1000` by default. + nms_thresh(float): Threshold in NMS. The data type must be float32. `0.5` by default. + min_size(float): Remove predicted boxes with either height or + width < min_size. The data type must be float32. `0.1` by default. + eta(float): Apply in adaptive NMS, if adaptive `threshold > 0.5`, + `adaptive_threshold = adaptive_threshold * eta` in each iteration. + return_rois_num(bool): When setting True, it will return a 1D Tensor with shape [N, ] that includes Rois's + num of each image in one batch. The N is the image's num. For example, the tensor has values [4,5] that represents + the first image has 4 Rois, the second image has 5 Rois. It only used in rcnn model. + 'False' by default. + name(str, optional): For detailed information, please refer + to :ref:`api_guide_Name`. Usually name is no need to set and + None by default. + + Returns: + tuple: + A tuple with format ``(rpn_rois, rpn_roi_probs)``. + - **rpn_rois**: The generated RoIs. 2-D Tensor with shape ``[N, 4]`` while ``N`` is the number of RoIs. The data type is the same as ``scores``. + - **rpn_roi_probs**: The scores of generated RoIs. 2-D Tensor with shape ``[N, 1]`` while ``N`` is the number of RoIs. The data type is the same as ``scores``. + + Examples: + .. code-block:: python + + import paddle + from ppdet.modeling import ops + paddle.enable_static() + scores = paddle.static.data(name='scores', shape=[None, 4, 5, 5], dtype='float32') + bbox_deltas = paddle.static.data(name='bbox_deltas', shape=[None, 16, 5, 5], dtype='float32') + im_shape = paddle.static.data(name='im_shape', shape=[None, 2], dtype='float32') + anchors = paddle.static.data(name='anchors', shape=[None, 5, 4, 4], dtype='float32') + variances = paddle.static.data(name='variances', shape=[None, 5, 10, 4], dtype='float32') + rois, roi_probs = ops.generate_proposals(scores, bbox_deltas, + im_shape, anchors, variances) + """ + if in_dygraph_mode(): + assert return_rois_num, "return_rois_num should be True in dygraph mode." + attrs = ('pre_nms_topN', pre_nms_top_n, 'post_nms_topN', post_nms_top_n, + 'nms_thresh', nms_thresh, 'min_size', min_size, 'eta', eta, + 'pixel_offset', pixel_offset) + rpn_rois, rpn_roi_probs, rpn_rois_num = core.ops.generate_proposals_v2( + scores, bbox_deltas, im_shape, anchors, variances, *attrs) + return rpn_rois, rpn_roi_probs, rpn_rois_num + + else: + helper = LayerHelper('generate_proposals_v2', **locals()) + + check_variable_and_dtype(scores, 'scores', ['float32'], + 'generate_proposals_v2') + check_variable_and_dtype(bbox_deltas, 'bbox_deltas', ['float32'], + 'generate_proposals_v2') + check_variable_and_dtype(im_shape, 'im_shape', ['float32', 'float64'], + 'generate_proposals_v2') + check_variable_and_dtype(anchors, 'anchors', ['float32'], + 'generate_proposals_v2') + check_variable_and_dtype(variances, 'variances', ['float32'], + 'generate_proposals_v2') + + rpn_rois = helper.create_variable_for_type_inference( + dtype=bbox_deltas.dtype) + rpn_roi_probs = helper.create_variable_for_type_inference( + dtype=scores.dtype) + outputs = { + 'RpnRois': rpn_rois, + 'RpnRoiProbs': rpn_roi_probs, + } + if return_rois_num: + rpn_rois_num = helper.create_variable_for_type_inference( + dtype='int32') + rpn_rois_num.stop_gradient = True + outputs['RpnRoisNum'] = rpn_rois_num + + helper.append_op( + type="generate_proposals_v2", + inputs={ + 'Scores': scores, + 'BboxDeltas': bbox_deltas, + 'ImShape': im_shape, + 'Anchors': anchors, + 'Variances': variances + }, + attrs={ + 'pre_nms_topN': pre_nms_top_n, + 'post_nms_topN': post_nms_top_n, + 'nms_thresh': nms_thresh, + 'min_size': min_size, + 'eta': eta, + 'pixel_offset': pixel_offset + }, + outputs=outputs) + rpn_rois.stop_gradient = True + rpn_roi_probs.stop_gradient = True + + return rpn_rois, rpn_roi_probs, rpn_rois_num + + +def sigmoid_cross_entropy_with_logits(input, + label, + ignore_index=-100, + normalize=False): + output = F.binary_cross_entropy_with_logits(input, label, reduction='none') + mask_tensor = paddle.cast(label != ignore_index, 'float32') + output = paddle.multiply(output, mask_tensor) + if normalize: + sum_valid_mask = paddle.sum(mask_tensor) + output = output / sum_valid_mask + return output + + +def smooth_l1(input, label, inside_weight=None, outside_weight=None, + sigma=None): + input_new = paddle.multiply(input, inside_weight) + label_new = paddle.multiply(label, inside_weight) + delta = 1 / (sigma * sigma) + out = F.smooth_l1_loss(input_new, label_new, reduction='none', delta=delta) + out = paddle.multiply(out, outside_weight) + out = out / delta + out = paddle.reshape(out, shape=[out.shape[0], -1]) + out = paddle.sum(out, axis=1) + return out + + +def channel_shuffle(x, groups): + batch_size, num_channels, height, width = x.shape[0:4] + assert num_channels % groups == 0, 'num_channels should be divisible by groups' + channels_per_group = num_channels // groups + x = paddle.reshape( + x=x, shape=[batch_size, groups, channels_per_group, height, width]) + x = paddle.transpose(x=x, perm=[0, 2, 1, 3, 4]) + x = paddle.reshape(x=x, shape=[batch_size, num_channels, height, width]) + return x diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/post_process.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/post_process.py new file mode 100644 index 000000000..679e09134 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/post_process.py @@ -0,0 +1,656 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from ppdet.core.workspace import register +from ppdet.modeling.bbox_utils import nonempty_bbox, rbox2poly +from ppdet.modeling.layers import TTFBox +from .transformers import bbox_cxcywh_to_xyxy +try: + from collections.abc import Sequence +except Exception: + from collections import Sequence + +__all__ = [ + 'BBoxPostProcess', 'MaskPostProcess', 'FCOSPostProcess', + 'S2ANetBBoxPostProcess', 'JDEBBoxPostProcess', 'CenterNetPostProcess', + 'DETRBBoxPostProcess', 'SparsePostProcess' +] + + +@register +class BBoxPostProcess(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['decode', 'nms'] + + def __init__(self, num_classes=80, decode=None, nms=None): + super(BBoxPostProcess, self).__init__() + self.num_classes = num_classes + self.decode = decode + self.nms = nms + self.fake_bboxes = paddle.to_tensor( + np.array( + [[-1, 0.0, 0.0, 0.0, 0.0, 0.0]], dtype='float32')) + self.fake_bbox_num = paddle.to_tensor(np.array([1], dtype='int32')) + + def forward(self, head_out, rois, im_shape, scale_factor): + """ + Decode the bbox and do NMS if needed. + + Args: + head_out (tuple): bbox_pred and cls_prob of bbox_head output. + rois (tuple): roi and rois_num of rpn_head output. + im_shape (Tensor): The shape of the input image. + scale_factor (Tensor): The scale factor of the input image. + Returns: + bbox_pred (Tensor): The output prediction with shape [N, 6], including + labels, scores and bboxes. The size of bboxes are corresponding + to the input image, the bboxes may be used in other branch. + bbox_num (Tensor): The number of prediction boxes of each batch with + shape [1], and is N. + """ + if self.nms is not None: + bboxes, score = self.decode(head_out, rois, im_shape, scale_factor) + bbox_pred, bbox_num, _ = self.nms(bboxes, score, self.num_classes) + else: + bbox_pred, bbox_num = self.decode(head_out, rois, im_shape, + scale_factor) + return bbox_pred, bbox_num + + def get_pred(self, bboxes, bbox_num, im_shape, scale_factor): + """ + Rescale, clip and filter the bbox from the output of NMS to + get final prediction. + + Notes: + Currently only support bs = 1. + + Args: + bboxes (Tensor): The output bboxes with shape [N, 6] after decode + and NMS, including labels, scores and bboxes. + bbox_num (Tensor): The number of prediction boxes of each batch with + shape [1], and is N. + im_shape (Tensor): The shape of the input image. + scale_factor (Tensor): The scale factor of the input image. + Returns: + pred_result (Tensor): The final prediction results with shape [N, 6] + including labels, scores and bboxes. + """ + + if bboxes.shape[0] == 0: + bboxes = self.fake_bboxes + bbox_num = self.fake_bbox_num + + origin_shape = paddle.floor(im_shape / scale_factor + 0.5) + + origin_shape_list = [] + scale_factor_list = [] + # scale_factor: scale_y, scale_x + for i in range(bbox_num.shape[0]): + expand_shape = paddle.expand(origin_shape[i:i + 1, :], + [bbox_num[i], 2]) + scale_y, scale_x = scale_factor[i][0], scale_factor[i][1] + scale = paddle.concat([scale_x, scale_y, scale_x, scale_y]) + expand_scale = paddle.expand(scale, [bbox_num[i], 4]) + origin_shape_list.append(expand_shape) + scale_factor_list.append(expand_scale) + + self.origin_shape_list = paddle.concat(origin_shape_list) + scale_factor_list = paddle.concat(scale_factor_list) + + # bboxes: [N, 6], label, score, bbox + pred_label = bboxes[:, 0:1] + pred_score = bboxes[:, 1:2] + pred_bbox = bboxes[:, 2:] + # rescale bbox to original image + scaled_bbox = pred_bbox / scale_factor_list + origin_h = self.origin_shape_list[:, 0] + origin_w = self.origin_shape_list[:, 1] + zeros = paddle.zeros_like(origin_h) + # clip bbox to [0, original_size] + x1 = paddle.maximum(paddle.minimum(scaled_bbox[:, 0], origin_w), zeros) + y1 = paddle.maximum(paddle.minimum(scaled_bbox[:, 1], origin_h), zeros) + x2 = paddle.maximum(paddle.minimum(scaled_bbox[:, 2], origin_w), zeros) + y2 = paddle.maximum(paddle.minimum(scaled_bbox[:, 3], origin_h), zeros) + pred_bbox = paddle.stack([x1, y1, x2, y2], axis=-1) + # filter empty bbox + keep_mask = nonempty_bbox(pred_bbox, return_mask=True) + keep_mask = paddle.unsqueeze(keep_mask, [1]) + pred_label = paddle.where(keep_mask, pred_label, + paddle.ones_like(pred_label) * -1) + pred_result = paddle.concat([pred_label, pred_score, pred_bbox], axis=1) + return pred_result + + def get_origin_shape(self, ): + return self.origin_shape_list + + +@register +class MaskPostProcess(object): + """ + refer to: + https://github.com/facebookresearch/detectron2/layers/mask_ops.py + + Get Mask output according to the output from model + """ + + def __init__(self, binary_thresh=0.5): + super(MaskPostProcess, self).__init__() + self.binary_thresh = binary_thresh + + def paste_mask(self, masks, boxes, im_h, im_w): + """ + Paste the mask prediction to the original image. + """ + x0, y0, x1, y1 = paddle.split(boxes, 4, axis=1) + masks = paddle.unsqueeze(masks, [0, 1]) + img_y = paddle.arange(0, im_h, dtype='float32') + 0.5 + img_x = paddle.arange(0, im_w, dtype='float32') + 0.5 + img_y = (img_y - y0) / (y1 - y0) * 2 - 1 + img_x = (img_x - x0) / (x1 - x0) * 2 - 1 + img_x = paddle.unsqueeze(img_x, [1]) + img_y = paddle.unsqueeze(img_y, [2]) + N = boxes.shape[0] + + gx = paddle.expand(img_x, [N, img_y.shape[1], img_x.shape[2]]) + gy = paddle.expand(img_y, [N, img_y.shape[1], img_x.shape[2]]) + grid = paddle.stack([gx, gy], axis=3) + img_masks = F.grid_sample(masks, grid, align_corners=False) + return img_masks[:, 0] + + def __call__(self, mask_out, bboxes, bbox_num, origin_shape): + """ + Decode the mask_out and paste the mask to the origin image. + + Args: + mask_out (Tensor): mask_head output with shape [N, 28, 28]. + bbox_pred (Tensor): The output bboxes with shape [N, 6] after decode + and NMS, including labels, scores and bboxes. + bbox_num (Tensor): The number of prediction boxes of each batch with + shape [1], and is N. + origin_shape (Tensor): The origin shape of the input image, the tensor + shape is [N, 2], and each row is [h, w]. + Returns: + pred_result (Tensor): The final prediction mask results with shape + [N, h, w] in binary mask style. + """ + num_mask = mask_out.shape[0] + origin_shape = paddle.cast(origin_shape, 'int32') + # TODO: support bs > 1 and mask output dtype is bool + pred_result = paddle.zeros( + [num_mask, origin_shape[0][0], origin_shape[0][1]], dtype='int32') + if bbox_num == 1 and bboxes[0][0] == -1: + return pred_result + + # TODO: optimize chunk paste + pred_result = [] + for i in range(bboxes.shape[0]): + im_h, im_w = origin_shape[i][0], origin_shape[i][1] + pred_mask = self.paste_mask(mask_out[i], bboxes[i:i + 1, 2:], im_h, + im_w) + pred_mask = pred_mask >= self.binary_thresh + pred_mask = paddle.cast(pred_mask, 'int32') + pred_result.append(pred_mask) + pred_result = paddle.concat(pred_result) + return pred_result + + +@register +class FCOSPostProcess(object): + __inject__ = ['decode', 'nms'] + + def __init__(self, decode=None, nms=None): + super(FCOSPostProcess, self).__init__() + self.decode = decode + self.nms = nms + + def __call__(self, fcos_head_outs, scale_factor): + """ + Decode the bbox and do NMS in FCOS. + """ + locations, cls_logits, bboxes_reg, centerness = fcos_head_outs + bboxes, score = self.decode(locations, cls_logits, bboxes_reg, + centerness, scale_factor) + bbox_pred, bbox_num, _ = self.nms(bboxes, score) + return bbox_pred, bbox_num + + +@register +class S2ANetBBoxPostProcess(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['nms'] + + def __init__(self, num_classes=15, nms_pre=2000, min_bbox_size=0, nms=None): + super(S2ANetBBoxPostProcess, self).__init__() + self.num_classes = num_classes + self.nms_pre = paddle.to_tensor(nms_pre) + self.min_bbox_size = min_bbox_size + self.nms = nms + self.origin_shape_list = [] + self.fake_pred_cls_score_bbox = paddle.to_tensor( + np.array( + [[-1, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]], + dtype='float32')) + self.fake_bbox_num = paddle.to_tensor(np.array([1], dtype='int32')) + + def forward(self, pred_scores, pred_bboxes): + """ + pred_scores : [N, M] score + pred_bboxes : [N, 5] xc, yc, w, h, a + im_shape : [N, 2] im_shape + scale_factor : [N, 2] scale_factor + """ + pred_ploys0 = rbox2poly(pred_bboxes) + pred_ploys = paddle.unsqueeze(pred_ploys0, axis=0) + + # pred_scores [NA, 16] --> [16, NA] + pred_scores0 = paddle.transpose(pred_scores, [1, 0]) + pred_scores = paddle.unsqueeze(pred_scores0, axis=0) + + pred_cls_score_bbox, bbox_num, _ = self.nms(pred_ploys, pred_scores, + self.num_classes) + # Prevent empty bbox_pred from decode or NMS. + # Bboxes and score before NMS may be empty due to the score threshold. + if pred_cls_score_bbox.shape[0] <= 0 or pred_cls_score_bbox.shape[ + 1] <= 1: + pred_cls_score_bbox = self.fake_pred_cls_score_bbox + bbox_num = self.fake_bbox_num + + pred_cls_score_bbox = paddle.reshape(pred_cls_score_bbox, [-1, 10]) + return pred_cls_score_bbox, bbox_num + + def get_pred(self, bboxes, bbox_num, im_shape, scale_factor): + """ + Rescale, clip and filter the bbox from the output of NMS to + get final prediction. + Args: + bboxes(Tensor): bboxes [N, 10] + bbox_num(Tensor): bbox_num + im_shape(Tensor): [1 2] + scale_factor(Tensor): [1 2] + Returns: + bbox_pred(Tensor): The output is the prediction with shape [N, 8] + including labels, scores and bboxes. The size of + bboxes are corresponding to the original image. + """ + origin_shape = paddle.floor(im_shape / scale_factor + 0.5) + + origin_shape_list = [] + scale_factor_list = [] + # scale_factor: scale_y, scale_x + for i in range(bbox_num.shape[0]): + expand_shape = paddle.expand(origin_shape[i:i + 1, :], + [bbox_num[i], 2]) + scale_y, scale_x = scale_factor[i][0], scale_factor[i][1] + scale = paddle.concat([ + scale_x, scale_y, scale_x, scale_y, scale_x, scale_y, scale_x, + scale_y + ]) + expand_scale = paddle.expand(scale, [bbox_num[i], 8]) + origin_shape_list.append(expand_shape) + scale_factor_list.append(expand_scale) + + origin_shape_list = paddle.concat(origin_shape_list) + scale_factor_list = paddle.concat(scale_factor_list) + + # bboxes: [N, 10], label, score, bbox + pred_label_score = bboxes[:, 0:2] + pred_bbox = bboxes[:, 2:] + + # rescale bbox to original image + pred_bbox = pred_bbox.reshape([-1, 8]) + scaled_bbox = pred_bbox / scale_factor_list + origin_h = origin_shape_list[:, 0] + origin_w = origin_shape_list[:, 1] + + bboxes = scaled_bbox + zeros = paddle.zeros_like(origin_h) + x1 = paddle.maximum(paddle.minimum(bboxes[:, 0], origin_w - 1), zeros) + y1 = paddle.maximum(paddle.minimum(bboxes[:, 1], origin_h - 1), zeros) + x2 = paddle.maximum(paddle.minimum(bboxes[:, 2], origin_w - 1), zeros) + y2 = paddle.maximum(paddle.minimum(bboxes[:, 3], origin_h - 1), zeros) + x3 = paddle.maximum(paddle.minimum(bboxes[:, 4], origin_w - 1), zeros) + y3 = paddle.maximum(paddle.minimum(bboxes[:, 5], origin_h - 1), zeros) + x4 = paddle.maximum(paddle.minimum(bboxes[:, 6], origin_w - 1), zeros) + y4 = paddle.maximum(paddle.minimum(bboxes[:, 7], origin_h - 1), zeros) + pred_bbox = paddle.stack([x1, y1, x2, y2, x3, y3, x4, y4], axis=-1) + pred_result = paddle.concat([pred_label_score, pred_bbox], axis=1) + return pred_result + + +@register +class JDEBBoxPostProcess(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['decode', 'nms'] + + def __init__(self, num_classes=1, decode=None, nms=None, return_idx=True): + super(JDEBBoxPostProcess, self).__init__() + self.num_classes = num_classes + self.decode = decode + self.nms = nms + self.return_idx = return_idx + + self.fake_bbox_pred = paddle.to_tensor( + np.array( + [[-1, 0.0, 0.0, 0.0, 0.0, 0.0]], dtype='float32')) + self.fake_bbox_num = paddle.to_tensor(np.array([1], dtype='int32')) + self.fake_nms_keep_idx = paddle.to_tensor( + np.array( + [[0]], dtype='int32')) + + self.fake_yolo_boxes_out = paddle.to_tensor( + np.array( + [[[0.0, 0.0, 0.0, 0.0]]], dtype='float32')) + self.fake_yolo_scores_out = paddle.to_tensor( + np.array( + [[[0.0]]], dtype='float32')) + self.fake_boxes_idx = paddle.to_tensor(np.array([[0]], dtype='int64')) + + def forward(self, head_out, anchors): + """ + Decode the bbox and do NMS for JDE model. + + Args: + head_out (list): Bbox_pred and cls_prob of bbox_head output. + anchors (list): Anchors of JDE model. + + Returns: + boxes_idx (Tensor): The index of kept bboxes after decode 'JDEBox'. + bbox_pred (Tensor): The output is the prediction with shape [N, 6] + including labels, scores and bboxes. + bbox_num (Tensor): The number of prediction of each batch with shape [N]. + nms_keep_idx (Tensor): The index of kept bboxes after NMS. + """ + boxes_idx, yolo_boxes_scores = self.decode(head_out, anchors) + + if len(boxes_idx) == 0: + boxes_idx = self.fake_boxes_idx + yolo_boxes_out = self.fake_yolo_boxes_out + yolo_scores_out = self.fake_yolo_scores_out + else: + yolo_boxes = paddle.gather_nd(yolo_boxes_scores, boxes_idx) + # TODO: only support bs=1 now + yolo_boxes_out = paddle.reshape( + yolo_boxes[:, :4], shape=[1, len(boxes_idx), 4]) + yolo_scores_out = paddle.reshape( + yolo_boxes[:, 4:5], shape=[1, 1, len(boxes_idx)]) + boxes_idx = boxes_idx[:, 1:] + + if self.return_idx: + bbox_pred, bbox_num, nms_keep_idx = self.nms( + yolo_boxes_out, yolo_scores_out, self.num_classes) + if bbox_pred.shape[0] == 0: + bbox_pred = self.fake_bbox_pred + bbox_num = self.fake_bbox_num + nms_keep_idx = self.fake_nms_keep_idx + return boxes_idx, bbox_pred, bbox_num, nms_keep_idx + else: + bbox_pred, bbox_num, _ = self.nms(yolo_boxes_out, yolo_scores_out, + self.num_classes) + if bbox_pred.shape[0] == 0: + bbox_pred = self.fake_bbox_pred + bbox_num = self.fake_bbox_num + return _, bbox_pred, bbox_num, _ + + +@register +class CenterNetPostProcess(TTFBox): + """ + Postprocess the model outputs to get final prediction: + 1. Do NMS for heatmap to get top `max_per_img` bboxes. + 2. Decode bboxes using center offset and box size. + 3. Rescale decoded bboxes reference to the origin image shape. + + Args: + max_per_img(int): the maximum number of predicted objects in a image, + 500 by default. + down_ratio(int): the down ratio from images to heatmap, 4 by default. + regress_ltrb (bool): whether to regress left/top/right/bottom or + width/height for a box, true by default. + for_mot (bool): whether return other features used in tracking model. + """ + + __shared__ = ['down_ratio', 'for_mot'] + + def __init__(self, + max_per_img=500, + down_ratio=4, + regress_ltrb=True, + for_mot=False): + super(TTFBox, self).__init__() + self.max_per_img = max_per_img + self.down_ratio = down_ratio + self.regress_ltrb = regress_ltrb + self.for_mot = for_mot + + def __call__(self, hm, wh, reg, im_shape, scale_factor): + heat = self._simple_nms(hm) + scores, inds, topk_clses, ys, xs = self._topk(heat) + scores = scores.unsqueeze(1) + clses = topk_clses.unsqueeze(1) + + reg_t = paddle.transpose(reg, [0, 2, 3, 1]) + # Like TTFBox, batch size is 1. + # TODO: support batch size > 1 + reg = paddle.reshape(reg_t, [-1, reg_t.shape[-1]]) + reg = paddle.gather(reg, inds) + xs = paddle.cast(xs, 'float32') + ys = paddle.cast(ys, 'float32') + xs = xs + reg[:, 0:1] + ys = ys + reg[:, 1:2] + + wh_t = paddle.transpose(wh, [0, 2, 3, 1]) + wh = paddle.reshape(wh_t, [-1, wh_t.shape[-1]]) + wh = paddle.gather(wh, inds) + + if self.regress_ltrb: + x1 = xs - wh[:, 0:1] + y1 = ys - wh[:, 1:2] + x2 = xs + wh[:, 2:3] + y2 = ys + wh[:, 3:4] + else: + x1 = xs - wh[:, 0:1] / 2 + y1 = ys - wh[:, 1:2] / 2 + x2 = xs + wh[:, 0:1] / 2 + y2 = ys + wh[:, 1:2] / 2 + + n, c, feat_h, feat_w = hm.shape[:] + padw = (feat_w * self.down_ratio - im_shape[0, 1]) / 2 + padh = (feat_h * self.down_ratio - im_shape[0, 0]) / 2 + x1 = x1 * self.down_ratio + y1 = y1 * self.down_ratio + x2 = x2 * self.down_ratio + y2 = y2 * self.down_ratio + + x1 = x1 - padw + y1 = y1 - padh + x2 = x2 - padw + y2 = y2 - padh + + bboxes = paddle.concat([x1, y1, x2, y2], axis=1) + scale_y = scale_factor[:, 0:1] + scale_x = scale_factor[:, 1:2] + scale_expand = paddle.concat( + [scale_x, scale_y, scale_x, scale_y], axis=1) + boxes_shape = bboxes.shape[:] + scale_expand = paddle.expand(scale_expand, shape=boxes_shape) + bboxes = paddle.divide(bboxes, scale_expand) + if self.for_mot: + results = paddle.concat([bboxes, scores, clses], axis=1) + return results, inds, topk_clses + else: + results = paddle.concat([clses, scores, bboxes], axis=1) + return results, paddle.shape(results)[0:1], topk_clses + + +@register +class DETRBBoxPostProcess(object): + __shared__ = ['num_classes', 'use_focal_loss'] + __inject__ = [] + + def __init__(self, + num_classes=80, + num_top_queries=100, + use_focal_loss=False): + super(DETRBBoxPostProcess, self).__init__() + self.num_classes = num_classes + self.num_top_queries = num_top_queries + self.use_focal_loss = use_focal_loss + + def __call__(self, head_out, im_shape, scale_factor): + """ + Decode the bbox. + + Args: + head_out (tuple): bbox_pred, cls_logit and masks of bbox_head output. + im_shape (Tensor): The shape of the input image. + scale_factor (Tensor): The scale factor of the input image. + Returns: + bbox_pred (Tensor): The output prediction with shape [N, 6], including + labels, scores and bboxes. The size of bboxes are corresponding + to the input image, the bboxes may be used in other branch. + bbox_num (Tensor): The number of prediction boxes of each batch with + shape [bs], and is N. + """ + bboxes, logits, masks = head_out + + bbox_pred = bbox_cxcywh_to_xyxy(bboxes) + origin_shape = paddle.floor(im_shape / scale_factor + 0.5) + img_h, img_w = origin_shape.unbind(1) + origin_shape = paddle.stack( + [img_w, img_h, img_w, img_h], axis=-1).unsqueeze(0) + bbox_pred *= origin_shape + + scores = F.sigmoid(logits) if self.use_focal_loss else F.softmax( + logits)[:, :, :-1] + + if not self.use_focal_loss: + scores, labels = scores.max(-1), scores.argmax(-1) + if scores.shape[1] > self.num_top_queries: + scores, index = paddle.topk( + scores, self.num_top_queries, axis=-1) + labels = paddle.stack( + [paddle.gather(l, i) for l, i in zip(labels, index)]) + bbox_pred = paddle.stack( + [paddle.gather(b, i) for b, i in zip(bbox_pred, index)]) + else: + scores, index = paddle.topk( + scores.reshape([logits.shape[0], -1]), + self.num_top_queries, + axis=-1) + labels = index % logits.shape[2] + index = index // logits.shape[2] + bbox_pred = paddle.stack( + [paddle.gather(b, i) for b, i in zip(bbox_pred, index)]) + + bbox_pred = paddle.concat( + [ + labels.unsqueeze(-1).astype('float32'), scores.unsqueeze(-1), + bbox_pred + ], + axis=-1) + bbox_num = paddle.to_tensor( + bbox_pred.shape[1], dtype='int32').tile([bbox_pred.shape[0]]) + bbox_pred = bbox_pred.reshape([-1, 6]) + return bbox_pred, bbox_num + + +@register +class SparsePostProcess(object): + __shared__ = ['num_classes'] + + def __init__(self, num_proposals, num_classes=80): + super(SparsePostProcess, self).__init__() + self.num_classes = num_classes + self.num_proposals = num_proposals + + def __call__(self, box_cls, box_pred, scale_factor_wh, img_whwh): + """ + Arguments: + box_cls (Tensor): tensor of shape (batch_size, num_proposals, K). + The tensor predicts the classification probability for each proposal. + box_pred (Tensor): tensors of shape (batch_size, num_proposals, 4). + The tensor predicts 4-vector (x,y,w,h) box + regression values for every proposal + scale_factor_wh (Tensor): tensors of shape [batch_size, 2] the scalor of per img + img_whwh (Tensor): tensors of shape [batch_size, 4] + Returns: + bbox_pred (Tensor): tensors of shape [num_boxes, 6] Each row has 6 values: + [label, confidence, xmin, ymin, xmax, ymax] + bbox_num (Tensor): tensors of shape [batch_size] the number of RoIs in each image. + """ + assert len(box_cls) == len(scale_factor_wh) == len(img_whwh) + + img_wh = img_whwh[:, :2] + + scores = F.sigmoid(box_cls) + labels = paddle.arange(0, self.num_classes). \ + unsqueeze(0).tile([self.num_proposals, 1]).flatten(start_axis=0, stop_axis=1) + + classes_all = [] + scores_all = [] + boxes_all = [] + for i, (scores_per_image, + box_pred_per_image) in enumerate(zip(scores, box_pred)): + + scores_per_image, topk_indices = scores_per_image.flatten( + 0, 1).topk( + self.num_proposals, sorted=False) + labels_per_image = paddle.gather(labels, topk_indices, axis=0) + + box_pred_per_image = box_pred_per_image.reshape([-1, 1, 4]).tile( + [1, self.num_classes, 1]).reshape([-1, 4]) + box_pred_per_image = paddle.gather( + box_pred_per_image, topk_indices, axis=0) + + classes_all.append(labels_per_image) + scores_all.append(scores_per_image) + boxes_all.append(box_pred_per_image) + + bbox_num = paddle.zeros([len(scale_factor_wh)], dtype="int32") + boxes_final = [] + + for i in range(len(scale_factor_wh)): + classes = classes_all[i] + boxes = boxes_all[i] + scores = scores_all[i] + + boxes[:, 0::2] = paddle.clip( + boxes[:, 0::2], min=0, max=img_wh[i][0]) / scale_factor_wh[i][0] + boxes[:, 1::2] = paddle.clip( + boxes[:, 1::2], min=0, max=img_wh[i][1]) / scale_factor_wh[i][1] + boxes_w, boxes_h = (boxes[:, 2] - boxes[:, 0]).numpy(), ( + boxes[:, 3] - boxes[:, 1]).numpy() + + keep = (boxes_w > 1.) & (boxes_h > 1.) + + if (keep.sum() == 0): + bboxes = paddle.zeros([1, 6]).astype("float32") + else: + boxes = paddle.to_tensor(boxes.numpy()[keep]).astype("float32") + classes = paddle.to_tensor(classes.numpy()[keep]).astype( + "float32").unsqueeze(-1) + scores = paddle.to_tensor(scores.numpy()[keep]).astype( + "float32").unsqueeze(-1) + + bboxes = paddle.concat([classes, scores, boxes], axis=-1) + + boxes_final.append(bboxes) + bbox_num[i] = bboxes.shape[0] + + bbox_pred = paddle.concat(boxes_final) + return bbox_pred, bbox_num diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__init__.py new file mode 100644 index 000000000..9fb518f2a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__init__.py @@ -0,0 +1,2 @@ +from . import rpn_head +from .rpn_head import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..3e5c82d31 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/anchor_generator.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/anchor_generator.cpython-37.pyc new file mode 100644 index 000000000..db9b88dd8 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/anchor_generator.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/proposal_generator.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/proposal_generator.cpython-37.pyc new file mode 100644 index 000000000..950055728 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/proposal_generator.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/rpn_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/rpn_head.cpython-37.pyc new file mode 100644 index 000000000..797bb0a85 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/rpn_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target.cpython-37.pyc new file mode 100644 index 000000000..8d5c58349 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target_layer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target_layer.cpython-37.pyc new file mode 100644 index 000000000..79ee00500 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/__pycache__/target_layer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/anchor_generator.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/anchor_generator.py new file mode 100644 index 000000000..34f03c0ef --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/anchor_generator.py @@ -0,0 +1,131 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on +# https://github.com/facebookresearch/detectron2/blob/main/detectron2/modeling/anchor_generator.py + +import math + +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register + + +@register +class AnchorGenerator(nn.Layer): + """ + Generate anchors according to the feature maps + + Args: + anchor_sizes (list[float] | list[list[float]]): The anchor sizes at + each feature point. list[float] means all feature levels share the + same sizes. list[list[float]] means the anchor sizes for + each level. The sizes stand for the scale of input size. + aspect_ratios (list[float] | list[list[float]]): The aspect ratios at + each feature point. list[float] means all feature levels share the + same ratios. list[list[float]] means the aspect ratios for + each level. + strides (list[float]): The strides of feature maps which generate + anchors + offset (float): The offset of the coordinate of anchors, default 0. + + """ + + def __init__(self, + anchor_sizes=[32, 64, 128, 256, 512], + aspect_ratios=[0.5, 1.0, 2.0], + strides=[16.0], + variance=[1.0, 1.0, 1.0, 1.0], + offset=0.): + super(AnchorGenerator, self).__init__() + self.anchor_sizes = anchor_sizes + self.aspect_ratios = aspect_ratios + self.strides = strides + self.variance = variance + self.cell_anchors = self._calculate_anchors(len(strides)) + self.offset = offset + + def _broadcast_params(self, params, num_features): + if not isinstance(params[0], (list, tuple)): # list[float] + return [params] * num_features + if len(params) == 1: + return list(params) * num_features + return params + + def generate_cell_anchors(self, sizes, aspect_ratios): + anchors = [] + for size in sizes: + area = size**2.0 + for aspect_ratio in aspect_ratios: + w = math.sqrt(area / aspect_ratio) + h = aspect_ratio * w + x0, y0, x1, y1 = -w / 2.0, -h / 2.0, w / 2.0, h / 2.0 + anchors.append([x0, y0, x1, y1]) + return paddle.to_tensor(anchors, dtype='float32') + + def _calculate_anchors(self, num_features): + sizes = self._broadcast_params(self.anchor_sizes, num_features) + aspect_ratios = self._broadcast_params(self.aspect_ratios, num_features) + cell_anchors = [ + self.generate_cell_anchors(s, a) + for s, a in zip(sizes, aspect_ratios) + ] + [ + self.register_buffer( + t.name, t, persistable=False) for t in cell_anchors + ] + return cell_anchors + + def _create_grid_offsets(self, size, stride, offset): + grid_height, grid_width = size[0], size[1] + shifts_x = paddle.arange( + offset * stride, grid_width * stride, step=stride, dtype='float32') + shifts_y = paddle.arange( + offset * stride, grid_height * stride, step=stride, dtype='float32') + shift_y, shift_x = paddle.meshgrid(shifts_y, shifts_x) + shift_x = paddle.reshape(shift_x, [-1]) + shift_y = paddle.reshape(shift_y, [-1]) + return shift_x, shift_y + + def _grid_anchors(self, grid_sizes): + anchors = [] + for size, stride, base_anchors in zip(grid_sizes, self.strides, + self.cell_anchors): + shift_x, shift_y = self._create_grid_offsets(size, stride, + self.offset) + shifts = paddle.stack((shift_x, shift_y, shift_x, shift_y), axis=1) + shifts = paddle.reshape(shifts, [-1, 1, 4]) + base_anchors = paddle.reshape(base_anchors, [1, -1, 4]) + + anchors.append(paddle.reshape(shifts + base_anchors, [-1, 4])) + + return anchors + + def forward(self, input): + grid_sizes = [paddle.shape(feature_map)[-2:] for feature_map in input] + anchors_over_all_feature_maps = self._grid_anchors(grid_sizes) + return anchors_over_all_feature_maps + + @property + def num_anchors(self): + """ + Returns: + int: number of anchors at every pixel + location, on that feature map. + For example, if at every pixel we use anchors of 3 aspect + ratios and 5 sizes, the number of anchors is 15. + For FPN models, `num_anchors` on every feature map is the same. + """ + return len(self.cell_anchors[0]) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/proposal_generator.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/proposal_generator.py new file mode 100644 index 000000000..1fcb8b1e2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/proposal_generator.py @@ -0,0 +1,77 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle + +from ppdet.core.workspace import register, serializable +from .. import ops + + +@register +@serializable +class ProposalGenerator(object): + """ + Proposal generation module + + For more details, please refer to the document of generate_proposals + in ppdet/modeing/ops.py + + Args: + pre_nms_top_n (int): Number of total bboxes to be kept per + image before NMS. default 6000 + post_nms_top_n (int): Number of total bboxes to be kept per + image after NMS. default 1000 + nms_thresh (float): Threshold in NMS. default 0.5 + min_size (flaot): Remove predicted boxes with either height or + width < min_size. default 0.1 + eta (float): Apply in adaptive NMS, if adaptive `threshold > 0.5`, + `adaptive_threshold = adaptive_threshold * eta` in each iteration. + default 1. + topk_after_collect (bool): whether to adopt topk after batch + collection. If topk_after_collect is true, box filter will not be + used after NMS at each image in proposal generation. default false + """ + + def __init__(self, + pre_nms_top_n=12000, + post_nms_top_n=2000, + nms_thresh=.5, + min_size=.1, + eta=1., + topk_after_collect=False): + super(ProposalGenerator, self).__init__() + self.pre_nms_top_n = pre_nms_top_n + self.post_nms_top_n = post_nms_top_n + self.nms_thresh = nms_thresh + self.min_size = min_size + self.eta = eta + self.topk_after_collect = topk_after_collect + + def __call__(self, scores, bbox_deltas, anchors, im_shape): + + top_n = self.pre_nms_top_n if self.topk_after_collect else self.post_nms_top_n + variances = paddle.ones_like(anchors) + rpn_rois, rpn_rois_prob, rpn_rois_num = ops.generate_proposals( + scores, + bbox_deltas, + im_shape, + anchors, + variances, + pre_nms_top_n=self.pre_nms_top_n, + post_nms_top_n=top_n, + nms_thresh=self.nms_thresh, + min_size=self.min_size, + eta=self.eta, + return_rois_num=True) + return rpn_rois, rpn_rois_prob, rpn_rois_num, self.post_nms_top_n diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/rpn_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/rpn_head.py new file mode 100644 index 000000000..1664d7839 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/rpn_head.py @@ -0,0 +1,259 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal + +from ppdet.core.workspace import register +from .anchor_generator import AnchorGenerator +from .target_layer import RPNTargetAssign +from .proposal_generator import ProposalGenerator + + +class RPNFeat(nn.Layer): + """ + Feature extraction in RPN head + + Args: + in_channel (int): Input channel + out_channel (int): Output channel + """ + + def __init__(self, in_channel=1024, out_channel=1024): + super(RPNFeat, self).__init__() + # rpn feat is shared with each level + self.rpn_conv = nn.Conv2D( + in_channels=in_channel, + out_channels=out_channel, + kernel_size=3, + padding=1, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0., std=0.01))) + self.rpn_conv.skip_quant = True + + def forward(self, feats): + rpn_feats = [] + for feat in feats: + rpn_feats.append(F.relu(self.rpn_conv(feat))) + return rpn_feats + + +@register +class RPNHead(nn.Layer): + """ + Region Proposal Network + + Args: + anchor_generator (dict): configure of anchor generation + rpn_target_assign (dict): configure of rpn targets assignment + train_proposal (dict): configure of proposals generation + at the stage of training + test_proposal (dict): configure of proposals generation + at the stage of prediction + in_channel (int): channel of input feature maps which can be + derived by from_config + """ + + def __init__(self, + anchor_generator=AnchorGenerator().__dict__, + rpn_target_assign=RPNTargetAssign().__dict__, + train_proposal=ProposalGenerator(12000, 2000).__dict__, + test_proposal=ProposalGenerator().__dict__, + in_channel=1024): + super(RPNHead, self).__init__() + self.anchor_generator = anchor_generator + self.rpn_target_assign = rpn_target_assign + self.train_proposal = train_proposal + self.test_proposal = test_proposal + if isinstance(anchor_generator, dict): + self.anchor_generator = AnchorGenerator(**anchor_generator) + if isinstance(rpn_target_assign, dict): + self.rpn_target_assign = RPNTargetAssign(**rpn_target_assign) + if isinstance(train_proposal, dict): + self.train_proposal = ProposalGenerator(**train_proposal) + if isinstance(test_proposal, dict): + self.test_proposal = ProposalGenerator(**test_proposal) + + num_anchors = self.anchor_generator.num_anchors + self.rpn_feat = RPNFeat(in_channel, in_channel) + # rpn head is shared with each level + # rpn roi classification scores + self.rpn_rois_score = nn.Conv2D( + in_channels=in_channel, + out_channels=num_anchors, + kernel_size=1, + padding=0, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0., std=0.01))) + self.rpn_rois_score.skip_quant = True + + # rpn roi bbox regression deltas + self.rpn_rois_delta = nn.Conv2D( + in_channels=in_channel, + out_channels=4 * num_anchors, + kernel_size=1, + padding=0, + weight_attr=paddle.ParamAttr(initializer=Normal( + mean=0., std=0.01))) + self.rpn_rois_delta.skip_quant = True + + @classmethod + def from_config(cls, cfg, input_shape): + # FPN share same rpn head + if isinstance(input_shape, (list, tuple)): + input_shape = input_shape[0] + return {'in_channel': input_shape.channels} + + def forward(self, feats, inputs): + rpn_feats = self.rpn_feat(feats) + scores = [] + deltas = [] + + for rpn_feat in rpn_feats: + rrs = self.rpn_rois_score(rpn_feat) + rrd = self.rpn_rois_delta(rpn_feat) + scores.append(rrs) + deltas.append(rrd) + + anchors = self.anchor_generator(rpn_feats) + + rois, rois_num = self._gen_proposal(scores, deltas, anchors, inputs) + if self.training: + loss = self.get_loss(scores, deltas, anchors, inputs) + return rois, rois_num, loss + else: + return rois, rois_num, None + + def _gen_proposal(self, scores, bbox_deltas, anchors, inputs): + """ + scores (list[Tensor]): Multi-level scores prediction + bbox_deltas (list[Tensor]): Multi-level deltas prediction + anchors (list[Tensor]): Multi-level anchors + inputs (dict): ground truth info + """ + prop_gen = self.train_proposal if self.training else self.test_proposal + im_shape = inputs['im_shape'] + + # Collect multi-level proposals for each batch + # Get 'topk' of them as final output + bs_rois_collect = [] + bs_rois_num_collect = [] + batch_size = paddle.slice(paddle.shape(im_shape), [0], [0], [1]) + + # Generate proposals for each level and each batch. + # Discard batch-computing to avoid sorting bbox cross different batches. + for i in range(batch_size): + rpn_rois_list = [] + rpn_prob_list = [] + rpn_rois_num_list = [] + + for rpn_score, rpn_delta, anchor in zip(scores, bbox_deltas, + anchors): + rpn_rois, rpn_rois_prob, rpn_rois_num, post_nms_top_n = prop_gen( + scores=rpn_score[i:i + 1], + bbox_deltas=rpn_delta[i:i + 1], + anchors=anchor, + im_shape=im_shape[i:i + 1]) + if rpn_rois.shape[0] > 0: + rpn_rois_list.append(rpn_rois) + rpn_prob_list.append(rpn_rois_prob) + rpn_rois_num_list.append(rpn_rois_num) + + if len(scores) > 1: + rpn_rois = paddle.concat(rpn_rois_list) + rpn_prob = paddle.concat(rpn_prob_list).flatten() + + if rpn_prob.shape[0] > post_nms_top_n: + topk_prob, topk_inds = paddle.topk(rpn_prob, post_nms_top_n) + topk_rois = paddle.gather(rpn_rois, topk_inds) + else: + topk_rois = rpn_rois + topk_prob = rpn_prob + else: + topk_rois = rpn_rois_list[0] + topk_prob = rpn_prob_list[0].flatten() + + bs_rois_collect.append(topk_rois) + bs_rois_num_collect.append(paddle.shape(topk_rois)[0]) + + bs_rois_num_collect = paddle.concat(bs_rois_num_collect) + + return bs_rois_collect, bs_rois_num_collect + + def get_loss(self, pred_scores, pred_deltas, anchors, inputs): + """ + pred_scores (list[Tensor]): Multi-level scores prediction + pred_deltas (list[Tensor]): Multi-level deltas prediction + anchors (list[Tensor]): Multi-level anchors + inputs (dict): ground truth info, including im, gt_bbox, gt_score + """ + anchors = [paddle.reshape(a, shape=(-1, 4)) for a in anchors] + anchors = paddle.concat(anchors) + + scores = [ + paddle.reshape( + paddle.transpose( + v, perm=[0, 2, 3, 1]), + shape=(v.shape[0], -1, 1)) for v in pred_scores + ] + scores = paddle.concat(scores, axis=1) + + deltas = [ + paddle.reshape( + paddle.transpose( + v, perm=[0, 2, 3, 1]), + shape=(v.shape[0], -1, 4)) for v in pred_deltas + ] + deltas = paddle.concat(deltas, axis=1) + + score_tgt, bbox_tgt, loc_tgt, norm = self.rpn_target_assign(inputs, + anchors) + + scores = paddle.reshape(x=scores, shape=(-1, )) + deltas = paddle.reshape(x=deltas, shape=(-1, 4)) + + score_tgt = paddle.concat(score_tgt) + score_tgt.stop_gradient = True + + pos_mask = score_tgt == 1 + pos_ind = paddle.nonzero(pos_mask) + + valid_mask = score_tgt >= 0 + valid_ind = paddle.nonzero(valid_mask) + + # cls loss + if valid_ind.shape[0] == 0: + loss_rpn_cls = paddle.zeros([1], dtype='float32') + else: + score_pred = paddle.gather(scores, valid_ind) + score_label = paddle.gather(score_tgt, valid_ind).cast('float32') + score_label.stop_gradient = True + loss_rpn_cls = F.binary_cross_entropy_with_logits( + logit=score_pred, label=score_label, reduction="sum") + + # reg loss + if pos_ind.shape[0] == 0: + loss_rpn_reg = paddle.zeros([1], dtype='float32') + else: + loc_pred = paddle.gather(deltas, pos_ind) + loc_tgt = paddle.concat(loc_tgt) + loc_tgt = paddle.gather(loc_tgt, pos_ind) + loc_tgt.stop_gradient = True + loss_rpn_reg = paddle.abs(loc_pred - loc_tgt).sum() + return { + 'loss_rpn_cls': loss_rpn_cls / norm, + 'loss_rpn_reg': loss_rpn_reg / norm + } diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target.py new file mode 100644 index 000000000..af83cfdb8 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target.py @@ -0,0 +1,675 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +import paddle +from ..bbox_utils import bbox2delta, bbox_overlaps + + +def rpn_anchor_target(anchors, + gt_boxes, + rpn_batch_size_per_im, + rpn_positive_overlap, + rpn_negative_overlap, + rpn_fg_fraction, + use_random=True, + batch_size=1, + ignore_thresh=-1, + is_crowd=None, + weights=[1., 1., 1., 1.], + assign_on_cpu=False): + tgt_labels = [] + tgt_bboxes = [] + tgt_deltas = [] + for i in range(batch_size): + gt_bbox = gt_boxes[i] + is_crowd_i = is_crowd[i] if is_crowd else None + # Step1: match anchor and gt_bbox + matches, match_labels = label_box( + anchors, gt_bbox, rpn_positive_overlap, rpn_negative_overlap, True, + ignore_thresh, is_crowd_i, assign_on_cpu) + # Step2: sample anchor + fg_inds, bg_inds = subsample_labels(match_labels, rpn_batch_size_per_im, + rpn_fg_fraction, 0, use_random) + # Fill with the ignore label (-1), then set positive and negative labels + labels = paddle.full(match_labels.shape, -1, dtype='int32') + if bg_inds.shape[0] > 0: + labels = paddle.scatter(labels, bg_inds, paddle.zeros_like(bg_inds)) + if fg_inds.shape[0] > 0: + labels = paddle.scatter(labels, fg_inds, paddle.ones_like(fg_inds)) + # Step3: make output + if gt_bbox.shape[0] == 0: + matched_gt_boxes = paddle.zeros([0, 4]) + tgt_delta = paddle.zeros([0, 4]) + else: + matched_gt_boxes = paddle.gather(gt_bbox, matches) + tgt_delta = bbox2delta(anchors, matched_gt_boxes, weights) + matched_gt_boxes.stop_gradient = True + tgt_delta.stop_gradient = True + labels.stop_gradient = True + tgt_labels.append(labels) + tgt_bboxes.append(matched_gt_boxes) + tgt_deltas.append(tgt_delta) + + return tgt_labels, tgt_bboxes, tgt_deltas + + +def label_box(anchors, + gt_boxes, + positive_overlap, + negative_overlap, + allow_low_quality, + ignore_thresh, + is_crowd=None, + assign_on_cpu=False): + if assign_on_cpu: + paddle.set_device("cpu") + iou = bbox_overlaps(gt_boxes, anchors) + paddle.set_device("gpu") + else: + iou = bbox_overlaps(gt_boxes, anchors) + n_gt = gt_boxes.shape[0] + if n_gt == 0 or is_crowd is None: + n_gt_crowd = 0 + else: + n_gt_crowd = paddle.nonzero(is_crowd).shape[0] + if iou.shape[0] == 0 or n_gt_crowd == n_gt: + # No truth, assign everything to background + default_matches = paddle.full((iou.shape[1], ), 0, dtype='int64') + default_match_labels = paddle.full((iou.shape[1], ), 0, dtype='int32') + return default_matches, default_match_labels + # if ignore_thresh > 0, remove anchor if it is closed to + # one of the crowded ground-truth + if n_gt_crowd > 0: + N_a = anchors.shape[0] + ones = paddle.ones([N_a]) + mask = is_crowd * ones + + if ignore_thresh > 0: + crowd_iou = iou * mask + valid = (paddle.sum((crowd_iou > ignore_thresh).cast('int32'), + axis=0) > 0).cast('float32') + iou = iou * (1 - valid) - valid + + # ignore the iou between anchor and crowded ground-truth + iou = iou * (1 - mask) - mask + + matched_vals, matches = paddle.topk(iou, k=1, axis=0) + match_labels = paddle.full(matches.shape, -1, dtype='int32') + # set ignored anchor with iou = -1 + neg_cond = paddle.logical_and(matched_vals > -1, + matched_vals < negative_overlap) + match_labels = paddle.where(neg_cond, + paddle.zeros_like(match_labels), match_labels) + match_labels = paddle.where(matched_vals >= positive_overlap, + paddle.ones_like(match_labels), match_labels) + if allow_low_quality: + highest_quality_foreach_gt = iou.max(axis=1, keepdim=True) + pred_inds_with_highest_quality = paddle.logical_and( + iou > 0, iou == highest_quality_foreach_gt).cast('int32').sum( + 0, keepdim=True) + match_labels = paddle.where(pred_inds_with_highest_quality > 0, + paddle.ones_like(match_labels), + match_labels) + + matches = matches.flatten() + match_labels = match_labels.flatten() + + return matches, match_labels + + +def subsample_labels(labels, + num_samples, + fg_fraction, + bg_label=0, + use_random=True): + positive = paddle.nonzero( + paddle.logical_and(labels != -1, labels != bg_label)) + negative = paddle.nonzero(labels == bg_label) + + fg_num = int(num_samples * fg_fraction) + fg_num = min(positive.numel(), fg_num) + bg_num = num_samples - fg_num + bg_num = min(negative.numel(), bg_num) + if fg_num == 0 and bg_num == 0: + fg_inds = paddle.zeros([0], dtype='int32') + bg_inds = paddle.zeros([0], dtype='int32') + return fg_inds, bg_inds + + # randomly select positive and negative examples + + negative = negative.cast('int32').flatten() + bg_perm = paddle.randperm(negative.numel(), dtype='int32') + bg_perm = paddle.slice(bg_perm, axes=[0], starts=[0], ends=[bg_num]) + if use_random: + bg_inds = paddle.gather(negative, bg_perm) + else: + bg_inds = paddle.slice(negative, axes=[0], starts=[0], ends=[bg_num]) + if fg_num == 0: + fg_inds = paddle.zeros([0], dtype='int32') + return fg_inds, bg_inds + + positive = positive.cast('int32').flatten() + fg_perm = paddle.randperm(positive.numel(), dtype='int32') + fg_perm = paddle.slice(fg_perm, axes=[0], starts=[0], ends=[fg_num]) + if use_random: + fg_inds = paddle.gather(positive, fg_perm) + else: + fg_inds = paddle.slice(positive, axes=[0], starts=[0], ends=[fg_num]) + + return fg_inds, bg_inds + + +def generate_proposal_target(rpn_rois, + gt_classes, + gt_boxes, + batch_size_per_im, + fg_fraction, + fg_thresh, + bg_thresh, + num_classes, + ignore_thresh=-1., + is_crowd=None, + use_random=True, + is_cascade=False, + cascade_iou=0.5, + assign_on_cpu=False): + + rois_with_gt = [] + tgt_labels = [] + tgt_bboxes = [] + tgt_gt_inds = [] + new_rois_num = [] + + # In cascade rcnn, the threshold for foreground and background + # is used from cascade_iou + fg_thresh = cascade_iou if is_cascade else fg_thresh + bg_thresh = cascade_iou if is_cascade else bg_thresh + for i, rpn_roi in enumerate(rpn_rois): + gt_bbox = gt_boxes[i] + is_crowd_i = is_crowd[i] if is_crowd else None + gt_class = paddle.squeeze(gt_classes[i], axis=-1) + + # Concat RoIs and gt boxes except cascade rcnn or none gt + if not is_cascade and gt_bbox.shape[0] > 0: + bbox = paddle.concat([rpn_roi, gt_bbox]) + else: + bbox = rpn_roi + + # Step1: label bbox + matches, match_labels = label_box(bbox, gt_bbox, fg_thresh, bg_thresh, + False, ignore_thresh, is_crowd_i, + assign_on_cpu) + # Step2: sample bbox + sampled_inds, sampled_gt_classes = sample_bbox( + matches, match_labels, gt_class, batch_size_per_im, fg_fraction, + num_classes, use_random, is_cascade) + + # Step3: make output + rois_per_image = bbox if is_cascade else paddle.gather(bbox, + sampled_inds) + sampled_gt_ind = matches if is_cascade else paddle.gather(matches, + sampled_inds) + if gt_bbox.shape[0] > 0: + sampled_bbox = paddle.gather(gt_bbox, sampled_gt_ind) + else: + num = rois_per_image.shape[0] + sampled_bbox = paddle.zeros([num, 4], dtype='float32') + + rois_per_image.stop_gradient = True + sampled_gt_ind.stop_gradient = True + sampled_bbox.stop_gradient = True + tgt_labels.append(sampled_gt_classes) + tgt_bboxes.append(sampled_bbox) + rois_with_gt.append(rois_per_image) + tgt_gt_inds.append(sampled_gt_ind) + new_rois_num.append(paddle.shape(sampled_inds)[0]) + new_rois_num = paddle.concat(new_rois_num) + return rois_with_gt, tgt_labels, tgt_bboxes, tgt_gt_inds, new_rois_num + + +def sample_bbox(matches, + match_labels, + gt_classes, + batch_size_per_im, + fg_fraction, + num_classes, + use_random=True, + is_cascade=False): + + n_gt = gt_classes.shape[0] + if n_gt == 0: + # No truth, assign everything to background + gt_classes = paddle.ones(matches.shape, dtype='int32') * num_classes + #return matches, match_labels + num_classes + else: + gt_classes = paddle.gather(gt_classes, matches) + gt_classes = paddle.where(match_labels == 0, + paddle.ones_like(gt_classes) * num_classes, + gt_classes) + gt_classes = paddle.where(match_labels == -1, + paddle.ones_like(gt_classes) * -1, gt_classes) + if is_cascade: + index = paddle.arange(matches.shape[0]) + return index, gt_classes + rois_per_image = int(batch_size_per_im) + + fg_inds, bg_inds = subsample_labels(gt_classes, rois_per_image, fg_fraction, + num_classes, use_random) + if fg_inds.shape[0] == 0 and bg_inds.shape[0] == 0: + # fake output labeled with -1 when all boxes are neither + # foreground nor background + sampled_inds = paddle.zeros([1], dtype='int32') + else: + sampled_inds = paddle.concat([fg_inds, bg_inds]) + sampled_gt_classes = paddle.gather(gt_classes, sampled_inds) + return sampled_inds, sampled_gt_classes + + +def polygons_to_mask(polygons, height, width): + """ + Convert the polygons to mask format + + Args: + polygons (list[ndarray]): each array has shape (Nx2,) + height (int): mask height + width (int): mask width + Returns: + ndarray: a bool mask of shape (height, width) + """ + import pycocotools.mask as mask_util + assert len(polygons) > 0, "COCOAPI does not support empty polygons" + rles = mask_util.frPyObjects(polygons, height, width) + rle = mask_util.merge(rles) + return mask_util.decode(rle).astype(np.bool) + + +def rasterize_polygons_within_box(poly, box, resolution): + w, h = box[2] - box[0], box[3] - box[1] + polygons = [np.asarray(p, dtype=np.float64) for p in poly] + for p in polygons: + p[0::2] = p[0::2] - box[0] + p[1::2] = p[1::2] - box[1] + + ratio_h = resolution / max(h, 0.1) + ratio_w = resolution / max(w, 0.1) + + if ratio_h == ratio_w: + for p in polygons: + p *= ratio_h + else: + for p in polygons: + p[0::2] *= ratio_w + p[1::2] *= ratio_h + + # 3. Rasterize the polygons with coco api + mask = polygons_to_mask(polygons, resolution, resolution) + mask = paddle.to_tensor(mask, dtype='int32') + return mask + + +def generate_mask_target(gt_segms, rois, labels_int32, sampled_gt_inds, + num_classes, resolution): + mask_rois = [] + mask_rois_num = [] + tgt_masks = [] + tgt_classes = [] + mask_index = [] + tgt_weights = [] + for k in range(len(rois)): + labels_per_im = labels_int32[k] + # select rois labeled with foreground + fg_inds = paddle.nonzero( + paddle.logical_and(labels_per_im != -1, labels_per_im != + num_classes)) + has_fg = True + # generate fake roi if foreground is empty + if fg_inds.numel() == 0: + has_fg = False + fg_inds = paddle.ones([1], dtype='int32') + inds_per_im = sampled_gt_inds[k] + inds_per_im = paddle.gather(inds_per_im, fg_inds) + + rois_per_im = rois[k] + fg_rois = paddle.gather(rois_per_im, fg_inds) + # Copy the foreground roi to cpu + # to generate mask target with ground-truth + boxes = fg_rois.numpy() + gt_segms_per_im = gt_segms[k] + + new_segm = [] + inds_per_im = inds_per_im.numpy() + if len(gt_segms_per_im) > 0: + for i in inds_per_im: + new_segm.append(gt_segms_per_im[i]) + fg_inds_new = fg_inds.reshape([-1]).numpy() + results = [] + if len(gt_segms_per_im) > 0: + for j in fg_inds_new: + results.append( + rasterize_polygons_within_box(new_segm[j], boxes[j], + resolution)) + else: + results.append(paddle.ones([resolution, resolution], dtype='int32')) + + fg_classes = paddle.gather(labels_per_im, fg_inds) + weight = paddle.ones([fg_rois.shape[0]], dtype='float32') + if not has_fg: + # now all sampled classes are background + # which will cause error in loss calculation, + # make fake classes with weight of 0. + fg_classes = paddle.zeros([1], dtype='int32') + weight = weight - 1 + tgt_mask = paddle.stack(results) + tgt_mask.stop_gradient = True + fg_rois.stop_gradient = True + + mask_index.append(fg_inds) + mask_rois.append(fg_rois) + mask_rois_num.append(paddle.shape(fg_rois)[0]) + tgt_classes.append(fg_classes) + tgt_masks.append(tgt_mask) + tgt_weights.append(weight) + + mask_index = paddle.concat(mask_index) + mask_rois_num = paddle.concat(mask_rois_num) + tgt_classes = paddle.concat(tgt_classes, axis=0) + tgt_masks = paddle.concat(tgt_masks, axis=0) + tgt_weights = paddle.concat(tgt_weights, axis=0) + + return mask_rois, mask_rois_num, tgt_classes, tgt_masks, mask_index, tgt_weights + + +def libra_sample_pos(max_overlaps, max_classes, pos_inds, num_expected): + if len(pos_inds) <= num_expected: + return pos_inds + else: + unique_gt_inds = np.unique(max_classes[pos_inds]) + num_gts = len(unique_gt_inds) + num_per_gt = int(round(num_expected / float(num_gts)) + 1) + + sampled_inds = [] + for i in unique_gt_inds: + inds = np.nonzero(max_classes == i)[0] + before_len = len(inds) + inds = list(set(inds) & set(pos_inds)) + after_len = len(inds) + if len(inds) > num_per_gt: + inds = np.random.choice(inds, size=num_per_gt, replace=False) + sampled_inds.extend(list(inds)) # combine as a new sampler + if len(sampled_inds) < num_expected: + num_extra = num_expected - len(sampled_inds) + extra_inds = np.array(list(set(pos_inds) - set(sampled_inds))) + assert len(sampled_inds) + len(extra_inds) == len(pos_inds), \ + "sum of sampled_inds({}) and extra_inds({}) length must be equal with pos_inds({})!".format( + len(sampled_inds), len(extra_inds), len(pos_inds)) + if len(extra_inds) > num_extra: + extra_inds = np.random.choice( + extra_inds, size=num_extra, replace=False) + sampled_inds.extend(extra_inds.tolist()) + elif len(sampled_inds) > num_expected: + sampled_inds = np.random.choice( + sampled_inds, size=num_expected, replace=False) + return paddle.to_tensor(sampled_inds) + + +def libra_sample_via_interval(max_overlaps, full_set, num_expected, floor_thr, + num_bins, bg_thresh): + max_iou = max_overlaps.max() + iou_interval = (max_iou - floor_thr) / num_bins + per_num_expected = int(num_expected / num_bins) + + sampled_inds = [] + for i in range(num_bins): + start_iou = floor_thr + i * iou_interval + end_iou = floor_thr + (i + 1) * iou_interval + + tmp_set = set( + np.where( + np.logical_and(max_overlaps >= start_iou, max_overlaps < + end_iou))[0]) + tmp_inds = list(tmp_set & full_set) + + if len(tmp_inds) > per_num_expected: + tmp_sampled_set = np.random.choice( + tmp_inds, size=per_num_expected, replace=False) + else: + tmp_sampled_set = np.array(tmp_inds, dtype=np.int) + sampled_inds.append(tmp_sampled_set) + + sampled_inds = np.concatenate(sampled_inds) + if len(sampled_inds) < num_expected: + num_extra = num_expected - len(sampled_inds) + extra_inds = np.array(list(full_set - set(sampled_inds))) + assert len(sampled_inds) + len(extra_inds) == len(full_set), \ + "sum of sampled_inds({}) and extra_inds({}) length must be equal with full_set({})!".format( + len(sampled_inds), len(extra_inds), len(full_set)) + + if len(extra_inds) > num_extra: + extra_inds = np.random.choice(extra_inds, num_extra, replace=False) + sampled_inds = np.concatenate([sampled_inds, extra_inds]) + + return sampled_inds + + +def libra_sample_neg(max_overlaps, + max_classes, + neg_inds, + num_expected, + floor_thr=-1, + floor_fraction=0, + num_bins=3, + bg_thresh=0.5): + if len(neg_inds) <= num_expected: + return neg_inds + else: + # balance sampling for negative samples + neg_set = set(neg_inds.tolist()) + if floor_thr > 0: + floor_set = set( + np.where( + np.logical_and(max_overlaps >= 0, max_overlaps < floor_thr)) + [0]) + iou_sampling_set = set(np.where(max_overlaps >= floor_thr)[0]) + elif floor_thr == 0: + floor_set = set(np.where(max_overlaps == 0)[0]) + iou_sampling_set = set(np.where(max_overlaps > floor_thr)[0]) + else: + floor_set = set() + iou_sampling_set = set(np.where(max_overlaps > floor_thr)[0]) + floor_thr = 0 + + floor_neg_inds = list(floor_set & neg_set) + iou_sampling_neg_inds = list(iou_sampling_set & neg_set) + + num_expected_iou_sampling = int(num_expected * (1 - floor_fraction)) + if len(iou_sampling_neg_inds) > num_expected_iou_sampling: + if num_bins >= 2: + iou_sampled_inds = libra_sample_via_interval( + max_overlaps, + set(iou_sampling_neg_inds), num_expected_iou_sampling, + floor_thr, num_bins, bg_thresh) + else: + iou_sampled_inds = np.random.choice( + iou_sampling_neg_inds, + size=num_expected_iou_sampling, + replace=False) + else: + iou_sampled_inds = np.array(iou_sampling_neg_inds, dtype=np.int) + num_expected_floor = num_expected - len(iou_sampled_inds) + if len(floor_neg_inds) > num_expected_floor: + sampled_floor_inds = np.random.choice( + floor_neg_inds, size=num_expected_floor, replace=False) + else: + sampled_floor_inds = np.array(floor_neg_inds, dtype=np.int) + sampled_inds = np.concatenate((sampled_floor_inds, iou_sampled_inds)) + if len(sampled_inds) < num_expected: + num_extra = num_expected - len(sampled_inds) + extra_inds = np.array(list(neg_set - set(sampled_inds))) + if len(extra_inds) > num_extra: + extra_inds = np.random.choice( + extra_inds, size=num_extra, replace=False) + sampled_inds = np.concatenate((sampled_inds, extra_inds)) + return paddle.to_tensor(sampled_inds) + + +def libra_label_box(anchors, gt_boxes, gt_classes, positive_overlap, + negative_overlap, num_classes): + # TODO: use paddle API to speed up + gt_classes = gt_classes.numpy() + gt_overlaps = np.zeros((anchors.shape[0], num_classes)) + matches = np.zeros((anchors.shape[0]), dtype=np.int32) + if len(gt_boxes) > 0: + proposal_to_gt_overlaps = bbox_overlaps(anchors, gt_boxes).numpy() + overlaps_argmax = proposal_to_gt_overlaps.argmax(axis=1) + overlaps_max = proposal_to_gt_overlaps.max(axis=1) + # Boxes which with non-zero overlap with gt boxes + overlapped_boxes_ind = np.where(overlaps_max > 0)[0] + overlapped_boxes_gt_classes = gt_classes[overlaps_argmax[ + overlapped_boxes_ind]] + + for idx in range(len(overlapped_boxes_ind)): + gt_overlaps[overlapped_boxes_ind[idx], overlapped_boxes_gt_classes[ + idx]] = overlaps_max[overlapped_boxes_ind[idx]] + matches[overlapped_boxes_ind[idx]] = overlaps_argmax[ + overlapped_boxes_ind[idx]] + + gt_overlaps = paddle.to_tensor(gt_overlaps) + matches = paddle.to_tensor(matches) + + matched_vals = paddle.max(gt_overlaps, axis=1) + match_labels = paddle.full(matches.shape, -1, dtype='int32') + match_labels = paddle.where(matched_vals < negative_overlap, + paddle.zeros_like(match_labels), match_labels) + match_labels = paddle.where(matched_vals >= positive_overlap, + paddle.ones_like(match_labels), match_labels) + + return matches, match_labels, matched_vals + + +def libra_sample_bbox(matches, + match_labels, + matched_vals, + gt_classes, + batch_size_per_im, + num_classes, + fg_fraction, + fg_thresh, + bg_thresh, + num_bins, + use_random=True, + is_cascade_rcnn=False): + rois_per_image = int(batch_size_per_im) + fg_rois_per_im = int(np.round(fg_fraction * rois_per_image)) + bg_rois_per_im = rois_per_image - fg_rois_per_im + + if is_cascade_rcnn: + fg_inds = paddle.nonzero(matched_vals >= fg_thresh) + bg_inds = paddle.nonzero(matched_vals < bg_thresh) + else: + matched_vals_np = matched_vals.numpy() + match_labels_np = match_labels.numpy() + + # sample fg + fg_inds = paddle.nonzero(matched_vals >= fg_thresh).flatten() + fg_nums = int(np.minimum(fg_rois_per_im, fg_inds.shape[0])) + if (fg_inds.shape[0] > fg_nums) and use_random: + fg_inds = libra_sample_pos(matched_vals_np, match_labels_np, + fg_inds.numpy(), fg_rois_per_im) + fg_inds = fg_inds[:fg_nums] + + # sample bg + bg_inds = paddle.nonzero(matched_vals < bg_thresh).flatten() + bg_nums = int(np.minimum(rois_per_image - fg_nums, bg_inds.shape[0])) + if (bg_inds.shape[0] > bg_nums) and use_random: + bg_inds = libra_sample_neg( + matched_vals_np, + match_labels_np, + bg_inds.numpy(), + bg_rois_per_im, + num_bins=num_bins, + bg_thresh=bg_thresh) + bg_inds = bg_inds[:bg_nums] + + sampled_inds = paddle.concat([fg_inds, bg_inds]) + + gt_classes = paddle.gather(gt_classes, matches) + gt_classes = paddle.where(match_labels == 0, + paddle.ones_like(gt_classes) * num_classes, + gt_classes) + gt_classes = paddle.where(match_labels == -1, + paddle.ones_like(gt_classes) * -1, gt_classes) + sampled_gt_classes = paddle.gather(gt_classes, sampled_inds) + + return sampled_inds, sampled_gt_classes + + +def libra_generate_proposal_target(rpn_rois, + gt_classes, + gt_boxes, + batch_size_per_im, + fg_fraction, + fg_thresh, + bg_thresh, + num_classes, + use_random=True, + is_cascade_rcnn=False, + max_overlaps=None, + num_bins=3): + + rois_with_gt = [] + tgt_labels = [] + tgt_bboxes = [] + sampled_max_overlaps = [] + tgt_gt_inds = [] + new_rois_num = [] + + for i, rpn_roi in enumerate(rpn_rois): + max_overlap = max_overlaps[i] if is_cascade_rcnn else None + gt_bbox = gt_boxes[i] + gt_class = paddle.squeeze(gt_classes[i], axis=-1) + if is_cascade_rcnn: + rpn_roi = filter_roi(rpn_roi, max_overlap) + bbox = paddle.concat([rpn_roi, gt_bbox]) + + # Step1: label bbox + matches, match_labels, matched_vals = libra_label_box( + bbox, gt_bbox, gt_class, fg_thresh, bg_thresh, num_classes) + + # Step2: sample bbox + sampled_inds, sampled_gt_classes = libra_sample_bbox( + matches, match_labels, matched_vals, gt_class, batch_size_per_im, + num_classes, fg_fraction, fg_thresh, bg_thresh, num_bins, + use_random, is_cascade_rcnn) + + # Step3: make output + rois_per_image = paddle.gather(bbox, sampled_inds) + sampled_gt_ind = paddle.gather(matches, sampled_inds) + sampled_bbox = paddle.gather(gt_bbox, sampled_gt_ind) + sampled_overlap = paddle.gather(matched_vals, sampled_inds) + + rois_per_image.stop_gradient = True + sampled_gt_ind.stop_gradient = True + sampled_bbox.stop_gradient = True + sampled_overlap.stop_gradient = True + + tgt_labels.append(sampled_gt_classes) + tgt_bboxes.append(sampled_bbox) + rois_with_gt.append(rois_per_image) + sampled_max_overlaps.append(sampled_overlap) + tgt_gt_inds.append(sampled_gt_ind) + new_rois_num.append(paddle.shape(sampled_inds)[0]) + new_rois_num = paddle.concat(new_rois_num) + # rois_with_gt, tgt_labels, tgt_bboxes, tgt_gt_inds, new_rois_num + return rois_with_gt, tgt_labels, tgt_bboxes, tgt_gt_inds, new_rois_num diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target_layer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target_layer.py new file mode 100644 index 000000000..3b5a09601 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/proposal_generator/target_layer.py @@ -0,0 +1,490 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import sys +import paddle +from ppdet.core.workspace import register, serializable + +from .target import rpn_anchor_target, generate_proposal_target, generate_mask_target, libra_generate_proposal_target +import numpy as np + + +@register +@serializable +class RPNTargetAssign(object): + __shared__ = ['assign_on_cpu'] + """ + RPN targets assignment module + + The assignment consists of three steps: + 1. Match anchor and ground-truth box, label the anchor with foreground + or background sample + 2. Sample anchors to keep the properly ratio between foreground and + background + 3. Generate the targets for classification and regression branch + + + Args: + batch_size_per_im (int): Total number of RPN samples per image. + default 256 + fg_fraction (float): Fraction of anchors that is labeled + foreground, default 0.5 + positive_overlap (float): Minimum overlap required between an anchor + and ground-truth box for the (anchor, gt box) pair to be + a foreground sample. default 0.7 + negative_overlap (float): Maximum overlap allowed between an anchor + and ground-truth box for the (anchor, gt box) pair to be + a background sample. default 0.3 + ignore_thresh(float): Threshold for ignoring the is_crowd ground-truth + if the value is larger than zero. + use_random (bool): Use random sampling to choose foreground and + background boxes, default true. + assign_on_cpu (bool): In case the number of gt box is too large, + compute IoU on CPU, default false. + """ + + def __init__(self, + batch_size_per_im=256, + fg_fraction=0.5, + positive_overlap=0.7, + negative_overlap=0.3, + ignore_thresh=-1., + use_random=True, + assign_on_cpu=False): + super(RPNTargetAssign, self).__init__() + self.batch_size_per_im = batch_size_per_im + self.fg_fraction = fg_fraction + self.positive_overlap = positive_overlap + self.negative_overlap = negative_overlap + self.ignore_thresh = ignore_thresh + self.use_random = use_random + self.assign_on_cpu = assign_on_cpu + + def __call__(self, inputs, anchors): + """ + inputs: ground-truth instances. + anchor_box (Tensor): [num_anchors, 4], num_anchors are all anchors in all feature maps. + """ + gt_boxes = inputs['gt_bbox'] + is_crowd = inputs.get('is_crowd', None) + batch_size = len(gt_boxes) + tgt_labels, tgt_bboxes, tgt_deltas = rpn_anchor_target( + anchors, + gt_boxes, + self.batch_size_per_im, + self.positive_overlap, + self.negative_overlap, + self.fg_fraction, + self.use_random, + batch_size, + self.ignore_thresh, + is_crowd, + assign_on_cpu=self.assign_on_cpu) + norm = self.batch_size_per_im * batch_size + + return tgt_labels, tgt_bboxes, tgt_deltas, norm + + +@register +class BBoxAssigner(object): + __shared__ = ['num_classes', 'assign_on_cpu'] + """ + RCNN targets assignment module + + The assignment consists of three steps: + 1. Match RoIs and ground-truth box, label the RoIs with foreground + or background sample + 2. Sample anchors to keep the properly ratio between foreground and + background + 3. Generate the targets for classification and regression branch + + Args: + batch_size_per_im (int): Total number of RoIs per image. + default 512 + fg_fraction (float): Fraction of RoIs that is labeled + foreground, default 0.25 + fg_thresh (float): Minimum overlap required between a RoI + and ground-truth box for the (roi, gt box) pair to be + a foreground sample. default 0.5 + bg_thresh (float): Maximum overlap allowed between a RoI + and ground-truth box for the (roi, gt box) pair to be + a background sample. default 0.5 + ignore_thresh(float): Threshold for ignoring the is_crowd ground-truth + if the value is larger than zero. + use_random (bool): Use random sampling to choose foreground and + background boxes, default true + cascade_iou (list[iou]): The list of overlap to select foreground and + background of each stage, which is only used In Cascade RCNN. + num_classes (int): The number of class. + assign_on_cpu (bool): In case the number of gt box is too large, + compute IoU on CPU, default false. + """ + + def __init__(self, + batch_size_per_im=512, + fg_fraction=.25, + fg_thresh=.5, + bg_thresh=.5, + ignore_thresh=-1., + use_random=True, + cascade_iou=[0.5, 0.6, 0.7], + num_classes=80, + assign_on_cpu=False): + super(BBoxAssigner, self).__init__() + self.batch_size_per_im = batch_size_per_im + self.fg_fraction = fg_fraction + self.fg_thresh = fg_thresh + self.bg_thresh = bg_thresh + self.ignore_thresh = ignore_thresh + self.use_random = use_random + self.cascade_iou = cascade_iou + self.num_classes = num_classes + self.assign_on_cpu = assign_on_cpu + + def __call__(self, + rpn_rois, + rpn_rois_num, + inputs, + stage=0, + is_cascade=False): + gt_classes = inputs['gt_class'] + gt_boxes = inputs['gt_bbox'] + is_crowd = inputs.get('is_crowd', None) + # rois, tgt_labels, tgt_bboxes, tgt_gt_inds + # new_rois_num + outs = generate_proposal_target( + rpn_rois, gt_classes, gt_boxes, self.batch_size_per_im, + self.fg_fraction, self.fg_thresh, self.bg_thresh, self.num_classes, + self.ignore_thresh, is_crowd, self.use_random, is_cascade, + self.cascade_iou[stage], self.assign_on_cpu) + rois = outs[0] + rois_num = outs[-1] + # tgt_labels, tgt_bboxes, tgt_gt_inds + targets = outs[1:4] + return rois, rois_num, targets + + +@register +class BBoxLibraAssigner(object): + __shared__ = ['num_classes'] + """ + Libra-RCNN targets assignment module + + The assignment consists of three steps: + 1. Match RoIs and ground-truth box, label the RoIs with foreground + or background sample + 2. Sample anchors to keep the properly ratio between foreground and + background + 3. Generate the targets for classification and regression branch + + Args: + batch_size_per_im (int): Total number of RoIs per image. + default 512 + fg_fraction (float): Fraction of RoIs that is labeled + foreground, default 0.25 + fg_thresh (float): Minimum overlap required between a RoI + and ground-truth box for the (roi, gt box) pair to be + a foreground sample. default 0.5 + bg_thresh (float): Maximum overlap allowed between a RoI + and ground-truth box for the (roi, gt box) pair to be + a background sample. default 0.5 + use_random (bool): Use random sampling to choose foreground and + background boxes, default true + cascade_iou (list[iou]): The list of overlap to select foreground and + background of each stage, which is only used In Cascade RCNN. + num_classes (int): The number of class. + num_bins (int): The number of libra_sample. + """ + + def __init__(self, + batch_size_per_im=512, + fg_fraction=.25, + fg_thresh=.5, + bg_thresh=.5, + use_random=True, + cascade_iou=[0.5, 0.6, 0.7], + num_classes=80, + num_bins=3): + super(BBoxLibraAssigner, self).__init__() + self.batch_size_per_im = batch_size_per_im + self.fg_fraction = fg_fraction + self.fg_thresh = fg_thresh + self.bg_thresh = bg_thresh + self.use_random = use_random + self.cascade_iou = cascade_iou + self.num_classes = num_classes + self.num_bins = num_bins + + def __call__(self, + rpn_rois, + rpn_rois_num, + inputs, + stage=0, + is_cascade=False): + gt_classes = inputs['gt_class'] + gt_boxes = inputs['gt_bbox'] + # rois, tgt_labels, tgt_bboxes, tgt_gt_inds + outs = libra_generate_proposal_target( + rpn_rois, gt_classes, gt_boxes, self.batch_size_per_im, + self.fg_fraction, self.fg_thresh, self.bg_thresh, self.num_classes, + self.use_random, is_cascade, self.cascade_iou[stage], self.num_bins) + rois = outs[0] + rois_num = outs[-1] + # tgt_labels, tgt_bboxes, tgt_gt_inds + targets = outs[1:4] + return rois, rois_num, targets + + +@register +@serializable +class MaskAssigner(object): + __shared__ = ['num_classes', 'mask_resolution'] + """ + Mask targets assignment module + + The assignment consists of three steps: + 1. Select RoIs labels with foreground. + 2. Encode the RoIs and corresponding gt polygons to generate + mask target + + Args: + num_classes (int): The number of class + mask_resolution (int): The resolution of mask target, default 14 + """ + + def __init__(self, num_classes=80, mask_resolution=14): + super(MaskAssigner, self).__init__() + self.num_classes = num_classes + self.mask_resolution = mask_resolution + + def __call__(self, rois, tgt_labels, tgt_gt_inds, inputs): + gt_segms = inputs['gt_poly'] + + outs = generate_mask_target(gt_segms, rois, tgt_labels, tgt_gt_inds, + self.num_classes, self.mask_resolution) + + # mask_rois, mask_rois_num, tgt_classes, tgt_masks, mask_index, tgt_weights + return outs + + +@register +class RBoxAssigner(object): + """ + assigner of rbox + Args: + pos_iou_thr (float): threshold of pos samples + neg_iou_thr (float): threshold of neg samples + min_iou_thr (float): the min threshold of samples + ignore_iof_thr (int): the ignored threshold + """ + + def __init__(self, + pos_iou_thr=0.5, + neg_iou_thr=0.4, + min_iou_thr=0.0, + ignore_iof_thr=-2): + super(RBoxAssigner, self).__init__() + + self.pos_iou_thr = pos_iou_thr + self.neg_iou_thr = neg_iou_thr + self.min_iou_thr = min_iou_thr + self.ignore_iof_thr = ignore_iof_thr + + def anchor_valid(self, anchors): + """ + + Args: + anchor: M x 4 + + Returns: + + """ + if anchors.ndim == 3: + anchors = anchors.reshape(-1, anchors.shape[-1]) + assert anchors.ndim == 2 + anchor_num = anchors.shape[0] + anchor_valid = np.ones((anchor_num), np.int32) + anchor_inds = np.arange(anchor_num) + return anchor_inds + + def rbox2delta(self, + proposals, + gt, + means=[0, 0, 0, 0, 0], + stds=[1, 1, 1, 1, 1]): + """ + Args: + proposals: tensor [N, 5] + gt: gt [N, 5] + means: means [5] + stds: stds [5] + Returns: + + """ + proposals = proposals.astype(np.float64) + + PI = np.pi + + gt_widths = gt[..., 2] + gt_heights = gt[..., 3] + gt_angle = gt[..., 4] + + proposals_widths = proposals[..., 2] + proposals_heights = proposals[..., 3] + proposals_angle = proposals[..., 4] + + coord = gt[..., 0:2] - proposals[..., 0:2] + dx = (np.cos(proposals[..., 4]) * coord[..., 0] + + np.sin(proposals[..., 4]) * coord[..., 1]) / proposals_widths + dy = (-np.sin(proposals[..., 4]) * coord[..., 0] + + np.cos(proposals[..., 4]) * coord[..., 1]) / proposals_heights + dw = np.log(gt_widths / proposals_widths) + dh = np.log(gt_heights / proposals_heights) + da = (gt_angle - proposals_angle) + + da = (da + PI / 4) % PI - PI / 4 + da /= PI + + deltas = np.stack([dx, dy, dw, dh, da], axis=-1) + means = np.array(means, dtype=deltas.dtype) + stds = np.array(stds, dtype=deltas.dtype) + deltas = (deltas - means) / stds + deltas = deltas.astype(np.float32) + return deltas + + def assign_anchor(self, + anchors, + gt_bboxes, + gt_lables, + pos_iou_thr, + neg_iou_thr, + min_iou_thr=0.0, + ignore_iof_thr=-2): + """ + + Args: + anchors: + gt_bboxes:[M, 5] rc,yc,w,h,angle + gt_lables: + + Returns: + + """ + assert anchors.shape[1] == 4 or anchors.shape[1] == 5 + assert gt_bboxes.shape[1] == 4 or gt_bboxes.shape[1] == 5 + anchors_xc_yc = anchors + gt_bboxes_xc_yc = gt_bboxes + + # calc rbox iou + anchors_xc_yc = anchors_xc_yc.astype(np.float32) + gt_bboxes_xc_yc = gt_bboxes_xc_yc.astype(np.float32) + anchors_xc_yc = paddle.to_tensor(anchors_xc_yc) + gt_bboxes_xc_yc = paddle.to_tensor(gt_bboxes_xc_yc) + + try: + from rbox_iou_ops import rbox_iou + except Exception as e: + print("import custom_ops error, try install rbox_iou_ops " \ + "following ppdet/ext_op/README.md", e) + sys.stdout.flush() + sys.exit(-1) + + iou = rbox_iou(gt_bboxes_xc_yc, anchors_xc_yc) + iou = iou.numpy() + iou = iou.T + + # every gt's anchor's index + gt_bbox_anchor_inds = iou.argmax(axis=0) + gt_bbox_anchor_iou = iou[gt_bbox_anchor_inds, np.arange(iou.shape[1])] + gt_bbox_anchor_iou_inds = np.where(iou == gt_bbox_anchor_iou)[0] + + # every anchor's gt bbox's index + anchor_gt_bbox_inds = iou.argmax(axis=1) + anchor_gt_bbox_iou = iou[np.arange(iou.shape[0]), anchor_gt_bbox_inds] + + # (1) set labels=-2 as default + labels = np.ones((iou.shape[0], ), dtype=np.int32) * ignore_iof_thr + + # (2) assign ignore + labels[anchor_gt_bbox_iou < min_iou_thr] = ignore_iof_thr + + # (3) assign neg_ids -1 + assign_neg_ids1 = anchor_gt_bbox_iou >= min_iou_thr + assign_neg_ids2 = anchor_gt_bbox_iou < neg_iou_thr + assign_neg_ids = np.logical_and(assign_neg_ids1, assign_neg_ids2) + labels[assign_neg_ids] = -1 + + # anchor_gt_bbox_iou_inds + # (4) assign max_iou as pos_ids >=0 + anchor_gt_bbox_iou_inds = anchor_gt_bbox_inds[gt_bbox_anchor_iou_inds] + # gt_bbox_anchor_iou_inds = np.logical_and(gt_bbox_anchor_iou_inds, anchor_gt_bbox_iou >= min_iou_thr) + labels[gt_bbox_anchor_iou_inds] = gt_lables[anchor_gt_bbox_iou_inds] + + # (5) assign >= pos_iou_thr as pos_ids + iou_pos_iou_thr_ids = anchor_gt_bbox_iou >= pos_iou_thr + iou_pos_iou_thr_ids_box_inds = anchor_gt_bbox_inds[iou_pos_iou_thr_ids] + labels[iou_pos_iou_thr_ids] = gt_lables[iou_pos_iou_thr_ids_box_inds] + return anchor_gt_bbox_inds, anchor_gt_bbox_iou, labels + + def __call__(self, anchors, gt_bboxes, gt_labels, is_crowd): + + assert anchors.ndim == 2 + assert anchors.shape[1] == 5 + assert gt_bboxes.ndim == 2 + assert gt_bboxes.shape[1] == 5 + + pos_iou_thr = self.pos_iou_thr + neg_iou_thr = self.neg_iou_thr + min_iou_thr = self.min_iou_thr + ignore_iof_thr = self.ignore_iof_thr + + anchor_num = anchors.shape[0] + + gt_bboxes = gt_bboxes + is_crowd_slice = is_crowd + not_crowd_inds = np.where(is_crowd_slice == 0) + + # Step1: match anchor and gt_bbox + anchor_gt_bbox_inds, anchor_gt_bbox_iou, labels = self.assign_anchor( + anchors, gt_bboxes, + gt_labels.reshape(-1), pos_iou_thr, neg_iou_thr, min_iou_thr, + ignore_iof_thr) + + # Step2: sample anchor + pos_inds = np.where(labels >= 0)[0] + neg_inds = np.where(labels == -1)[0] + + # Step3: make output + anchors_num = anchors.shape[0] + bbox_targets = np.zeros_like(anchors) + bbox_weights = np.zeros_like(anchors) + bbox_gt_bboxes = np.zeros_like(anchors) + pos_labels = np.zeros(anchors_num, dtype=np.int32) + pos_labels_weights = np.zeros(anchors_num, dtype=np.float32) + + pos_sampled_anchors = anchors[pos_inds] + pos_sampled_gt_boxes = gt_bboxes[anchor_gt_bbox_inds[pos_inds]] + if len(pos_inds) > 0: + pos_bbox_targets = self.rbox2delta(pos_sampled_anchors, + pos_sampled_gt_boxes) + bbox_targets[pos_inds, :] = pos_bbox_targets + bbox_gt_bboxes[pos_inds, :] = pos_sampled_gt_boxes + bbox_weights[pos_inds, :] = 1.0 + + pos_labels[pos_inds] = labels[pos_inds] + pos_labels_weights[pos_inds] = 1.0 + + if len(neg_inds) > 0: + pos_labels_weights[neg_inds] = 1.0 + return (pos_labels, pos_labels_weights, bbox_targets, bbox_weights, + bbox_gt_bboxes, pos_inds, neg_inds) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__init__.py new file mode 100644 index 000000000..3b461325f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__init__.py @@ -0,0 +1,25 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import jde_embedding_head +from . import fairmot_embedding_head +from . import resnet +from . import pyramidal_embedding +from . import pplcnet_embedding + +from .fairmot_embedding_head import * +from .jde_embedding_head import * +from .resnet import * +from .pyramidal_embedding import * +from .pplcnet_embedding import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..c892d8565 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/fairmot_embedding_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/fairmot_embedding_head.cpython-37.pyc new file mode 100644 index 000000000..67188be62 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/fairmot_embedding_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/jde_embedding_head.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/jde_embedding_head.cpython-37.pyc new file mode 100644 index 000000000..997d37bc1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/jde_embedding_head.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pplcnet_embedding.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pplcnet_embedding.cpython-37.pyc new file mode 100644 index 000000000..52c1121b2 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pplcnet_embedding.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pyramidal_embedding.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pyramidal_embedding.cpython-37.pyc new file mode 100644 index 000000000..24c90d066 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/pyramidal_embedding.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/resnet.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/resnet.cpython-37.pyc new file mode 100644 index 000000000..43525e482 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/__pycache__/resnet.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/fairmot_embedding_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/fairmot_embedding_head.py new file mode 100644 index 000000000..98ca257fd --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/fairmot_embedding_head.py @@ -0,0 +1,224 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import numpy as np +import math +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import KaimingUniform, Uniform +from ppdet.core.workspace import register +from ppdet.modeling.heads.centernet_head import ConvLayer + +__all__ = ['FairMOTEmbeddingHead'] + + +@register +class FairMOTEmbeddingHead(nn.Layer): + __shared__ = ['num_classes'] + """ + Args: + in_channels (int): the channel number of input to FairMOTEmbeddingHead. + ch_head (int): the channel of features before fed into embedding, 256 by default. + ch_emb (int): the channel of the embedding feature, 128 by default. + num_identities_dict (dict): the number of identities of each category, + support single class and multi-calss, {0: 14455} as default. + """ + + def __init__(self, + in_channels, + ch_head=256, + ch_emb=128, + num_classes=1, + num_identities_dict={0: 14455}): + super(FairMOTEmbeddingHead, self).__init__() + assert num_classes >= 1 + self.num_classes = num_classes + self.ch_emb = ch_emb + self.num_identities_dict = num_identities_dict + self.reid = nn.Sequential( + ConvLayer( + in_channels, ch_head, kernel_size=3, padding=1, bias=True), + nn.ReLU(), + ConvLayer( + ch_head, ch_emb, kernel_size=1, stride=1, padding=0, bias=True)) + param_attr = paddle.ParamAttr(initializer=KaimingUniform()) + bound = 1 / math.sqrt(ch_emb) + bias_attr = paddle.ParamAttr(initializer=Uniform(-bound, bound)) + self.reid_loss = nn.CrossEntropyLoss(ignore_index=-1, reduction='sum') + + if num_classes == 1: + nID = self.num_identities_dict[0] # single class + self.classifier = nn.Linear( + ch_emb, nID, weight_attr=param_attr, bias_attr=bias_attr) + # When num_identities(nID) is 1, emb_scale is set as 1 + self.emb_scale = math.sqrt(2) * math.log(nID - 1) if nID > 1 else 1 + else: + self.classifiers = dict() + self.emb_scale_dict = dict() + for cls_id, nID in self.num_identities_dict.items(): + self.classifiers[str(cls_id)] = nn.Linear( + ch_emb, nID, weight_attr=param_attr, bias_attr=bias_attr) + # When num_identities(nID) is 1, emb_scale is set as 1 + self.emb_scale_dict[str(cls_id)] = math.sqrt(2) * math.log( + nID - 1) if nID > 1 else 1 + + @classmethod + def from_config(cls, cfg, input_shape): + if isinstance(input_shape, (list, tuple)): + input_shape = input_shape[0] + return {'in_channels': input_shape.channels} + + def process_by_class(self, bboxes, embedding, bbox_inds, topk_clses): + pred_dets, pred_embs = [], [] + for cls_id in range(self.num_classes): + inds_masks = topk_clses == cls_id + inds_masks = paddle.cast(inds_masks, 'float32') + + pos_num = inds_masks.sum().numpy() + if pos_num == 0: + continue + + cls_inds_mask = inds_masks > 0 + + bbox_mask = paddle.nonzero(cls_inds_mask) + cls_bboxes = paddle.gather_nd(bboxes, bbox_mask) + pred_dets.append(cls_bboxes) + + cls_inds = paddle.masked_select(bbox_inds, cls_inds_mask) + cls_inds = cls_inds.unsqueeze(-1) + cls_embedding = paddle.gather_nd(embedding, cls_inds) + pred_embs.append(cls_embedding) + + return paddle.concat(pred_dets), paddle.concat(pred_embs) + + def forward(self, + neck_feat, + inputs, + bboxes=None, + bbox_inds=None, + topk_clses=None): + reid_feat = self.reid(neck_feat) + if self.training: + if self.num_classes == 1: + loss = self.get_loss(reid_feat, inputs) + else: + loss = self.get_mc_loss(reid_feat, inputs) + return loss + else: + assert bboxes is not None and bbox_inds is not None + reid_feat = F.normalize(reid_feat) + embedding = paddle.transpose(reid_feat, [0, 2, 3, 1]) + embedding = paddle.reshape(embedding, [-1, self.ch_emb]) + # embedding shape: [bs * h * w, ch_emb] + + if self.num_classes == 1: + pred_dets = bboxes + pred_embs = paddle.gather(embedding, bbox_inds) + else: + pred_dets, pred_embs = self.process_by_class( + bboxes, embedding, bbox_inds, topk_clses) + return pred_dets, pred_embs + + def get_loss(self, feat, inputs): + index = inputs['index'] + mask = inputs['index_mask'] + target = inputs['reid'] + target = paddle.masked_select(target, mask > 0) + target = paddle.unsqueeze(target, 1) + + feat = paddle.transpose(feat, perm=[0, 2, 3, 1]) + feat_n, feat_h, feat_w, feat_c = feat.shape + feat = paddle.reshape(feat, shape=[feat_n, -1, feat_c]) + index = paddle.unsqueeze(index, 2) + batch_inds = list() + for i in range(feat_n): + batch_ind = paddle.full( + shape=[1, index.shape[1], 1], fill_value=i, dtype='int64') + batch_inds.append(batch_ind) + batch_inds = paddle.concat(batch_inds, axis=0) + index = paddle.concat(x=[batch_inds, index], axis=2) + feat = paddle.gather_nd(feat, index=index) + + mask = paddle.unsqueeze(mask, axis=2) + mask = paddle.expand_as(mask, feat) + mask.stop_gradient = True + feat = paddle.masked_select(feat, mask > 0) + feat = paddle.reshape(feat, shape=[-1, feat_c]) + feat = F.normalize(feat) + feat = self.emb_scale * feat + logit = self.classifier(feat) + target.stop_gradient = True + loss = self.reid_loss(logit, target) + valid = (target != self.reid_loss.ignore_index) + valid.stop_gradient = True + count = paddle.sum((paddle.cast(valid, dtype=np.int32))) + count.stop_gradient = True + if count > 0: + loss = loss / count + + return loss + + def get_mc_loss(self, feat, inputs): + # feat.shape = [bs, ch_emb, h, w] + assert 'cls_id_map' in inputs and 'cls_tr_ids' in inputs + index = inputs['index'] + mask = inputs['index_mask'] + cls_id_map = inputs['cls_id_map'] # [bs, h, w] + cls_tr_ids = inputs['cls_tr_ids'] # [bs, num_classes, h, w] + + feat = paddle.transpose(feat, perm=[0, 2, 3, 1]) + feat_n, feat_h, feat_w, feat_c = feat.shape + feat = paddle.reshape(feat, shape=[feat_n, -1, feat_c]) + + index = paddle.unsqueeze(index, 2) + batch_inds = list() + for i in range(feat_n): + batch_ind = paddle.full( + shape=[1, index.shape[1], 1], fill_value=i, dtype='int64') + batch_inds.append(batch_ind) + batch_inds = paddle.concat(batch_inds, axis=0) + index = paddle.concat(x=[batch_inds, index], axis=2) + feat = paddle.gather_nd(feat, index=index) + + mask = paddle.unsqueeze(mask, axis=2) + mask = paddle.expand_as(mask, feat) + mask.stop_gradient = True + feat = paddle.masked_select(feat, mask > 0) + feat = paddle.reshape(feat, shape=[-1, feat_c]) + + reid_losses = 0 + for cls_id, id_num in self.num_identities_dict.items(): + # target + cur_cls_tr_ids = paddle.reshape( + cls_tr_ids[:, cls_id, :, :], shape=[feat_n, -1]) # [bs, h*w] + cls_id_target = paddle.gather_nd(cur_cls_tr_ids, index=index) + mask = inputs['index_mask'] + cls_id_target = paddle.masked_select(cls_id_target, mask > 0) + cls_id_target.stop_gradient = True + + # feat + cls_id_feat = self.emb_scale_dict[str(cls_id)] * F.normalize(feat) + cls_id_pred = self.classifiers[str(cls_id)](cls_id_feat) + + loss = self.reid_loss(cls_id_pred, cls_id_target) + valid = (cls_id_target != self.reid_loss.ignore_index) + valid.stop_gradient = True + count = paddle.sum((paddle.cast(valid, dtype=np.int32))) + count.stop_gradient = True + if count > 0: + loss = loss / count + reid_losses += loss + + return reid_losses diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/jde_embedding_head.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/jde_embedding_head.py new file mode 100644 index 000000000..c35f8cfb0 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/jde_embedding_head.py @@ -0,0 +1,212 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import numpy as np +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr +from paddle.regularizer import L2Decay +from ppdet.core.workspace import register +from paddle.nn.initializer import Normal, Constant + +__all__ = ['JDEEmbeddingHead'] + + +class LossParam(nn.Layer): + def __init__(self, init_value=0., use_uncertainy=True): + super(LossParam, self).__init__() + self.loss_param = self.create_parameter( + shape=[1], + attr=ParamAttr(initializer=Constant(value=init_value)), + dtype="float32") + + def forward(self, inputs): + out = paddle.exp(-self.loss_param) * inputs + self.loss_param + return out * 0.5 + + +@register +class JDEEmbeddingHead(nn.Layer): + __shared__ = ['num_classes'] + __inject__ = ['emb_loss', 'jde_loss'] + """ + JDEEmbeddingHead + Args: + num_classes(int): Number of classes. Only support one class tracking. + num_identities(int): Number of identities. + anchor_levels(int): Number of anchor levels, same as FPN levels. + anchor_scales(int): Number of anchor scales on each FPN level. + embedding_dim(int): Embedding dimension. Default: 512. + emb_loss(object): Instance of 'JDEEmbeddingLoss' + jde_loss(object): Instance of 'JDELoss' + """ + + def __init__( + self, + num_classes=1, + num_identities=14455, # dataset.num_identities_dict[0] + anchor_levels=3, + anchor_scales=4, + embedding_dim=512, + emb_loss='JDEEmbeddingLoss', + jde_loss='JDELoss'): + super(JDEEmbeddingHead, self).__init__() + self.num_classes = num_classes + self.num_identities = num_identities + self.anchor_levels = anchor_levels + self.anchor_scales = anchor_scales + self.embedding_dim = embedding_dim + self.emb_loss = emb_loss + self.jde_loss = jde_loss + + self.emb_scale = math.sqrt(2) * math.log( + self.num_identities - 1) if self.num_identities > 1 else 1 + + self.identify_outputs = [] + self.loss_params_cls = [] + self.loss_params_reg = [] + self.loss_params_ide = [] + for i in range(self.anchor_levels): + name = 'identify_output.{}'.format(i) + identify_output = self.add_sublayer( + name, + nn.Conv2D( + in_channels=64 * (2**self.anchor_levels) // (2**i), + out_channels=self.embedding_dim, + kernel_size=3, + stride=1, + padding=1, + bias_attr=ParamAttr(regularizer=L2Decay(0.)))) + self.identify_outputs.append(identify_output) + + loss_p_cls = self.add_sublayer('cls.{}'.format(i), LossParam(-4.15)) + self.loss_params_cls.append(loss_p_cls) + loss_p_reg = self.add_sublayer('reg.{}'.format(i), LossParam(-4.85)) + self.loss_params_reg.append(loss_p_reg) + loss_p_ide = self.add_sublayer('ide.{}'.format(i), LossParam(-2.3)) + self.loss_params_ide.append(loss_p_ide) + + self.classifier = self.add_sublayer( + 'classifier', + nn.Linear( + self.embedding_dim, + self.num_identities, + weight_attr=ParamAttr( + learning_rate=1., initializer=Normal( + mean=0.0, std=0.01)), + bias_attr=ParamAttr( + learning_rate=2., regularizer=L2Decay(0.)))) + + def forward(self, + identify_feats, + targets, + loss_confs=None, + loss_boxes=None, + bboxes=None, + boxes_idx=None, + nms_keep_idx=None): + assert self.num_classes == 1, 'JDE only support sindle class MOT.' + assert len(identify_feats) == self.anchor_levels + ide_outs = [] + for feat, ide_head in zip(identify_feats, self.identify_outputs): + ide_outs.append(ide_head(feat)) + + if self.training: + assert len(loss_confs) == len(loss_boxes) == self.anchor_levels + loss_ides = self.emb_loss(ide_outs, targets, self.emb_scale, + self.classifier) + jde_losses = self.jde_loss( + loss_confs, loss_boxes, loss_ides, self.loss_params_cls, + self.loss_params_reg, self.loss_params_ide, targets) + return jde_losses + else: + assert bboxes is not None + assert boxes_idx is not None + assert nms_keep_idx is not None + + emb_outs = self.get_emb_outs(ide_outs) + emb_valid = paddle.gather_nd(emb_outs, boxes_idx) + pred_embs = paddle.gather_nd(emb_valid, nms_keep_idx) + + input_shape = targets['image'].shape[2:] + # input_shape: [h, w], before data transforms, set in model config + im_shape = targets['im_shape'][0].numpy() + # im_shape: [new_h, new_w], after data transforms + scale_factor = targets['scale_factor'][0].numpy() + bboxes[:, 2:] = self.scale_coords(bboxes[:, 2:], input_shape, + im_shape, scale_factor) + # tlwhs, scores, cls_ids + pred_dets = paddle.concat( + (bboxes[:, 2:], bboxes[:, 1:2], bboxes[:, 0:1]), axis=1) + return pred_dets, pred_embs + + def scale_coords(self, coords, input_shape, im_shape, scale_factor): + ratio = scale_factor[0] + pad_w = (input_shape[1] - int(im_shape[1])) / 2 + pad_h = (input_shape[0] - int(im_shape[0])) / 2 + coords = paddle.cast(coords, 'float32') + coords[:, 0::2] -= pad_w + coords[:, 1::2] -= pad_h + coords[:, 0:4] /= ratio + coords[:, :4] = paddle.clip( + coords[:, :4], min=0, max=coords[:, :4].max()) + return coords.round() + + def get_emb_and_gt_outs(self, ide_outs, targets): + emb_and_gts = [] + for i, p_ide in enumerate(ide_outs): + t_conf = targets['tconf{}'.format(i)] + t_ide = targets['tide{}'.format(i)] + + p_ide = p_ide.transpose((0, 2, 3, 1)) + p_ide_flatten = paddle.reshape(p_ide, [-1, self.embedding_dim]) + + mask = t_conf > 0 + mask = paddle.cast(mask, dtype="int64") + emb_mask = mask.max(1).flatten() + emb_mask_inds = paddle.nonzero(emb_mask > 0).flatten() + if len(emb_mask_inds) > 0: + t_ide_flatten = paddle.reshape(t_ide.max(1), [-1, 1]) + tids = paddle.gather(t_ide_flatten, emb_mask_inds) + + embedding = paddle.gather(p_ide_flatten, emb_mask_inds) + embedding = self.emb_scale * F.normalize(embedding) + emb_and_gt = paddle.concat([embedding, tids], axis=1) + emb_and_gts.append(emb_and_gt) + + if len(emb_and_gts) > 0: + return paddle.concat(emb_and_gts, axis=0) + else: + return paddle.zeros((1, self.embedding_dim + 1)) + + def get_emb_outs(self, ide_outs): + emb_outs = [] + for i, p_ide in enumerate(ide_outs): + p_ide = p_ide.transpose((0, 2, 3, 1)) + + p_ide_repeat = paddle.tile(p_ide, [self.anchor_scales, 1, 1, 1]) + embedding = F.normalize(p_ide_repeat, axis=-1) + emb = paddle.reshape(embedding, [-1, self.embedding_dim]) + emb_outs.append(emb) + + if len(emb_outs) > 0: + return paddle.concat(emb_outs, axis=0) + else: + return paddle.zeros((1, self.embedding_dim)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pplcnet_embedding.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pplcnet_embedding.py new file mode 100644 index 000000000..cad9f85be --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pplcnet_embedding.py @@ -0,0 +1,281 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, Constant +from paddle import ParamAttr +from paddle.nn import AdaptiveAvgPool2D, BatchNorm, Conv2D, Linear +from paddle.regularizer import L2Decay +from paddle.nn.initializer import KaimingNormal, XavierNormal +from ppdet.core.workspace import register + +__all__ = ['PPLCNetEmbedding'] + + +# Each element(list) represents a depthwise block, which is composed of k, in_c, out_c, s, use_se. +# k: kernel_size +# in_c: input channel number in depthwise block +# out_c: output channel number in depthwise block +# s: stride in depthwise block +# use_se: whether to use SE block + +NET_CONFIG = { + "blocks2": + #k, in_c, out_c, s, use_se + [[3, 16, 32, 1, False]], + "blocks3": [[3, 32, 64, 2, False], [3, 64, 64, 1, False]], + "blocks4": [[3, 64, 128, 2, False], [3, 128, 128, 1, False]], + "blocks5": [[3, 128, 256, 2, False], [5, 256, 256, 1, False], + [5, 256, 256, 1, False], [5, 256, 256, 1, False], + [5, 256, 256, 1, False], [5, 256, 256, 1, False]], + "blocks6": [[5, 256, 512, 2, True], [5, 512, 512, 1, True]] +} + + +def make_divisible(v, divisor=8, min_value=None): + if min_value is None: + min_value = divisor + new_v = max(min_value, int(v + divisor / 2) // divisor * divisor) + if new_v < 0.9 * v: + new_v += divisor + return new_v + + +class ConvBNLayer(nn.Layer): + def __init__(self, + num_channels, + filter_size, + num_filters, + stride, + num_groups=1): + super().__init__() + + self.conv = Conv2D( + in_channels=num_channels, + out_channels=num_filters, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + groups=num_groups, + weight_attr=ParamAttr(initializer=KaimingNormal()), + bias_attr=False) + + self.bn = BatchNorm( + num_filters, + param_attr=ParamAttr(regularizer=L2Decay(0.0)), + bias_attr=ParamAttr(regularizer=L2Decay(0.0))) + self.hardswish = nn.Hardswish() + + def forward(self, x): + x = self.conv(x) + x = self.bn(x) + x = self.hardswish(x) + return x + + +class DepthwiseSeparable(nn.Layer): + def __init__(self, + num_channels, + num_filters, + stride, + dw_size=3, + use_se=False): + super().__init__() + self.use_se = use_se + self.dw_conv = ConvBNLayer( + num_channels=num_channels, + num_filters=num_channels, + filter_size=dw_size, + stride=stride, + num_groups=num_channels) + if use_se: + self.se = SEModule(num_channels) + self.pw_conv = ConvBNLayer( + num_channels=num_channels, + filter_size=1, + num_filters=num_filters, + stride=1) + + def forward(self, x): + x = self.dw_conv(x) + if self.use_se: + x = self.se(x) + x = self.pw_conv(x) + return x + + +class SEModule(nn.Layer): + def __init__(self, channel, reduction=4): + super().__init__() + self.avg_pool = AdaptiveAvgPool2D(1) + self.conv1 = Conv2D( + in_channels=channel, + out_channels=channel // reduction, + kernel_size=1, + stride=1, + padding=0) + self.relu = nn.ReLU() + self.conv2 = Conv2D( + in_channels=channel // reduction, + out_channels=channel, + kernel_size=1, + stride=1, + padding=0) + self.hardsigmoid = nn.Hardsigmoid() + + def forward(self, x): + identity = x + x = self.avg_pool(x) + x = self.conv1(x) + x = self.relu(x) + x = self.conv2(x) + x = self.hardsigmoid(x) + x = paddle.multiply(x=identity, y=x) + return x + + +class PPLCNet(nn.Layer): + """ + PP-LCNet, see https://arxiv.org/abs/2109.15099. + This code is different from PPLCNet in ppdet/modeling/backbones/lcnet.py + or in PaddleClas, because the output is the flatten feature of last_conv. + + Args: + scale (float): Scale ratio of channels. + class_expand (int): Number of channels of conv feature. + """ + + def __init__(self, scale=1.0, class_expand=1280): + super(PPLCNet, self).__init__() + self.scale = scale + self.class_expand = class_expand + + self.conv1 = ConvBNLayer( + num_channels=3, + filter_size=3, + num_filters=make_divisible(16 * scale), + stride=2) + + self.blocks2 = nn.Sequential(*[ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks2"]) + ]) + + self.blocks3 = nn.Sequential(*[ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks3"]) + ]) + + self.blocks4 = nn.Sequential(*[ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks4"]) + ]) + + self.blocks5 = nn.Sequential(*[ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks5"]) + ]) + + self.blocks6 = nn.Sequential(*[ + DepthwiseSeparable( + num_channels=make_divisible(in_c * scale), + num_filters=make_divisible(out_c * scale), + dw_size=k, + stride=s, + use_se=se) + for i, (k, in_c, out_c, s, se) in enumerate(NET_CONFIG["blocks6"]) + ]) + + self.avg_pool = AdaptiveAvgPool2D(1) + self.last_conv = Conv2D( + in_channels=make_divisible(NET_CONFIG["blocks6"][-1][2] * scale), + out_channels=self.class_expand, + kernel_size=1, + stride=1, + padding=0, + bias_attr=False) + self.hardswish = nn.Hardswish() + self.flatten = nn.Flatten(start_axis=1, stop_axis=-1) + + def forward(self, x): + x = self.conv1(x) + + x = self.blocks2(x) + x = self.blocks3(x) + x = self.blocks4(x) + x = self.blocks5(x) + x = self.blocks6(x) + + x = self.avg_pool(x) + x = self.last_conv(x) + x = self.hardswish(x) + x = self.flatten(x) + return x + + +class FC(nn.Layer): + def __init__(self, input_ch, output_ch): + super(FC, self).__init__() + weight_attr = ParamAttr(initializer=XavierNormal()) + self.fc = paddle.nn.Linear(input_ch, output_ch, weight_attr=weight_attr) + + def forward(self, x): + out = self.fc(x) + return out + + +@register +class PPLCNetEmbedding(nn.Layer): + """ + PPLCNet Embedding + + Args: + input_ch (int): Number of channels of input conv feature. + output_ch (int): Number of channels of output conv feature. + """ + def __init__(self, scale=2.5, input_ch=1280, output_ch=512): + super(PPLCNetEmbedding, self).__init__() + self.backbone = PPLCNet(scale=scale) + self.neck = FC(input_ch, output_ch) + + def forward(self, x): + feat = self.backbone(x) + feat_out = self.neck(feat) + return feat_out diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pyramidal_embedding.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pyramidal_embedding.py new file mode 100644 index 000000000..a90d4e1ef --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/pyramidal_embedding.py @@ -0,0 +1,144 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal, Constant +from paddle import ParamAttr +from .resnet import * +from ppdet.core.workspace import register + +__all__ = ['PCBPyramid'] + + +@register +class PCBPyramid(nn.Layer): + """ + PCB (Part-based Convolutional Baseline), see https://arxiv.org/abs/1711.09349, + Pyramidal Person Re-IDentification, see https://arxiv.org/abs/1810.12193 + + Args: + input_ch (int): Number of channels of the input feature. + num_stripes (int): Number of sub-parts. + used_levels (tuple): Whether the level is used, 1 means used. + num_classes (int): Number of classes for identities, default 751 in + Market-1501 dataset. + last_conv_stride (int): Stride of the last conv. + last_conv_dilation (int): Dilation of the last conv. + num_conv_out_channels (int): Number of channels of conv feature. + """ + + def __init__(self, + input_ch=2048, + num_stripes=6, + used_levels=(1, 1, 1, 1, 1, 1), + num_classes=751, + last_conv_stride=1, + last_conv_dilation=1, + num_conv_out_channels=128): + super(PCBPyramid, self).__init__() + self.num_stripes = num_stripes + self.used_levels = used_levels + self.num_classes = num_classes + + self.num_in_each_level = [i for i in range(self.num_stripes, 0, -1)] + self.num_branches = sum(self.num_in_each_level) + + self.base = ResNet101( + lr_mult=0.1, + last_conv_stride=last_conv_stride, + last_conv_dilation=last_conv_dilation) + self.dropout_layer = nn.Dropout(p=0.2) + self.pyramid_conv_list0, self.pyramid_fc_list0 = self.basic_branch( + num_conv_out_channels, input_ch) + + def basic_branch(self, num_conv_out_channels, input_ch): + # the level indexes are defined from fine to coarse, + # the branch will contain one more part than that of its previous level + # the sliding step is set to 1 + pyramid_conv_list = nn.LayerList() + pyramid_fc_list = nn.LayerList() + + idx_levels = 0 + for idx_branches in range(self.num_branches): + if idx_branches >= sum(self.num_in_each_level[0:idx_levels + 1]): + idx_levels += 1 + + pyramid_conv_list.append( + nn.Sequential( + nn.Conv2D(input_ch, num_conv_out_channels, 1), + nn.BatchNorm2D(num_conv_out_channels), nn.ReLU())) + + idx_levels = 0 + for idx_branches in range(self.num_branches): + if idx_branches >= sum(self.num_in_each_level[0:idx_levels + 1]): + idx_levels += 1 + + fc = nn.Linear( + in_features=num_conv_out_channels, + out_features=self.num_classes, + weight_attr=ParamAttr(initializer=Normal( + mean=0., std=0.001)), + bias_attr=ParamAttr(initializer=Constant(value=0.))) + pyramid_fc_list.append(fc) + return pyramid_conv_list, pyramid_fc_list + + def pyramid_forward(self, feat): + each_stripe_size = int(feat.shape[2] / self.num_stripes) + + feat_list, logits_list = [], [] + idx_levels = 0 + used_branches = 0 + for idx_branches in range(self.num_branches): + if idx_branches >= sum(self.num_in_each_level[0:idx_levels + 1]): + idx_levels += 1 + idx_in_each_level = idx_branches - sum(self.num_in_each_level[ + 0:idx_levels]) + stripe_size_in_each_level = each_stripe_size * (idx_levels + 1) + start = idx_in_each_level * each_stripe_size + end = start + stripe_size_in_each_level + + k = feat.shape[-1] + local_feat_avgpool = F.avg_pool2d( + feat[:, :, start:end, :], + kernel_size=(stripe_size_in_each_level, k)) + local_feat_maxpool = F.max_pool2d( + feat[:, :, start:end, :], + kernel_size=(stripe_size_in_each_level, k)) + local_feat = local_feat_avgpool + local_feat_maxpool + + local_feat = self.pyramid_conv_list0[used_branches](local_feat) + local_feat = paddle.reshape( + local_feat, shape=[local_feat.shape[0], -1]) + feat_list.append(local_feat) + + local_logits = self.pyramid_fc_list0[used_branches]( + self.dropout_layer(local_feat)) + logits_list.append(local_logits) + + used_branches += 1 + + return feat_list, logits_list + + def forward(self, x): + feat = self.base(x) + assert feat.shape[2] % self.num_stripes == 0 + feat_list, logits_list = self.pyramid_forward(feat) + feat_out = paddle.concat(feat_list, axis=-1) + return feat_out diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/resnet.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/resnet.py new file mode 100644 index 000000000..968fe9774 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/reid/resnet.py @@ -0,0 +1,310 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import math +import paddle +from paddle import ParamAttr +import paddle.nn as nn +import paddle.nn.functional as F +from paddle.nn.initializer import Normal + +__all__ = ["ResNet18", "ResNet34", "ResNet50", "ResNet101", "ResNet152"] + + +class ConvBNLayer(nn.Layer): + def __init__(self, + num_channels, + num_filters, + filter_size, + stride=1, + dilation=1, + groups=1, + act=None, + lr_mult=1.0, + name=None, + data_format="NCHW"): + super(ConvBNLayer, self).__init__() + conv_stdv = filter_size * filter_size * num_filters + self._conv = nn.Conv2D( + in_channels=num_channels, + out_channels=num_filters, + kernel_size=filter_size, + stride=stride, + padding=(filter_size - 1) // 2, + dilation=dilation, + groups=groups, + weight_attr=ParamAttr( + learning_rate=lr_mult, + initializer=Normal(0, math.sqrt(2. / conv_stdv))), + bias_attr=False, + data_format=data_format) + + self._batch_norm = nn.BatchNorm( + num_filters, act=act, data_layout=data_format) + + def forward(self, inputs): + y = self._conv(inputs) + y = self._batch_norm(y) + return y + + +class BottleneckBlock(nn.Layer): + def __init__(self, + num_channels, + num_filters, + stride, + shortcut=True, + name=None, + lr_mult=1.0, + dilation=1, + data_format="NCHW"): + super(BottleneckBlock, self).__init__() + self.conv0 = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters, + filter_size=1, + dilation=dilation, + act="relu", + lr_mult=lr_mult, + name=name + "_branch2a", + data_format=data_format) + self.conv1 = ConvBNLayer( + num_channels=num_filters, + num_filters=num_filters, + filter_size=3, + dilation=dilation, + stride=stride, + act="relu", + lr_mult=lr_mult, + name=name + "_branch2b", + data_format=data_format) + self.conv2 = ConvBNLayer( + num_channels=num_filters, + num_filters=num_filters * 4, + filter_size=1, + dilation=dilation, + act=None, + lr_mult=lr_mult, + name=name + "_branch2c", + data_format=data_format) + if not shortcut: + self.short = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters * 4, + filter_size=1, + dilation=dilation, + stride=stride, + lr_mult=lr_mult, + name=name + "_branch1", + data_format=data_format) + self.shortcut = shortcut + self._num_channels_out = num_filters * 4 + + def forward(self, inputs): + y = self.conv0(inputs) + conv1 = self.conv1(y) + conv2 = self.conv2(conv1) + if self.shortcut: + short = inputs + else: + short = self.short(inputs) + y = paddle.add(x=short, y=conv2) + y = F.relu(y) + return y + + +class BasicBlock(nn.Layer): + def __init__(self, + num_channels, + num_filters, + stride, + shortcut=True, + name=None, + data_format="NCHW"): + super(BasicBlock, self).__init__() + self.stride = stride + self.conv0 = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters, + filter_size=3, + stride=stride, + act="relu", + name=name + "_branch2a", + data_format=data_format) + self.conv1 = ConvBNLayer( + num_channels=num_filters, + num_filters=num_filters, + filter_size=3, + act=None, + name=name + "_branch2b", + data_format=data_format) + if not shortcut: + self.short = ConvBNLayer( + num_channels=num_channels, + num_filters=num_filters, + filter_size=1, + stride=stride, + name=name + "_branch1", + data_format=data_format) + self.shortcut = shortcut + + def forward(self, inputs): + y = self.conv0(inputs) + conv1 = self.conv1(y) + if self.shortcut: + short = inputs + else: + short = self.short(inputs) + y = paddle.add(x=short, y=conv1) + y = F.relu(y) + return y + + +class ResNet(nn.Layer): + def __init__(self, + layers=50, + lr_mult=1.0, + last_conv_stride=2, + last_conv_dilation=1): + super(ResNet, self).__init__() + self.layers = layers + self.data_format = "NCHW" + self.input_image_channel = 3 + supported_layers = [18, 34, 50, 101, 152] + assert layers in supported_layers, \ + "supported layers are {} but input layer is {}".format( + supported_layers, layers) + if layers == 18: + depth = [2, 2, 2, 2] + elif layers == 34 or layers == 50: + depth = [3, 4, 6, 3] + elif layers == 101: + depth = [3, 4, 23, 3] + elif layers == 152: + depth = [3, 8, 36, 3] + num_channels = [64, 256, 512, + 1024] if layers >= 50 else [64, 64, 128, 256] + num_filters = [64, 128, 256, 512] + self.conv = ConvBNLayer( + num_channels=self.input_image_channel, + num_filters=64, + filter_size=7, + stride=2, + act="relu", + lr_mult=lr_mult, + name="conv1", + data_format=self.data_format) + self.pool2d_max = nn.MaxPool2D( + kernel_size=3, stride=2, padding=1, data_format=self.data_format) + self.block_list = [] + if layers >= 50: + for block in range(len(depth)): + shortcut = False + for i in range(depth[block]): + if layers in [101, 152] and block == 2: + if i == 0: + conv_name = "res" + str(block + 2) + "a" + else: + conv_name = "res" + str(block + 2) + "b" + str(i) + else: + conv_name = "res" + str(block + 2) + chr(97 + i) + if i != 0 or block == 0: + stride = 1 + elif block == len(depth) - 1: + stride = last_conv_stride + else: + stride = 2 + bottleneck_block = self.add_sublayer( + conv_name, + BottleneckBlock( + num_channels=num_channels[block] + if i == 0 else num_filters[block] * 4, + num_filters=num_filters[block], + stride=stride, + shortcut=shortcut, + name=conv_name, + lr_mult=lr_mult, + dilation=last_conv_dilation + if block == len(depth) - 1 else 1, + data_format=self.data_format)) + self.block_list.append(bottleneck_block) + shortcut = True + else: + for block in range(len(depth)): + shortcut = False + for i in range(depth[block]): + conv_name = "res" + str(block + 2) + chr(97 + i) + basic_block = self.add_sublayer( + conv_name, + BasicBlock( + num_channels=num_channels[block] + if i == 0 else num_filters[block], + num_filters=num_filters[block], + stride=2 if i == 0 and block != 0 else 1, + shortcut=shortcut, + name=conv_name, + data_format=self.data_format)) + self.block_list.append(basic_block) + shortcut = True + + def forward(self, inputs): + y = self.conv(inputs) + y = self.pool2d_max(y) + for block in self.block_list: + y = block(y) + return y + + +def ResNet18(**args): + model = ResNet(layers=18, **args) + return model + + +def ResNet34(**args): + model = ResNet(layers=34, **args) + return model + + +def ResNet50(pretrained=None, **args): + model = ResNet(layers=50, **args) + if pretrained is not None: + if not (os.path.isdir(pretrained) or + os.path.exists(pretrained + '.pdparams')): + raise ValueError("Model pretrain path {} does not " + "exists.".format(pretrained)) + param_state_dict = paddle.load(pretrained + '.pdparams') + model.set_dict(param_state_dict) + return model + + +def ResNet101(pretrained=None, **args): + model = ResNet(layers=101, **args) + if pretrained is not None: + if not (os.path.isdir(pretrained) or + os.path.exists(pretrained + '.pdparams')): + raise ValueError("Model pretrain path {} does not " + "exists.".format(pretrained)) + param_state_dict = paddle.load(pretrained + '.pdparams') + model.set_dict(param_state_dict) + return model + + +def ResNet152(**args): + model = ResNet(layers=152, **args) + return model diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/shape_spec.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/shape_spec.py new file mode 100644 index 000000000..81601fd64 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/shape_spec.py @@ -0,0 +1,25 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +# The code is based on: +# https://github.com/facebookresearch/detectron2/blob/main/detectron2/layers/shape_spec.py + +from collections import namedtuple + + +class ShapeSpec( + namedtuple("_ShapeSpec", ["channels", "height", "width", "stride"])): + def __new__(cls, channels=None, height=None, width=None, stride=None): + return super(ShapeSpec, cls).__new__(cls, channels, height, width, + stride) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/__init__.py new file mode 100644 index 000000000..847ddc47a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/__init__.py @@ -0,0 +1,13 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_architectures.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_architectures.py new file mode 100644 index 000000000..25767e74a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_architectures.py @@ -0,0 +1,69 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import unittest +import ppdet + + +class TestFasterRCNN(unittest.TestCase): + def setUp(self): + self.set_config() + + def set_config(self): + self.cfg_file = 'configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.yml' + + def test_trainer(self): + # Trainer __init__ will build model and DataLoader + # 'train' and 'eval' mode include dataset loading + # use 'test' mode to simplify tests + cfg = ppdet.core.workspace.load_config(self.cfg_file) + trainer = ppdet.engine.Trainer(cfg, mode='test') + + +class TestMaskRCNN(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/mask_rcnn/mask_rcnn_r50_fpn_1x_coco.yml' + + +class TestCascadeRCNN(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/cascade_rcnn/cascade_rcnn_r50_fpn_1x_coco.yml' + + +class TestYolov3(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/yolov3/yolov3_darknet53_270e_coco.yml' + + +class TestSSD(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/ssd/ssd_vgg16_300_240e_voc.yml' + + +class TestGFL(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/gfl/gfl_r50_fpn_1x_coco.yml' + + +class TestPicoDet(TestFasterRCNN): + def set_config(self): + self.cfg_file = 'configs/picodet/picodet_s_320_coco.yml' + + +if __name__ == '__main__': + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_base.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_base.py new file mode 100644 index 000000000..cbb9033b3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_base.py @@ -0,0 +1,74 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import print_function +import unittest + +import contextlib + +import paddle +import paddle.fluid as fluid +from paddle.fluid.framework import Program +from paddle.fluid import core + + +class LayerTest(unittest.TestCase): + @classmethod + def setUpClass(cls): + cls.seed = 111 + + @classmethod + def tearDownClass(cls): + pass + + def _get_place(self, force_to_use_cpu=False): + # this option for ops that only have cpu kernel + if force_to_use_cpu: + return core.CPUPlace() + else: + if core.is_compiled_with_cuda(): + return core.CUDAPlace(0) + return core.CPUPlace() + + @contextlib.contextmanager + def static_graph(self): + paddle.enable_static() + scope = fluid.core.Scope() + program = Program() + with fluid.scope_guard(scope): + with fluid.program_guard(program): + paddle.seed(self.seed) + paddle.framework.random._manual_program_seed(self.seed) + yield + + def get_static_graph_result(self, + feed, + fetch_list, + with_lod=False, + force_to_use_cpu=False): + exe = fluid.Executor(self._get_place(force_to_use_cpu)) + exe.run(fluid.default_startup_program()) + return exe.run(fluid.default_main_program(), + feed=feed, + fetch_list=fetch_list, + return_numpy=(not with_lod)) + + @contextlib.contextmanager + def dynamic_graph(self, force_to_use_cpu=False): + paddle.disable_static() + with fluid.dygraph.guard( + self._get_place(force_to_use_cpu=force_to_use_cpu)): + paddle.seed(self.seed) + paddle.framework.random._manual_program_seed(self.seed) + yield diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_ops.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_ops.py new file mode 100644 index 000000000..d4b574748 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_ops.py @@ -0,0 +1,835 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import print_function +import os, sys +# add python path of PadleDetection to sys.path +parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4))) +if parent_path not in sys.path: + sys.path.append(parent_path) + +import unittest +import numpy as np + +import paddle +import paddle.fluid as fluid +from paddle.fluid.dygraph import base + +import ppdet.modeling.ops as ops +from ppdet.modeling.tests.test_base import LayerTest + + +def make_rois(h, w, rois_num, output_size): + rois = np.zeros((0, 4)).astype('float32') + for roi_num in rois_num: + roi = np.zeros((roi_num, 4)).astype('float32') + roi[:, 0] = np.random.randint(0, h - output_size[0], size=roi_num) + roi[:, 1] = np.random.randint(0, w - output_size[1], size=roi_num) + roi[:, 2] = np.random.randint(roi[:, 0] + output_size[0], h) + roi[:, 3] = np.random.randint(roi[:, 1] + output_size[1], w) + rois = np.vstack((rois, roi)) + return rois + + +def softmax(x): + # clip to shiftx, otherwise, when calc loss with + # log(exp(shiftx)), may get log(0)=INF + shiftx = (x - np.max(x)).clip(-64.) + exps = np.exp(shiftx) + return exps / np.sum(exps) + + +class TestCollectFpnProposals(LayerTest): + def test_collect_fpn_proposals(self): + multi_bboxes_np = [] + multi_scores_np = [] + rois_num_per_level_np = [] + for i in range(4): + bboxes_np = np.random.rand(5, 4).astype('float32') + scores_np = np.random.rand(5, 1).astype('float32') + rois_num = np.array([2, 3]).astype('int32') + multi_bboxes_np.append(bboxes_np) + multi_scores_np.append(scores_np) + rois_num_per_level_np.append(rois_num) + + with self.static_graph(): + multi_bboxes = [] + multi_scores = [] + rois_num_per_level = [] + for i in range(4): + bboxes = paddle.static.data( + name='rois' + str(i), + shape=[5, 4], + dtype='float32', + lod_level=1) + scores = paddle.static.data( + name='scores' + str(i), + shape=[5, 1], + dtype='float32', + lod_level=1) + rois_num = paddle.static.data( + name='rois_num' + str(i), shape=[None], dtype='int32') + + multi_bboxes.append(bboxes) + multi_scores.append(scores) + rois_num_per_level.append(rois_num) + + fpn_rois, rois_num = ops.collect_fpn_proposals( + multi_bboxes, + multi_scores, + 2, + 5, + 10, + rois_num_per_level=rois_num_per_level) + feed = {} + for i in range(4): + feed['rois' + str(i)] = multi_bboxes_np[i] + feed['scores' + str(i)] = multi_scores_np[i] + feed['rois_num' + str(i)] = rois_num_per_level_np[i] + fpn_rois_stat, rois_num_stat = self.get_static_graph_result( + feed=feed, fetch_list=[fpn_rois, rois_num], with_lod=True) + fpn_rois_stat = np.array(fpn_rois_stat) + rois_num_stat = np.array(rois_num_stat) + + with self.dynamic_graph(): + multi_bboxes_dy = [] + multi_scores_dy = [] + rois_num_per_level_dy = [] + for i in range(4): + bboxes_dy = base.to_variable(multi_bboxes_np[i]) + scores_dy = base.to_variable(multi_scores_np[i]) + rois_num_dy = base.to_variable(rois_num_per_level_np[i]) + multi_bboxes_dy.append(bboxes_dy) + multi_scores_dy.append(scores_dy) + rois_num_per_level_dy.append(rois_num_dy) + fpn_rois_dy, rois_num_dy = ops.collect_fpn_proposals( + multi_bboxes_dy, + multi_scores_dy, + 2, + 5, + 10, + rois_num_per_level=rois_num_per_level_dy) + fpn_rois_dy = fpn_rois_dy.numpy() + rois_num_dy = rois_num_dy.numpy() + + self.assertTrue(np.array_equal(fpn_rois_stat, fpn_rois_dy)) + self.assertTrue(np.array_equal(rois_num_stat, rois_num_dy)) + + def test_collect_fpn_proposals_error(self): + def generate_input(bbox_type, score_type, name): + multi_bboxes = [] + multi_scores = [] + for i in range(4): + bboxes = paddle.static.data( + name='rois' + name + str(i), + shape=[10, 4], + dtype=bbox_type, + lod_level=1) + scores = paddle.static.data( + name='scores' + name + str(i), + shape=[10, 1], + dtype=score_type, + lod_level=1) + multi_bboxes.append(bboxes) + multi_scores.append(scores) + return multi_bboxes, multi_scores + + with self.static_graph(): + bbox1 = paddle.static.data( + name='rois', shape=[5, 10, 4], dtype='float32', lod_level=1) + score1 = paddle.static.data( + name='scores', shape=[5, 10, 1], dtype='float32', lod_level=1) + bbox2, score2 = generate_input('int32', 'float32', '2') + self.assertRaises( + TypeError, + ops.collect_fpn_proposals, + multi_rois=bbox1, + multi_scores=score1, + min_level=2, + max_level=5, + post_nms_top_n=2000) + self.assertRaises( + TypeError, + ops.collect_fpn_proposals, + multi_rois=bbox2, + multi_scores=score2, + min_level=2, + max_level=5, + post_nms_top_n=2000) + + paddle.disable_static() + + +class TestDistributeFpnProposals(LayerTest): + def test_distribute_fpn_proposals(self): + rois_np = np.random.rand(10, 4).astype('float32') + rois_num_np = np.array([4, 6]).astype('int32') + with self.static_graph(): + rois = paddle.static.data( + name='rois', shape=[10, 4], dtype='float32') + rois_num = paddle.static.data( + name='rois_num', shape=[None], dtype='int32') + multi_rois, restore_ind, rois_num_per_level = ops.distribute_fpn_proposals( + fpn_rois=rois, + min_level=2, + max_level=5, + refer_level=4, + refer_scale=224, + rois_num=rois_num) + fetch_list = multi_rois + [restore_ind] + rois_num_per_level + output_stat = self.get_static_graph_result( + feed={'rois': rois_np, + 'rois_num': rois_num_np}, + fetch_list=fetch_list, + with_lod=True) + output_stat_np = [] + for output in output_stat: + output_np = np.array(output) + if len(output_np) > 0: + output_stat_np.append(output_np) + + with self.dynamic_graph(): + rois_dy = base.to_variable(rois_np) + rois_num_dy = base.to_variable(rois_num_np) + multi_rois_dy, restore_ind_dy, rois_num_per_level_dy = ops.distribute_fpn_proposals( + fpn_rois=rois_dy, + min_level=2, + max_level=5, + refer_level=4, + refer_scale=224, + rois_num=rois_num_dy) + output_dy = multi_rois_dy + [restore_ind_dy] + rois_num_per_level_dy + output_dy_np = [] + for output in output_dy: + output_np = output.numpy() + if len(output_np) > 0: + output_dy_np.append(output_np) + + for res_stat, res_dy in zip(output_stat_np, output_dy_np): + self.assertTrue(np.array_equal(res_stat, res_dy)) + + def test_distribute_fpn_proposals_error(self): + with self.static_graph(): + fpn_rois = paddle.static.data( + name='data_error', shape=[10, 4], dtype='int32', lod_level=1) + self.assertRaises( + TypeError, + ops.distribute_fpn_proposals, + fpn_rois=fpn_rois, + min_level=2, + max_level=5, + refer_level=4, + refer_scale=224) + + paddle.disable_static() + + +class TestROIAlign(LayerTest): + def test_roi_align(self): + b, c, h, w = 2, 12, 20, 20 + inputs_np = np.random.rand(b, c, h, w).astype('float32') + rois_num = [4, 6] + output_size = (7, 7) + rois_np = make_rois(h, w, rois_num, output_size) + rois_num_np = np.array(rois_num).astype('int32') + with self.static_graph(): + inputs = paddle.static.data( + name='inputs', shape=[b, c, h, w], dtype='float32') + rois = paddle.static.data( + name='rois', shape=[10, 4], dtype='float32') + rois_num = paddle.static.data( + name='rois_num', shape=[None], dtype='int32') + + output = ops.roi_align( + input=inputs, + rois=rois, + output_size=output_size, + rois_num=rois_num) + output_np, = self.get_static_graph_result( + feed={ + 'inputs': inputs_np, + 'rois': rois_np, + 'rois_num': rois_num_np + }, + fetch_list=output, + with_lod=False) + + with self.dynamic_graph(): + inputs_dy = base.to_variable(inputs_np) + rois_dy = base.to_variable(rois_np) + rois_num_dy = base.to_variable(rois_num_np) + + output_dy = ops.roi_align( + input=inputs_dy, + rois=rois_dy, + output_size=output_size, + rois_num=rois_num_dy) + output_dy_np = output_dy.numpy() + + self.assertTrue(np.array_equal(output_np, output_dy_np)) + + def test_roi_align_error(self): + with self.static_graph(): + inputs = paddle.static.data( + name='inputs', shape=[2, 12, 20, 20], dtype='float32') + rois = paddle.static.data( + name='data_error', shape=[10, 4], dtype='int32', lod_level=1) + self.assertRaises( + TypeError, + ops.roi_align, + input=inputs, + rois=rois, + output_size=(7, 7)) + + paddle.disable_static() + + +class TestROIPool(LayerTest): + def test_roi_pool(self): + b, c, h, w = 2, 12, 20, 20 + inputs_np = np.random.rand(b, c, h, w).astype('float32') + rois_num = [4, 6] + output_size = (7, 7) + rois_np = make_rois(h, w, rois_num, output_size) + rois_num_np = np.array(rois_num).astype('int32') + with self.static_graph(): + inputs = paddle.static.data( + name='inputs', shape=[b, c, h, w], dtype='float32') + rois = paddle.static.data( + name='rois', shape=[10, 4], dtype='float32') + rois_num = paddle.static.data( + name='rois_num', shape=[None], dtype='int32') + + output, _ = ops.roi_pool( + input=inputs, + rois=rois, + output_size=output_size, + rois_num=rois_num) + output_np, = self.get_static_graph_result( + feed={ + 'inputs': inputs_np, + 'rois': rois_np, + 'rois_num': rois_num_np + }, + fetch_list=[output], + with_lod=False) + + with self.dynamic_graph(): + inputs_dy = base.to_variable(inputs_np) + rois_dy = base.to_variable(rois_np) + rois_num_dy = base.to_variable(rois_num_np) + + output_dy, _ = ops.roi_pool( + input=inputs_dy, + rois=rois_dy, + output_size=output_size, + rois_num=rois_num_dy) + output_dy_np = output_dy.numpy() + + self.assertTrue(np.array_equal(output_np, output_dy_np)) + + def test_roi_pool_error(self): + with self.static_graph(): + inputs = paddle.static.data( + name='inputs', shape=[2, 12, 20, 20], dtype='float32') + rois = paddle.static.data( + name='data_error', shape=[10, 4], dtype='int32', lod_level=1) + self.assertRaises( + TypeError, + ops.roi_pool, + input=inputs, + rois=rois, + output_size=(7, 7)) + + paddle.disable_static() + + +class TestIoUSimilarity(LayerTest): + def test_iou_similarity(self): + b, c, h, w = 2, 12, 20, 20 + inputs_np = np.random.rand(b, c, h, w).astype('float32') + output_size = (7, 7) + x_np = make_rois(h, w, [20], output_size) + y_np = make_rois(h, w, [10], output_size) + with self.static_graph(): + x = paddle.static.data(name='x', shape=[20, 4], dtype='float32') + y = paddle.static.data(name='y', shape=[10, 4], dtype='float32') + + iou = ops.iou_similarity(x=x, y=y) + iou_np, = self.get_static_graph_result( + feed={ + 'x': x_np, + 'y': y_np, + }, fetch_list=[iou], with_lod=False) + + with self.dynamic_graph(): + x_dy = base.to_variable(x_np) + y_dy = base.to_variable(y_np) + + iou_dy = ops.iou_similarity(x=x_dy, y=y_dy) + iou_dy_np = iou_dy.numpy() + + self.assertTrue(np.array_equal(iou_np, iou_dy_np)) + + +class TestBipartiteMatch(LayerTest): + def test_bipartite_match(self): + distance = np.random.random((20, 10)).astype('float32') + with self.static_graph(): + x = paddle.static.data(name='x', shape=[20, 10], dtype='float32') + + match_indices, match_dist = ops.bipartite_match( + x, match_type='per_prediction', dist_threshold=0.5) + match_indices_np, match_dist_np = self.get_static_graph_result( + feed={'x': distance, }, + fetch_list=[match_indices, match_dist], + with_lod=False) + + with self.dynamic_graph(): + x_dy = base.to_variable(distance) + + match_indices_dy, match_dist_dy = ops.bipartite_match( + x_dy, match_type='per_prediction', dist_threshold=0.5) + match_indices_dy_np = match_indices_dy.numpy() + match_dist_dy_np = match_dist_dy.numpy() + + self.assertTrue(np.array_equal(match_indices_np, match_indices_dy_np)) + self.assertTrue(np.array_equal(match_dist_np, match_dist_dy_np)) + + +class TestYoloBox(LayerTest): + def test_yolo_box(self): + + # x shape [N C H W], C=K * (5 + class_num), class_num=10, K=2 + np_x = np.random.random([1, 30, 7, 7]).astype('float32') + np_origin_shape = np.array([[608, 608]], dtype='int32') + class_num = 10 + conf_thresh = 0.01 + downsample_ratio = 32 + scale_x_y = 1.2 + + # static + with self.static_graph(): + # x shape [N C H W], C=K * (5 + class_num), class_num=10, K=2 + x = paddle.static.data( + name='x', shape=[1, 30, 7, 7], dtype='float32') + origin_shape = paddle.static.data( + name='origin_shape', shape=[1, 2], dtype='int32') + + boxes, scores = ops.yolo_box( + x, + origin_shape, [10, 13, 30, 13], + class_num, + conf_thresh, + downsample_ratio, + scale_x_y=scale_x_y) + + boxes_np, scores_np = self.get_static_graph_result( + feed={ + 'x': np_x, + 'origin_shape': np_origin_shape, + }, + fetch_list=[boxes, scores], + with_lod=False) + + # dygraph + with self.dynamic_graph(): + x_dy = fluid.layers.assign(np_x) + origin_shape_dy = fluid.layers.assign(np_origin_shape) + + boxes_dy, scores_dy = ops.yolo_box( + x_dy, + origin_shape_dy, [10, 13, 30, 13], + 10, + 0.01, + 32, + scale_x_y=scale_x_y) + + boxes_dy_np = boxes_dy.numpy() + scores_dy_np = scores_dy.numpy() + + self.assertTrue(np.array_equal(boxes_np, boxes_dy_np)) + self.assertTrue(np.array_equal(scores_np, scores_dy_np)) + + def test_yolo_box_error(self): + with self.static_graph(): + # x shape [N C H W], C=K * (5 + class_num), class_num=10, K=2 + x = paddle.static.data( + name='x', shape=[1, 30, 7, 7], dtype='float32') + origin_shape = paddle.static.data( + name='origin_shape', shape=[1, 2], dtype='int32') + + self.assertRaises( + TypeError, + ops.yolo_box, + x, + origin_shape, [10, 13, 30, 13], + 10.123, + 0.01, + 32, + scale_x_y=1.2) + + paddle.disable_static() + + +class TestPriorBox(LayerTest): + def test_prior_box(self): + input_np = np.random.rand(2, 10, 32, 32).astype('float32') + image_np = np.random.rand(2, 10, 40, 40).astype('float32') + min_sizes = [2, 4] + with self.static_graph(): + input = paddle.static.data( + name='input', shape=[2, 10, 32, 32], dtype='float32') + image = paddle.static.data( + name='image', shape=[2, 10, 40, 40], dtype='float32') + + box, var = ops.prior_box( + input=input, + image=image, + min_sizes=min_sizes, + clip=True, + flip=True) + box_np, var_np = self.get_static_graph_result( + feed={ + 'input': input_np, + 'image': image_np, + }, + fetch_list=[box, var], + with_lod=False) + + with self.dynamic_graph(): + inputs_dy = base.to_variable(input_np) + image_dy = base.to_variable(image_np) + + box_dy, var_dy = ops.prior_box( + input=inputs_dy, + image=image_dy, + min_sizes=min_sizes, + clip=True, + flip=True) + box_dy_np = box_dy.numpy() + var_dy_np = var_dy.numpy() + + self.assertTrue(np.array_equal(box_np, box_dy_np)) + self.assertTrue(np.array_equal(var_np, var_dy_np)) + + def test_prior_box_error(self): + with self.static_graph(): + input = paddle.static.data( + name='input', shape=[2, 10, 32, 32], dtype='int32') + image = paddle.static.data( + name='image', shape=[2, 10, 40, 40], dtype='int32') + self.assertRaises( + TypeError, + ops.prior_box, + input=input, + image=image, + min_sizes=[2, 4], + clip=True, + flip=True) + + paddle.disable_static() + + +class TestMulticlassNms(LayerTest): + def test_multiclass_nms(self): + boxes_np = np.random.rand(10, 81, 4).astype('float32') + scores_np = np.random.rand(10, 81).astype('float32') + rois_num_np = np.array([2, 8]).astype('int32') + with self.static_graph(): + boxes = paddle.static.data( + name='bboxes', + shape=[None, 81, 4], + dtype='float32', + lod_level=1) + scores = paddle.static.data( + name='scores', shape=[None, 81], dtype='float32', lod_level=1) + rois_num = paddle.static.data( + name='rois_num', shape=[None], dtype='int32') + + output = ops.multiclass_nms( + bboxes=boxes, + scores=scores, + background_label=0, + score_threshold=0.5, + nms_top_k=400, + nms_threshold=0.3, + keep_top_k=200, + normalized=False, + return_index=True, + rois_num=rois_num) + out_np, index_np, nms_rois_num_np = self.get_static_graph_result( + feed={ + 'bboxes': boxes_np, + 'scores': scores_np, + 'rois_num': rois_num_np + }, + fetch_list=output, + with_lod=True) + out_np = np.array(out_np) + index_np = np.array(index_np) + nms_rois_num_np = np.array(nms_rois_num_np) + + with self.dynamic_graph(): + boxes_dy = base.to_variable(boxes_np) + scores_dy = base.to_variable(scores_np) + rois_num_dy = base.to_variable(rois_num_np) + + out_dy, index_dy, nms_rois_num_dy = ops.multiclass_nms( + bboxes=boxes_dy, + scores=scores_dy, + background_label=0, + score_threshold=0.5, + nms_top_k=400, + nms_threshold=0.3, + keep_top_k=200, + normalized=False, + return_index=True, + rois_num=rois_num_dy) + out_dy_np = out_dy.numpy() + index_dy_np = index_dy.numpy() + nms_rois_num_dy_np = nms_rois_num_dy.numpy() + + self.assertTrue(np.array_equal(out_np, out_dy_np)) + self.assertTrue(np.array_equal(index_np, index_dy_np)) + self.assertTrue(np.array_equal(nms_rois_num_np, nms_rois_num_dy_np)) + + def test_multiclass_nms_error(self): + with self.static_graph(): + boxes = paddle.static.data( + name='bboxes', shape=[81, 4], dtype='float32', lod_level=1) + scores = paddle.static.data( + name='scores', shape=[81], dtype='float32', lod_level=1) + rois_num = paddle.static.data( + name='rois_num', shape=[40, 41], dtype='int32') + self.assertRaises( + TypeError, + ops.multiclass_nms, + boxes=boxes, + scores=scores, + background_label=0, + score_threshold=0.5, + nms_top_k=400, + nms_threshold=0.3, + keep_top_k=200, + normalized=False, + return_index=True, + rois_num=rois_num) + + +class TestMatrixNMS(LayerTest): + def test_matrix_nms(self): + N, M, C = 7, 1200, 21 + BOX_SIZE = 4 + nms_top_k = 400 + keep_top_k = 200 + score_threshold = 0.01 + post_threshold = 0. + + scores_np = np.random.random((N * M, C)).astype('float32') + scores_np = np.apply_along_axis(softmax, 1, scores_np) + scores_np = np.reshape(scores_np, (N, M, C)) + scores_np = np.transpose(scores_np, (0, 2, 1)) + + boxes_np = np.random.random((N, M, BOX_SIZE)).astype('float32') + boxes_np[:, :, 0:2] = boxes_np[:, :, 0:2] * 0.5 + boxes_np[:, :, 2:4] = boxes_np[:, :, 2:4] * 0.5 + 0.5 + + with self.static_graph(): + boxes = paddle.static.data( + name='boxes', shape=[N, M, BOX_SIZE], dtype='float32') + scores = paddle.static.data( + name='scores', shape=[N, C, M], dtype='float32') + out, index, _ = ops.matrix_nms( + bboxes=boxes, + scores=scores, + score_threshold=score_threshold, + post_threshold=post_threshold, + nms_top_k=nms_top_k, + keep_top_k=keep_top_k, + return_index=True) + out_np, index_np = self.get_static_graph_result( + feed={'boxes': boxes_np, + 'scores': scores_np}, + fetch_list=[out, index], + with_lod=True) + + with self.dynamic_graph(): + boxes_dy = base.to_variable(boxes_np) + scores_dy = base.to_variable(scores_np) + + out_dy, index_dy, _ = ops.matrix_nms( + bboxes=boxes_dy, + scores=scores_dy, + score_threshold=score_threshold, + post_threshold=post_threshold, + nms_top_k=nms_top_k, + keep_top_k=keep_top_k, + return_index=True) + out_dy_np = out_dy.numpy() + index_dy_np = index_dy.numpy() + + self.assertTrue(np.array_equal(out_np, out_dy_np)) + self.assertTrue(np.array_equal(index_np, index_dy_np)) + + def test_matrix_nms_error(self): + with self.static_graph(): + bboxes = paddle.static.data( + name='bboxes', shape=[7, 1200, 4], dtype='float32') + scores = paddle.static.data( + name='data_error', shape=[7, 21, 1200], dtype='int32') + self.assertRaises( + TypeError, + ops.matrix_nms, + bboxes=bboxes, + scores=scores, + score_threshold=0.01, + post_threshold=0., + nms_top_k=400, + keep_top_k=200, + return_index=True) + + paddle.disable_static() + + +class TestBoxCoder(LayerTest): + def test_box_coder(self): + + prior_box_np = np.random.random((81, 4)).astype('float32') + prior_box_var_np = np.random.random((81, 4)).astype('float32') + target_box_np = np.random.random((20, 81, 4)).astype('float32') + + # static + with self.static_graph(): + prior_box = paddle.static.data( + name='prior_box', shape=[81, 4], dtype='float32') + prior_box_var = paddle.static.data( + name='prior_box_var', shape=[81, 4], dtype='float32') + target_box = paddle.static.data( + name='target_box', shape=[20, 81, 4], dtype='float32') + + boxes = ops.box_coder( + prior_box=prior_box, + prior_box_var=prior_box_var, + target_box=target_box, + code_type="decode_center_size", + box_normalized=False) + + boxes_np, = self.get_static_graph_result( + feed={ + 'prior_box': prior_box_np, + 'prior_box_var': prior_box_var_np, + 'target_box': target_box_np, + }, + fetch_list=[boxes], + with_lod=False) + + # dygraph + with self.dynamic_graph(): + prior_box_dy = base.to_variable(prior_box_np) + prior_box_var_dy = base.to_variable(prior_box_var_np) + target_box_dy = base.to_variable(target_box_np) + + boxes_dy = ops.box_coder( + prior_box=prior_box_dy, + prior_box_var=prior_box_var_dy, + target_box=target_box_dy, + code_type="decode_center_size", + box_normalized=False) + + boxes_dy_np = boxes_dy.numpy() + + self.assertTrue(np.array_equal(boxes_np, boxes_dy_np)) + + def test_box_coder_error(self): + with self.static_graph(): + prior_box = paddle.static.data( + name='prior_box', shape=[81, 4], dtype='int32') + prior_box_var = paddle.static.data( + name='prior_box_var', shape=[81, 4], dtype='float32') + target_box = paddle.static.data( + name='target_box', shape=[20, 81, 4], dtype='float32') + + self.assertRaises(TypeError, ops.box_coder, prior_box, + prior_box_var, target_box) + + paddle.disable_static() + + +class TestGenerateProposals(LayerTest): + def test_generate_proposals(self): + scores_np = np.random.rand(2, 3, 4, 4).astype('float32') + bbox_deltas_np = np.random.rand(2, 12, 4, 4).astype('float32') + im_shape_np = np.array([[8, 8], [6, 6]]).astype('float32') + anchors_np = np.reshape(np.arange(4 * 4 * 3 * 4), + [4, 4, 3, 4]).astype('float32') + variances_np = np.ones((4, 4, 3, 4)).astype('float32') + + with self.static_graph(): + scores = paddle.static.data( + name='scores', shape=[2, 3, 4, 4], dtype='float32') + bbox_deltas = paddle.static.data( + name='bbox_deltas', shape=[2, 12, 4, 4], dtype='float32') + im_shape = paddle.static.data( + name='im_shape', shape=[2, 2], dtype='float32') + anchors = paddle.static.data( + name='anchors', shape=[4, 4, 3, 4], dtype='float32') + variances = paddle.static.data( + name='var', shape=[4, 4, 3, 4], dtype='float32') + rois, roi_probs, rois_num = ops.generate_proposals( + scores, + bbox_deltas, + im_shape, + anchors, + variances, + pre_nms_top_n=10, + post_nms_top_n=5, + return_rois_num=True) + rois_stat, roi_probs_stat, rois_num_stat = self.get_static_graph_result( + feed={ + 'scores': scores_np, + 'bbox_deltas': bbox_deltas_np, + 'im_shape': im_shape_np, + 'anchors': anchors_np, + 'var': variances_np + }, + fetch_list=[rois, roi_probs, rois_num], + with_lod=True) + + with self.dynamic_graph(): + scores_dy = base.to_variable(scores_np) + bbox_deltas_dy = base.to_variable(bbox_deltas_np) + im_shape_dy = base.to_variable(im_shape_np) + anchors_dy = base.to_variable(anchors_np) + variances_dy = base.to_variable(variances_np) + rois, roi_probs, rois_num = ops.generate_proposals( + scores_dy, + bbox_deltas_dy, + im_shape_dy, + anchors_dy, + variances_dy, + pre_nms_top_n=10, + post_nms_top_n=5, + return_rois_num=True) + rois_dy = rois.numpy() + roi_probs_dy = roi_probs.numpy() + rois_num_dy = rois_num.numpy() + + self.assertTrue(np.array_equal(np.array(rois_stat), rois_dy)) + self.assertTrue(np.array_equal(np.array(roi_probs_stat), roi_probs_dy)) + self.assertTrue(np.array_equal(np.array(rois_num_stat), rois_num_dy)) + + +if __name__ == '__main__': + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_yolov3_loss.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_yolov3_loss.py new file mode 100644 index 000000000..cec8bc940 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/tests/test_yolov3_loss.py @@ -0,0 +1,414 @@ +# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import division + +import unittest + +import paddle +from paddle import fluid +# add python path of PadleDetection to sys.path +import os +import sys +parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 4))) +if parent_path not in sys.path: + sys.path.append(parent_path) + +from ppdet.modeling.losses import YOLOv3Loss +from ppdet.data.transform.op_helper import jaccard_overlap +import numpy as np + + +def _split_ioup(output, an_num, num_classes): + """ + Split output feature map to output, predicted iou + along channel dimension + """ + ioup = fluid.layers.slice(output, axes=[1], starts=[0], ends=[an_num]) + ioup = fluid.layers.sigmoid(ioup) + oriout = fluid.layers.slice( + output, axes=[1], starts=[an_num], ends=[an_num * (num_classes + 6)]) + return (ioup, oriout) + + +def _split_output(output, an_num, num_classes): + """ + Split output feature map to x, y, w, h, objectness, classification + along channel dimension + """ + x = fluid.layers.strided_slice( + output, + axes=[1], + starts=[0], + ends=[output.shape[1]], + strides=[5 + num_classes]) + y = fluid.layers.strided_slice( + output, + axes=[1], + starts=[1], + ends=[output.shape[1]], + strides=[5 + num_classes]) + w = fluid.layers.strided_slice( + output, + axes=[1], + starts=[2], + ends=[output.shape[1]], + strides=[5 + num_classes]) + h = fluid.layers.strided_slice( + output, + axes=[1], + starts=[3], + ends=[output.shape[1]], + strides=[5 + num_classes]) + obj = fluid.layers.strided_slice( + output, + axes=[1], + starts=[4], + ends=[output.shape[1]], + strides=[5 + num_classes]) + clss = [] + stride = output.shape[1] // an_num + for m in range(an_num): + clss.append( + fluid.layers.slice( + output, + axes=[1], + starts=[stride * m + 5], + ends=[stride * m + 5 + num_classes])) + cls = fluid.layers.transpose( + fluid.layers.stack( + clss, axis=1), perm=[0, 1, 3, 4, 2]) + return (x, y, w, h, obj, cls) + + +def _split_target(target): + """ + split target to x, y, w, h, objectness, classification + along dimension 2 + target is in shape [N, an_num, 6 + class_num, H, W] + """ + tx = target[:, :, 0, :, :] + ty = target[:, :, 1, :, :] + tw = target[:, :, 2, :, :] + th = target[:, :, 3, :, :] + tscale = target[:, :, 4, :, :] + tobj = target[:, :, 5, :, :] + tcls = fluid.layers.transpose(target[:, :, 6:, :, :], perm=[0, 1, 3, 4, 2]) + tcls.stop_gradient = True + return (tx, ty, tw, th, tscale, tobj, tcls) + + +def _calc_obj_loss(output, obj, tobj, gt_box, batch_size, anchors, num_classes, + downsample, ignore_thresh, scale_x_y): + # A prediction bbox overlap any gt_bbox over ignore_thresh, + # objectness loss will be ignored, process as follows: + # 1. get pred bbox, which is same with YOLOv3 infer mode, use yolo_box here + # NOTE: img_size is set as 1.0 to get noramlized pred bbox + bbox, prob = fluid.layers.yolo_box( + x=output, + img_size=fluid.layers.ones( + shape=[batch_size, 2], dtype="int32"), + anchors=anchors, + class_num=num_classes, + conf_thresh=0., + downsample_ratio=downsample, + clip_bbox=False, + scale_x_y=scale_x_y) + # 2. split pred bbox and gt bbox by sample, calculate IoU between pred bbox + # and gt bbox in each sample + if batch_size > 1: + preds = fluid.layers.split(bbox, batch_size, dim=0) + gts = fluid.layers.split(gt_box, batch_size, dim=0) + else: + preds = [bbox] + gts = [gt_box] + probs = [prob] + ious = [] + for pred, gt in zip(preds, gts): + + def box_xywh2xyxy(box): + x = box[:, 0] + y = box[:, 1] + w = box[:, 2] + h = box[:, 3] + return fluid.layers.stack( + [ + x - w / 2., + y - h / 2., + x + w / 2., + y + h / 2., + ], axis=1) + + pred = fluid.layers.squeeze(pred, axes=[0]) + gt = box_xywh2xyxy(fluid.layers.squeeze(gt, axes=[0])) + ious.append(fluid.layers.iou_similarity(pred, gt)) + iou = fluid.layers.stack(ious, axis=0) + # 3. Get iou_mask by IoU between gt bbox and prediction bbox, + # Get obj_mask by tobj(holds gt_score), calculate objectness loss + max_iou = fluid.layers.reduce_max(iou, dim=-1) + iou_mask = fluid.layers.cast(max_iou <= ignore_thresh, dtype="float32") + output_shape = fluid.layers.shape(output) + an_num = len(anchors) // 2 + iou_mask = fluid.layers.reshape(iou_mask, (-1, an_num, output_shape[2], + output_shape[3])) + iou_mask.stop_gradient = True + # NOTE: tobj holds gt_score, obj_mask holds object existence mask + obj_mask = fluid.layers.cast(tobj > 0., dtype="float32") + obj_mask.stop_gradient = True + # For positive objectness grids, objectness loss should be calculated + # For negative objectness grids, objectness loss is calculated only iou_mask == 1.0 + loss_obj = fluid.layers.sigmoid_cross_entropy_with_logits(obj, obj_mask) + loss_obj_pos = fluid.layers.reduce_sum(loss_obj * tobj, dim=[1, 2, 3]) + loss_obj_neg = fluid.layers.reduce_sum( + loss_obj * (1.0 - obj_mask) * iou_mask, dim=[1, 2, 3]) + return loss_obj_pos, loss_obj_neg + + +def fine_grained_loss(output, + target, + gt_box, + batch_size, + num_classes, + anchors, + ignore_thresh, + downsample, + scale_x_y=1., + eps=1e-10): + an_num = len(anchors) // 2 + x, y, w, h, obj, cls = _split_output(output, an_num, num_classes) + tx, ty, tw, th, tscale, tobj, tcls = _split_target(target) + + tscale_tobj = tscale * tobj + + scale_x_y = scale_x_y + + if (abs(scale_x_y - 1.0) < eps): + loss_x = fluid.layers.sigmoid_cross_entropy_with_logits( + x, tx) * tscale_tobj + loss_x = fluid.layers.reduce_sum(loss_x, dim=[1, 2, 3]) + loss_y = fluid.layers.sigmoid_cross_entropy_with_logits( + y, ty) * tscale_tobj + loss_y = fluid.layers.reduce_sum(loss_y, dim=[1, 2, 3]) + else: + dx = scale_x_y * fluid.layers.sigmoid(x) - 0.5 * (scale_x_y - 1.0) + dy = scale_x_y * fluid.layers.sigmoid(y) - 0.5 * (scale_x_y - 1.0) + loss_x = fluid.layers.abs(dx - tx) * tscale_tobj + loss_x = fluid.layers.reduce_sum(loss_x, dim=[1, 2, 3]) + loss_y = fluid.layers.abs(dy - ty) * tscale_tobj + loss_y = fluid.layers.reduce_sum(loss_y, dim=[1, 2, 3]) + + # NOTE: we refined loss function of (w, h) as L1Loss + loss_w = fluid.layers.abs(w - tw) * tscale_tobj + loss_w = fluid.layers.reduce_sum(loss_w, dim=[1, 2, 3]) + loss_h = fluid.layers.abs(h - th) * tscale_tobj + loss_h = fluid.layers.reduce_sum(loss_h, dim=[1, 2, 3]) + + loss_obj_pos, loss_obj_neg = _calc_obj_loss( + output, obj, tobj, gt_box, batch_size, anchors, num_classes, downsample, + ignore_thresh, scale_x_y) + + loss_cls = fluid.layers.sigmoid_cross_entropy_with_logits(cls, tcls) + loss_cls = fluid.layers.elementwise_mul(loss_cls, tobj, axis=0) + loss_cls = fluid.layers.reduce_sum(loss_cls, dim=[1, 2, 3, 4]) + + loss_xys = fluid.layers.reduce_mean(loss_x + loss_y) + loss_whs = fluid.layers.reduce_mean(loss_w + loss_h) + loss_objs = fluid.layers.reduce_mean(loss_obj_pos + loss_obj_neg) + loss_clss = fluid.layers.reduce_mean(loss_cls) + + losses_all = { + "loss_xy": fluid.layers.sum(loss_xys), + "loss_wh": fluid.layers.sum(loss_whs), + "loss_loc": fluid.layers.sum(loss_xys) + fluid.layers.sum(loss_whs), + "loss_obj": fluid.layers.sum(loss_objs), + "loss_cls": fluid.layers.sum(loss_clss), + } + return losses_all, x, y, tx, ty + + +def gt2yolotarget(gt_bbox, gt_class, gt_score, anchors, mask, num_classes, size, + stride): + grid_h, grid_w = size + h, w = grid_h * stride, grid_w * stride + an_hw = np.array(anchors) / np.array([[w, h]]) + target = np.zeros( + (len(mask), 6 + num_classes, grid_h, grid_w), dtype=np.float32) + for b in range(gt_bbox.shape[0]): + gx, gy, gw, gh = gt_bbox[b, :] + cls = gt_class[b] + score = gt_score[b] + if gw <= 0. or gh <= 0. or score <= 0.: + continue + + # find best match anchor index + best_iou = 0. + best_idx = -1 + for an_idx in range(an_hw.shape[0]): + iou = jaccard_overlap([0., 0., gw, gh], + [0., 0., an_hw[an_idx, 0], an_hw[an_idx, 1]]) + if iou > best_iou: + best_iou = iou + best_idx = an_idx + + gi = int(gx * grid_w) + gj = int(gy * grid_h) + + # gtbox should be regresed in this layes if best match + # anchor index in anchor mask of this layer + if best_idx in mask: + best_n = mask.index(best_idx) + + # x, y, w, h, scale + target[best_n, 0, gj, gi] = gx * grid_w - gi + target[best_n, 1, gj, gi] = gy * grid_h - gj + target[best_n, 2, gj, gi] = np.log(gw * w / anchors[best_idx][0]) + target[best_n, 3, gj, gi] = np.log(gh * h / anchors[best_idx][1]) + target[best_n, 4, gj, gi] = 2.0 - gw * gh + + # objectness record gt_score + # if target[best_n, 5, gj, gi] > 0: + # print('find 1 duplicate') + target[best_n, 5, gj, gi] = score + + # classification + target[best_n, 6 + cls, gj, gi] = 1. + + return target + + +class TestYolov3LossOp(unittest.TestCase): + def setUp(self): + self.initTestCase() + x = np.random.uniform(0, 1, self.x_shape).astype('float64') + gtbox = np.random.random(size=self.gtbox_shape).astype('float64') + gtlabel = np.random.randint(0, self.class_num, self.gtbox_shape[:2]) + gtmask = np.random.randint(0, 2, self.gtbox_shape[:2]) + gtbox = gtbox * gtmask[:, :, np.newaxis] + gtlabel = gtlabel * gtmask + + gtscore = np.ones(self.gtbox_shape[:2]).astype('float64') + if self.gtscore: + gtscore = np.random.random(self.gtbox_shape[:2]).astype('float64') + + target = [] + for box, label, score in zip(gtbox, gtlabel, gtscore): + target.append( + gt2yolotarget(box, label, score, self.anchors, self.anchor_mask, + self.class_num, (self.h, self.w + ), self.downsample_ratio)) + + self.target = np.array(target).astype('float64') + + self.mask_anchors = [] + for i in self.anchor_mask: + self.mask_anchors.extend(self.anchors[i]) + self.x = x + self.gtbox = gtbox + self.gtlabel = gtlabel + self.gtscore = gtscore + + def initTestCase(self): + self.b = 8 + self.h = 19 + self.w = 19 + self.anchors = [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], + [59, 119], [116, 90], [156, 198], [373, 326]] + self.anchor_mask = [6, 7, 8] + self.na = len(self.anchor_mask) + self.class_num = 80 + self.ignore_thresh = 0.7 + self.downsample_ratio = 32 + self.x_shape = (self.b, len(self.anchor_mask) * (5 + self.class_num), + self.h, self.w) + self.gtbox_shape = (self.b, 40, 4) + self.gtscore = True + self.use_label_smooth = False + self.scale_x_y = 1. + + def test_loss(self): + x, gtbox, gtlabel, gtscore, target = self.x, self.gtbox, self.gtlabel, self.gtscore, self.target + yolo_loss = YOLOv3Loss( + ignore_thresh=self.ignore_thresh, + label_smooth=self.use_label_smooth, + num_classes=self.class_num, + downsample=self.downsample_ratio, + scale_x_y=self.scale_x_y) + x = paddle.to_tensor(x.astype(np.float32)) + gtbox = paddle.to_tensor(gtbox.astype(np.float32)) + gtlabel = paddle.to_tensor(gtlabel.astype(np.float32)) + gtscore = paddle.to_tensor(gtscore.astype(np.float32)) + t = paddle.to_tensor(target.astype(np.float32)) + anchor = [self.anchors[i] for i in self.anchor_mask] + (yolo_loss1, px, py, tx, ty) = fine_grained_loss( + output=x, + target=t, + gt_box=gtbox, + batch_size=self.b, + num_classes=self.class_num, + anchors=self.mask_anchors, + ignore_thresh=self.ignore_thresh, + downsample=self.downsample_ratio, + scale_x_y=self.scale_x_y) + yolo_loss2 = yolo_loss.yolov3_loss( + x, t, gtbox, anchor, self.downsample_ratio, self.scale_x_y) + for k in yolo_loss2: + self.assertAlmostEqual( + yolo_loss1[k].numpy()[0], + yolo_loss2[k].numpy()[0], + delta=1e-2, + msg=k) + + +class TestYolov3LossNoGTScore(TestYolov3LossOp): + def initTestCase(self): + self.b = 1 + self.h = 76 + self.w = 76 + self.anchors = [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], + [59, 119], [116, 90], [156, 198], [373, 326]] + self.anchor_mask = [0, 1, 2] + self.na = len(self.anchor_mask) + self.class_num = 80 + self.ignore_thresh = 0.7 + self.downsample_ratio = 8 + self.x_shape = (self.b, len(self.anchor_mask) * (5 + self.class_num), + self.h, self.w) + self.gtbox_shape = (self.b, 40, 4) + self.gtscore = False + self.use_label_smooth = False + self.scale_x_y = 1. + + +class TestYolov3LossWithScaleXY(TestYolov3LossOp): + def initTestCase(self): + self.b = 5 + self.h = 38 + self.w = 38 + self.anchors = [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], + [59, 119], [116, 90], [156, 198], [373, 326]] + self.anchor_mask = [3, 4, 5] + self.na = len(self.anchor_mask) + self.class_num = 80 + self.ignore_thresh = 0.7 + self.downsample_ratio = 16 + self.x_shape = (self.b, len(self.anchor_mask) * (5 + self.class_num), + self.h, self.w) + self.gtbox_shape = (self.b, 40, 4) + self.gtscore = True + self.use_label_smooth = False + self.scale_x_y = 1.2 + + +if __name__ == "__main__": + unittest.main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__init__.py new file mode 100644 index 000000000..4aed815d7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__init__.py @@ -0,0 +1,25 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import detr_transformer +from . import utils +from . import matchers +from . import position_encoding +from . import deformable_transformer + +from .detr_transformer import * +from .utils import * +from .matchers import * +from .position_encoding import * +from .deformable_transformer import * diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..fae862b10 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/deformable_transformer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/deformable_transformer.cpython-37.pyc new file mode 100644 index 000000000..75ee28c4b Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/deformable_transformer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/detr_transformer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/detr_transformer.cpython-37.pyc new file mode 100644 index 000000000..11bce7aef Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/detr_transformer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/matchers.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/matchers.cpython-37.pyc new file mode 100644 index 000000000..fc05ed243 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/matchers.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/position_encoding.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/position_encoding.cpython-37.pyc new file mode 100644 index 000000000..195c55c50 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/position_encoding.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/utils.cpython-37.pyc new file mode 100644 index 000000000..f243af8b9 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/__pycache__/utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/deformable_transformer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/deformable_transformer.py new file mode 100644 index 000000000..0c2089a8b --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/deformable_transformer.py @@ -0,0 +1,517 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Modified from Deformable-DETR (https://github.com/fundamentalvision/Deformable-DETR) +# Copyright (c) 2020 SenseTime. All Rights Reserved. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from paddle import ParamAttr + +from ppdet.core.workspace import register +from ..layers import MultiHeadAttention +from .position_encoding import PositionEmbedding +from .utils import _get_clones, deformable_attention_core_func +from ..initializer import linear_init_, constant_, xavier_uniform_, normal_ + +__all__ = ['DeformableTransformer'] + + +class MSDeformableAttention(nn.Layer): + def __init__(self, + embed_dim=256, + num_heads=8, + num_levels=4, + num_points=4, + lr_mult=0.1): + """ + Multi-Scale Deformable Attention Module + """ + super(MSDeformableAttention, self).__init__() + self.embed_dim = embed_dim + self.num_heads = num_heads + self.num_levels = num_levels + self.num_points = num_points + self.total_points = num_heads * num_levels * num_points + + self.head_dim = embed_dim // num_heads + assert self.head_dim * num_heads == self.embed_dim, "embed_dim must be divisible by num_heads" + + self.sampling_offsets = nn.Linear( + embed_dim, + self.total_points * 2, + weight_attr=ParamAttr(learning_rate=lr_mult), + bias_attr=ParamAttr(learning_rate=lr_mult)) + + self.attention_weights = nn.Linear(embed_dim, self.total_points) + self.value_proj = nn.Linear(embed_dim, embed_dim) + self.output_proj = nn.Linear(embed_dim, embed_dim) + + self._reset_parameters() + + def _reset_parameters(self): + # sampling_offsets + constant_(self.sampling_offsets.weight) + thetas = paddle.arange( + self.num_heads, + dtype=paddle.float32) * (2.0 * math.pi / self.num_heads) + grid_init = paddle.stack([thetas.cos(), thetas.sin()], -1) + grid_init = grid_init / grid_init.abs().max(-1, keepdim=True) + grid_init = grid_init.reshape([self.num_heads, 1, 1, 2]).tile( + [1, self.num_levels, self.num_points, 1]) + scaling = paddle.arange( + 1, self.num_points + 1, + dtype=paddle.float32).reshape([1, 1, -1, 1]) + grid_init *= scaling + self.sampling_offsets.bias.set_value(grid_init.flatten()) + # attention_weights + constant_(self.attention_weights.weight) + constant_(self.attention_weights.bias) + # proj + xavier_uniform_(self.value_proj.weight) + constant_(self.value_proj.bias) + xavier_uniform_(self.output_proj.weight) + constant_(self.output_proj.bias) + + def forward(self, + query, + reference_points, + value, + value_spatial_shapes, + value_mask=None): + """ + Args: + query (Tensor): [bs, query_length, C] + reference_points (Tensor): [bs, query_length, n_levels, 2], range in [0, 1], top-left (0,0), + bottom-right (1, 1), including padding area + value (Tensor): [bs, value_length, C] + value_spatial_shapes (Tensor): [n_levels, 2], [(H_0, W_0), (H_1, W_1), ..., (H_{L-1}, W_{L-1})] + value_mask (Tensor): [bs, value_length], True for non-padding elements, False for padding elements + + Returns: + output (Tensor): [bs, Length_{query}, C] + """ + bs, Len_q = query.shape[:2] + Len_v = value.shape[1] + assert int(value_spatial_shapes.prod(1).sum()) == Len_v + + value = self.value_proj(value) + if value_mask is not None: + value_mask = value_mask.astype(value.dtype).unsqueeze(-1) + value *= value_mask + value = value.reshape([bs, Len_v, self.num_heads, self.head_dim]) + + sampling_offsets = self.sampling_offsets(query).reshape( + [bs, Len_q, self.num_heads, self.num_levels, self.num_points, 2]) + attention_weights = self.attention_weights(query).reshape( + [bs, Len_q, self.num_heads, self.num_levels * self.num_points]) + attention_weights = F.softmax(attention_weights, -1).reshape( + [bs, Len_q, self.num_heads, self.num_levels, self.num_points]) + + offset_normalizer = value_spatial_shapes.flip([1]).reshape( + [1, 1, 1, self.num_levels, 1, 2]) + sampling_locations = reference_points.reshape([ + bs, Len_q, 1, self.num_levels, 1, 2 + ]) + sampling_offsets / offset_normalizer + + output = deformable_attention_core_func( + value, value_spatial_shapes, sampling_locations, attention_weights) + output = self.output_proj(output) + + return output + + +class DeformableTransformerEncoderLayer(nn.Layer): + def __init__(self, + d_model=256, + n_head=8, + dim_feedforward=1024, + dropout=0.1, + activation="relu", + n_levels=4, + n_points=4, + weight_attr=None, + bias_attr=None): + super(DeformableTransformerEncoderLayer, self).__init__() + # self attention + self.self_attn = MSDeformableAttention(d_model, n_head, n_levels, + n_points) + self.dropout1 = nn.Dropout(dropout) + self.norm1 = nn.LayerNorm(d_model) + # ffn + self.linear1 = nn.Linear(d_model, dim_feedforward, weight_attr, + bias_attr) + self.activation = getattr(F, activation) + self.dropout2 = nn.Dropout(dropout) + self.linear2 = nn.Linear(dim_feedforward, d_model, weight_attr, + bias_attr) + self.dropout3 = nn.Dropout(dropout) + self.norm2 = nn.LayerNorm(d_model) + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.linear1) + linear_init_(self.linear2) + xavier_uniform_(self.linear1.weight) + xavier_uniform_(self.linear2.weight) + + def with_pos_embed(self, tensor, pos): + return tensor if pos is None else tensor + pos + + def forward_ffn(self, src): + src2 = self.linear2(self.dropout2(self.activation(self.linear1(src)))) + src = src + self.dropout3(src2) + src = self.norm2(src) + return src + + def forward(self, + src, + reference_points, + spatial_shapes, + src_mask=None, + pos_embed=None): + # self attention + src2 = self.self_attn( + self.with_pos_embed(src, pos_embed), reference_points, src, + spatial_shapes, src_mask) + src = src + self.dropout1(src2) + src = self.norm1(src) + # ffn + src = self.forward_ffn(src) + + return src + + +class DeformableTransformerEncoder(nn.Layer): + def __init__(self, encoder_layer, num_layers): + super(DeformableTransformerEncoder, self).__init__() + self.layers = _get_clones(encoder_layer, num_layers) + self.num_layers = num_layers + + @staticmethod + def get_reference_points(spatial_shapes, valid_ratios): + valid_ratios = valid_ratios.unsqueeze(1) + reference_points = [] + for i, (H, W) in enumerate(spatial_shapes.tolist()): + ref_y, ref_x = paddle.meshgrid( + paddle.linspace(0.5, H - 0.5, H), + paddle.linspace(0.5, W - 0.5, W)) + ref_y = ref_y.flatten().unsqueeze(0) / (valid_ratios[:, :, i, 1] * + H) + ref_x = ref_x.flatten().unsqueeze(0) / (valid_ratios[:, :, i, 0] * + W) + reference_points.append(paddle.stack((ref_x, ref_y), axis=-1)) + reference_points = paddle.concat(reference_points, 1).unsqueeze(2) + reference_points = reference_points * valid_ratios + return reference_points + + def forward(self, + src, + spatial_shapes, + src_mask=None, + pos_embed=None, + valid_ratios=None): + output = src + if valid_ratios is None: + valid_ratios = paddle.ones( + [src.shape[0], spatial_shapes.shape[0], 2]) + reference_points = self.get_reference_points(spatial_shapes, + valid_ratios) + for layer in self.layers: + output = layer(output, reference_points, spatial_shapes, src_mask, + pos_embed) + + return output + + +class DeformableTransformerDecoderLayer(nn.Layer): + def __init__(self, + d_model=256, + n_head=8, + dim_feedforward=1024, + dropout=0.1, + activation="relu", + n_levels=4, + n_points=4, + weight_attr=None, + bias_attr=None): + super(DeformableTransformerDecoderLayer, self).__init__() + + # self attention + self.self_attn = MultiHeadAttention(d_model, n_head, dropout=dropout) + self.dropout1 = nn.Dropout(dropout) + self.norm1 = nn.LayerNorm(d_model) + + # cross attention + self.cross_attn = MSDeformableAttention(d_model, n_head, n_levels, + n_points) + self.dropout2 = nn.Dropout(dropout) + self.norm2 = nn.LayerNorm(d_model) + + # ffn + self.linear1 = nn.Linear(d_model, dim_feedforward, weight_attr, + bias_attr) + self.activation = getattr(F, activation) + self.dropout3 = nn.Dropout(dropout) + self.linear2 = nn.Linear(dim_feedforward, d_model, weight_attr, + bias_attr) + self.dropout4 = nn.Dropout(dropout) + self.norm3 = nn.LayerNorm(d_model) + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.linear1) + linear_init_(self.linear2) + xavier_uniform_(self.linear1.weight) + xavier_uniform_(self.linear2.weight) + + def with_pos_embed(self, tensor, pos): + return tensor if pos is None else tensor + pos + + def forward_ffn(self, tgt): + tgt2 = self.linear2(self.dropout3(self.activation(self.linear1(tgt)))) + tgt = tgt + self.dropout4(tgt2) + tgt = self.norm3(tgt) + return tgt + + def forward(self, + tgt, + reference_points, + memory, + memory_spatial_shapes, + memory_mask=None, + query_pos_embed=None): + # self attention + q = k = self.with_pos_embed(tgt, query_pos_embed) + tgt2 = self.self_attn(q, k, value=tgt) + tgt = tgt + self.dropout1(tgt2) + tgt = self.norm1(tgt) + + # cross attention + tgt2 = self.cross_attn( + self.with_pos_embed(tgt, query_pos_embed), reference_points, memory, + memory_spatial_shapes, memory_mask) + tgt = tgt + self.dropout2(tgt2) + tgt = self.norm2(tgt) + + # ffn + tgt = self.forward_ffn(tgt) + + return tgt + + +class DeformableTransformerDecoder(nn.Layer): + def __init__(self, decoder_layer, num_layers, return_intermediate=False): + super(DeformableTransformerDecoder, self).__init__() + self.layers = _get_clones(decoder_layer, num_layers) + self.num_layers = num_layers + self.return_intermediate = return_intermediate + + def forward(self, + tgt, + reference_points, + memory, + memory_spatial_shapes, + memory_mask=None, + query_pos_embed=None): + output = tgt + intermediate = [] + for lid, layer in enumerate(self.layers): + output = layer(output, reference_points, memory, + memory_spatial_shapes, memory_mask, query_pos_embed) + + if self.return_intermediate: + intermediate.append(output) + + if self.return_intermediate: + return paddle.stack(intermediate) + + return output.unsqueeze(0) + + +@register +class DeformableTransformer(nn.Layer): + __shared__ = ['hidden_dim'] + + def __init__(self, + num_queries=300, + position_embed_type='sine', + return_intermediate_dec=True, + backbone_num_channels=[512, 1024, 2048], + num_feature_levels=4, + num_encoder_points=4, + num_decoder_points=4, + hidden_dim=256, + nhead=8, + num_encoder_layers=6, + num_decoder_layers=6, + dim_feedforward=1024, + dropout=0.1, + activation="relu", + lr_mult=0.1, + weight_attr=None, + bias_attr=None): + super(DeformableTransformer, self).__init__() + assert position_embed_type in ['sine', 'learned'], \ + f'ValueError: position_embed_type not supported {position_embed_type}!' + assert len(backbone_num_channels) <= num_feature_levels + + self.hidden_dim = hidden_dim + self.nhead = nhead + self.num_feature_levels = num_feature_levels + + encoder_layer = DeformableTransformerEncoderLayer( + hidden_dim, nhead, dim_feedforward, dropout, activation, + num_feature_levels, num_encoder_points, weight_attr, bias_attr) + self.encoder = DeformableTransformerEncoder(encoder_layer, + num_encoder_layers) + + decoder_layer = DeformableTransformerDecoderLayer( + hidden_dim, nhead, dim_feedforward, dropout, activation, + num_feature_levels, num_decoder_points, weight_attr, bias_attr) + self.decoder = DeformableTransformerDecoder( + decoder_layer, num_decoder_layers, return_intermediate_dec) + + self.level_embed = nn.Embedding(num_feature_levels, hidden_dim) + self.tgt_embed = nn.Embedding(num_queries, hidden_dim) + self.query_pos_embed = nn.Embedding(num_queries, hidden_dim) + + self.reference_points = nn.Linear( + hidden_dim, + 2, + weight_attr=ParamAttr(learning_rate=lr_mult), + bias_attr=ParamAttr(learning_rate=lr_mult)) + + self.input_proj = nn.LayerList() + for in_channels in backbone_num_channels: + self.input_proj.append( + nn.Sequential( + nn.Conv2D( + in_channels, + hidden_dim, + kernel_size=1, + weight_attr=weight_attr, + bias_attr=bias_attr), + nn.GroupNorm(32, hidden_dim))) + in_channels = backbone_num_channels[-1] + for _ in range(num_feature_levels - len(backbone_num_channels)): + self.input_proj.append( + nn.Sequential( + nn.Conv2D( + in_channels, + hidden_dim, + kernel_size=3, + stride=2, + padding=1, + weight_attr=weight_attr, + bias_attr=bias_attr), + nn.GroupNorm(32, hidden_dim))) + in_channels = hidden_dim + + self.position_embedding = PositionEmbedding( + hidden_dim // 2, + normalize=True if position_embed_type == 'sine' else False, + embed_type=position_embed_type, + offset=-0.5) + + self._reset_parameters() + + def _reset_parameters(self): + normal_(self.level_embed.weight) + normal_(self.tgt_embed.weight) + normal_(self.query_pos_embed.weight) + xavier_uniform_(self.reference_points.weight) + constant_(self.reference_points.bias) + for l in self.input_proj: + xavier_uniform_(l[0].weight) + constant_(l[0].bias) + + @classmethod + def from_config(cls, cfg, input_shape): + return {'backbone_num_channels': [i.channels for i in input_shape], } + + def _get_valid_ratio(self, mask): + mask = mask.astype(paddle.float32) + _, H, W = mask.shape + valid_ratio_h = paddle.sum(mask[:, :, 0], 1) / H + valid_ratio_w = paddle.sum(mask[:, 0, :], 1) / W + valid_ratio = paddle.stack([valid_ratio_w, valid_ratio_h], -1) + return valid_ratio + + def forward(self, src_feats, src_mask=None): + srcs = [] + for i in range(len(src_feats)): + srcs.append(self.input_proj[i](src_feats[i])) + if self.num_feature_levels > len(srcs): + len_srcs = len(srcs) + for i in range(len_srcs, self.num_feature_levels): + if i == len_srcs: + srcs.append(self.input_proj[i](src_feats[-1])) + else: + srcs.append(self.input_proj[i](srcs[-1])) + src_flatten = [] + mask_flatten = [] + lvl_pos_embed_flatten = [] + spatial_shapes = [] + valid_ratios = [] + for level, src in enumerate(srcs): + bs, c, h, w = src.shape + spatial_shapes.append([h, w]) + src = src.flatten(2).transpose([0, 2, 1]) + src_flatten.append(src) + if src_mask is not None: + mask = F.interpolate( + src_mask.unsqueeze(0).astype(src.dtype), + size=(h, w))[0].astype('bool') + else: + mask = paddle.ones([bs, h, w], dtype='bool') + valid_ratios.append(self._get_valid_ratio(mask)) + pos_embed = self.position_embedding(mask).flatten(2).transpose( + [0, 2, 1]) + lvl_pos_embed = pos_embed + self.level_embed.weight[level].reshape( + [1, 1, -1]) + lvl_pos_embed_flatten.append(lvl_pos_embed) + mask = mask.astype(src.dtype).flatten(1) + mask_flatten.append(mask) + src_flatten = paddle.concat(src_flatten, 1) + mask_flatten = paddle.concat(mask_flatten, 1) + lvl_pos_embed_flatten = paddle.concat(lvl_pos_embed_flatten, 1) + # [l, 2] + spatial_shapes = paddle.to_tensor(spatial_shapes, dtype='int64') + # [b, l, 2] + valid_ratios = paddle.stack(valid_ratios, 1) + + # encoder + memory = self.encoder(src_flatten, spatial_shapes, mask_flatten, + lvl_pos_embed_flatten, valid_ratios) + + # prepare input for decoder + bs, _, c = memory.shape + query_embed = self.query_pos_embed.weight.unsqueeze(0).tile([bs, 1, 1]) + tgt = self.tgt_embed.weight.unsqueeze(0).tile([bs, 1, 1]) + reference_points = F.sigmoid(self.reference_points(query_embed)) + reference_points_input = reference_points.unsqueeze( + 2) * valid_ratios.unsqueeze(1) + + # decoder + hs = self.decoder(tgt, reference_points_input, memory, spatial_shapes, + mask_flatten, query_embed) + + return (hs, memory, reference_points) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/detr_transformer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/detr_transformer.py new file mode 100644 index 000000000..bd513772d --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/detr_transformer.py @@ -0,0 +1,353 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Modified from DETR (https://github.com/facebookresearch/detr) +# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ppdet.core.workspace import register +from ..layers import MultiHeadAttention, _convert_attention_mask +from .position_encoding import PositionEmbedding +from .utils import _get_clones +from ..initializer import linear_init_, conv_init_, xavier_uniform_, normal_ + +__all__ = ['DETRTransformer'] + + +class TransformerEncoderLayer(nn.Layer): + def __init__(self, + d_model, + nhead, + dim_feedforward=2048, + dropout=0.1, + activation="relu", + attn_dropout=None, + act_dropout=None, + normalize_before=False): + super(TransformerEncoderLayer, self).__init__() + attn_dropout = dropout if attn_dropout is None else attn_dropout + act_dropout = dropout if act_dropout is None else act_dropout + self.normalize_before = normalize_before + + self.self_attn = MultiHeadAttention(d_model, nhead, attn_dropout) + # Implementation of Feedforward model + self.linear1 = nn.Linear(d_model, dim_feedforward) + self.dropout = nn.Dropout(act_dropout, mode="upscale_in_train") + self.linear2 = nn.Linear(dim_feedforward, d_model) + + self.norm1 = nn.LayerNorm(d_model) + self.norm2 = nn.LayerNorm(d_model) + self.dropout1 = nn.Dropout(dropout, mode="upscale_in_train") + self.dropout2 = nn.Dropout(dropout, mode="upscale_in_train") + self.activation = getattr(F, activation) + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.linear1) + linear_init_(self.linear2) + + @staticmethod + def with_pos_embed(tensor, pos_embed): + return tensor if pos_embed is None else tensor + pos_embed + + def forward(self, src, src_mask=None, pos_embed=None): + src_mask = _convert_attention_mask(src_mask, src.dtype) + + residual = src + if self.normalize_before: + src = self.norm1(src) + q = k = self.with_pos_embed(src, pos_embed) + src = self.self_attn(q, k, value=src, attn_mask=src_mask) + + src = residual + self.dropout1(src) + if not self.normalize_before: + src = self.norm1(src) + + residual = src + if self.normalize_before: + src = self.norm2(src) + src = self.linear2(self.dropout(self.activation(self.linear1(src)))) + src = residual + self.dropout2(src) + if not self.normalize_before: + src = self.norm2(src) + return src + + +class TransformerEncoder(nn.Layer): + def __init__(self, encoder_layer, num_layers, norm=None): + super(TransformerEncoder, self).__init__() + self.layers = _get_clones(encoder_layer, num_layers) + self.num_layers = num_layers + self.norm = norm + + def forward(self, src, src_mask=None, pos_embed=None): + src_mask = _convert_attention_mask(src_mask, src.dtype) + + output = src + for layer in self.layers: + output = layer(output, src_mask=src_mask, pos_embed=pos_embed) + + if self.norm is not None: + output = self.norm(output) + + return output + + +class TransformerDecoderLayer(nn.Layer): + def __init__(self, + d_model, + nhead, + dim_feedforward=2048, + dropout=0.1, + activation="relu", + attn_dropout=None, + act_dropout=None, + normalize_before=False): + super(TransformerDecoderLayer, self).__init__() + attn_dropout = dropout if attn_dropout is None else attn_dropout + act_dropout = dropout if act_dropout is None else act_dropout + self.normalize_before = normalize_before + + self.self_attn = MultiHeadAttention(d_model, nhead, attn_dropout) + self.cross_attn = MultiHeadAttention(d_model, nhead, attn_dropout) + # Implementation of Feedforward model + self.linear1 = nn.Linear(d_model, dim_feedforward) + self.dropout = nn.Dropout(act_dropout, mode="upscale_in_train") + self.linear2 = nn.Linear(dim_feedforward, d_model) + + self.norm1 = nn.LayerNorm(d_model) + self.norm2 = nn.LayerNorm(d_model) + self.norm3 = nn.LayerNorm(d_model) + self.dropout1 = nn.Dropout(dropout, mode="upscale_in_train") + self.dropout2 = nn.Dropout(dropout, mode="upscale_in_train") + self.dropout3 = nn.Dropout(dropout, mode="upscale_in_train") + self.activation = getattr(F, activation) + self._reset_parameters() + + def _reset_parameters(self): + linear_init_(self.linear1) + linear_init_(self.linear2) + + @staticmethod + def with_pos_embed(tensor, pos_embed): + return tensor if pos_embed is None else tensor + pos_embed + + def forward(self, + tgt, + memory, + tgt_mask=None, + memory_mask=None, + pos_embed=None, + query_pos_embed=None): + tgt_mask = _convert_attention_mask(tgt_mask, tgt.dtype) + memory_mask = _convert_attention_mask(memory_mask, memory.dtype) + + residual = tgt + if self.normalize_before: + tgt = self.norm1(tgt) + q = k = self.with_pos_embed(tgt, query_pos_embed) + tgt = self.self_attn(q, k, value=tgt, attn_mask=tgt_mask) + tgt = residual + self.dropout1(tgt) + if not self.normalize_before: + tgt = self.norm1(tgt) + + residual = tgt + if self.normalize_before: + tgt = self.norm2(tgt) + q = self.with_pos_embed(tgt, query_pos_embed) + k = self.with_pos_embed(memory, pos_embed) + tgt = self.cross_attn(q, k, value=memory, attn_mask=memory_mask) + tgt = residual + self.dropout2(tgt) + if not self.normalize_before: + tgt = self.norm2(tgt) + + residual = tgt + if self.normalize_before: + tgt = self.norm3(tgt) + tgt = self.linear2(self.dropout(self.activation(self.linear1(tgt)))) + tgt = residual + self.dropout3(tgt) + if not self.normalize_before: + tgt = self.norm3(tgt) + return tgt + + +class TransformerDecoder(nn.Layer): + def __init__(self, + decoder_layer, + num_layers, + norm=None, + return_intermediate=False): + super(TransformerDecoder, self).__init__() + self.layers = _get_clones(decoder_layer, num_layers) + self.num_layers = num_layers + self.norm = norm + self.return_intermediate = return_intermediate + + def forward(self, + tgt, + memory, + tgt_mask=None, + memory_mask=None, + pos_embed=None, + query_pos_embed=None): + tgt_mask = _convert_attention_mask(tgt_mask, tgt.dtype) + memory_mask = _convert_attention_mask(memory_mask, memory.dtype) + + output = tgt + intermediate = [] + for layer in self.layers: + output = layer( + output, + memory, + tgt_mask=tgt_mask, + memory_mask=memory_mask, + pos_embed=pos_embed, + query_pos_embed=query_pos_embed) + if self.return_intermediate: + intermediate.append(self.norm(output)) + + if self.norm is not None: + output = self.norm(output) + + if self.return_intermediate: + return paddle.stack(intermediate) + + return output.unsqueeze(0) + + +@register +class DETRTransformer(nn.Layer): + __shared__ = ['hidden_dim'] + + def __init__(self, + num_queries=100, + position_embed_type='sine', + return_intermediate_dec=True, + backbone_num_channels=2048, + hidden_dim=256, + nhead=8, + num_encoder_layers=6, + num_decoder_layers=6, + dim_feedforward=2048, + dropout=0.1, + activation="relu", + attn_dropout=None, + act_dropout=None, + normalize_before=False): + super(DETRTransformer, self).__init__() + assert position_embed_type in ['sine', 'learned'],\ + f'ValueError: position_embed_type not supported {position_embed_type}!' + self.hidden_dim = hidden_dim + self.nhead = nhead + + encoder_layer = TransformerEncoderLayer( + hidden_dim, nhead, dim_feedforward, dropout, activation, + attn_dropout, act_dropout, normalize_before) + encoder_norm = nn.LayerNorm(hidden_dim) if normalize_before else None + self.encoder = TransformerEncoder(encoder_layer, num_encoder_layers, + encoder_norm) + + decoder_layer = TransformerDecoderLayer( + hidden_dim, nhead, dim_feedforward, dropout, activation, + attn_dropout, act_dropout, normalize_before) + decoder_norm = nn.LayerNorm(hidden_dim) + self.decoder = TransformerDecoder( + decoder_layer, + num_decoder_layers, + decoder_norm, + return_intermediate=return_intermediate_dec) + + self.input_proj = nn.Conv2D( + backbone_num_channels, hidden_dim, kernel_size=1) + self.query_pos_embed = nn.Embedding(num_queries, hidden_dim) + self.position_embedding = PositionEmbedding( + hidden_dim // 2, + normalize=True if position_embed_type == 'sine' else False, + embed_type=position_embed_type) + + self._reset_parameters() + + def _reset_parameters(self): + for p in self.parameters(): + if p.dim() > 1: + xavier_uniform_(p) + conv_init_(self.input_proj) + normal_(self.query_pos_embed.weight) + + @classmethod + def from_config(cls, cfg, input_shape): + return { + 'backbone_num_channels': [i.channels for i in input_shape][-1], + } + + def forward(self, src, src_mask=None): + r""" + Applies a Transformer model on the inputs. + + Parameters: + src (List(Tensor)): Backbone feature maps with shape [[bs, c, h, w]]. + src_mask (Tensor, optional): A tensor used in multi-head attention + to prevents attention to some unwanted positions, usually the + paddings or the subsequent positions. It is a tensor with shape + [bs, H, W]`. When the data type is bool, the unwanted positions + have `False` values and the others have `True` values. When the + data type is int, the unwanted positions have 0 values and the + others have 1 values. When the data type is float, the unwanted + positions have `-INF` values and the others have 0 values. It + can be None when nothing wanted or needed to be prevented + attention to. Default None. + + Returns: + output (Tensor): [num_levels, batch_size, num_queries, hidden_dim] + memory (Tensor): [batch_size, hidden_dim, h, w] + """ + # use last level feature map + src_proj = self.input_proj(src[-1]) + bs, c, h, w = src_proj.shape + # flatten [B, C, H, W] to [B, HxW, C] + src_flatten = src_proj.flatten(2).transpose([0, 2, 1]) + if src_mask is not None: + src_mask = F.interpolate( + src_mask.unsqueeze(0).astype(src_flatten.dtype), + size=(h, w))[0].astype('bool') + else: + src_mask = paddle.ones([bs, h, w], dtype='bool') + pos_embed = self.position_embedding(src_mask).flatten(2).transpose( + [0, 2, 1]) + + src_mask = _convert_attention_mask(src_mask, src_flatten.dtype) + src_mask = src_mask.reshape([bs, 1, 1, -1]) + + memory = self.encoder( + src_flatten, src_mask=src_mask, pos_embed=pos_embed) + + query_pos_embed = self.query_pos_embed.weight.unsqueeze(0).tile( + [bs, 1, 1]) + tgt = paddle.zeros_like(query_pos_embed) + output = self.decoder( + tgt, + memory, + memory_mask=src_mask, + pos_embed=pos_embed, + query_pos_embed=query_pos_embed) + + return (output, memory.transpose([0, 2, 1]).reshape([bs, c, h, w]), + src_proj, src_mask.reshape([bs, 1, 1, h, w])) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/matchers.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/matchers.py new file mode 100644 index 000000000..794d86328 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/matchers.py @@ -0,0 +1,126 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Modified from DETR (https://github.com/facebookresearch/detr) +# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F +from scipy.optimize import linear_sum_assignment + +from ppdet.core.workspace import register, serializable +from ..losses.iou_loss import GIoULoss +from .utils import bbox_cxcywh_to_xyxy + +__all__ = ['HungarianMatcher'] + + +@register +@serializable +class HungarianMatcher(nn.Layer): + __shared__ = ['use_focal_loss'] + + def __init__(self, + matcher_coeff={'class': 1, + 'bbox': 5, + 'giou': 2}, + use_focal_loss=False, + alpha=0.25, + gamma=2.0): + r""" + Args: + matcher_coeff (dict): The coefficient of hungarian matcher cost. + """ + super(HungarianMatcher, self).__init__() + self.matcher_coeff = matcher_coeff + self.use_focal_loss = use_focal_loss + self.alpha = alpha + self.gamma = gamma + + self.giou_loss = GIoULoss() + + def forward(self, boxes, logits, gt_bbox, gt_class): + r""" + Args: + boxes (Tensor): [b, query, 4] + logits (Tensor): [b, query, num_classes] + gt_bbox (List(Tensor)): list[[n, 4]] + gt_class (List(Tensor)): list[[n, 1]] + + Returns: + A list of size batch_size, containing tuples of (index_i, index_j) where: + - index_i is the indices of the selected predictions (in order) + - index_j is the indices of the corresponding selected targets (in order) + For each batch element, it holds: + len(index_i) = len(index_j) = min(num_queries, num_target_boxes) + """ + bs, num_queries = boxes.shape[:2] + + num_gts = sum(len(a) for a in gt_class) + if num_gts == 0: + return [(paddle.to_tensor( + [], dtype=paddle.int64), paddle.to_tensor( + [], dtype=paddle.int64)) for _ in range(bs)] + + # We flatten to compute the cost matrices in a batch + # [batch_size * num_queries, num_classes] + out_prob = F.sigmoid(logits.flatten( + 0, 1)) if self.use_focal_loss else F.softmax(logits.flatten(0, 1)) + # [batch_size * num_queries, 4] + out_bbox = boxes.flatten(0, 1) + + # Also concat the target labels and boxes + tgt_ids = paddle.concat(gt_class).flatten() + tgt_bbox = paddle.concat(gt_bbox) + + # Compute the classification cost + if self.use_focal_loss: + neg_cost_class = (1 - self.alpha) * (out_prob**self.gamma) * (-( + 1 - out_prob + 1e-8).log()) + pos_cost_class = self.alpha * ( + (1 - out_prob)**self.gamma) * (-(out_prob + 1e-8).log()) + cost_class = paddle.gather( + pos_cost_class, tgt_ids, axis=1) - paddle.gather( + neg_cost_class, tgt_ids, axis=1) + else: + cost_class = -paddle.gather(out_prob, tgt_ids, axis=1) + + # Compute the L1 cost between boxes + cost_bbox = ( + out_bbox.unsqueeze(1) - tgt_bbox.unsqueeze(0)).abs().sum(-1) + + # Compute the giou cost betwen boxes + cost_giou = self.giou_loss( + bbox_cxcywh_to_xyxy(out_bbox.unsqueeze(1)), + bbox_cxcywh_to_xyxy(tgt_bbox.unsqueeze(0))).squeeze(-1) + + # Final cost matrix + C = self.matcher_coeff['class'] * cost_class + self.matcher_coeff['bbox'] * cost_bbox + \ + self.matcher_coeff['giou'] * cost_giou + C = C.reshape([bs, num_queries, -1]) + C = [a.squeeze(0) for a in C.chunk(bs)] + + sizes = [a.shape[0] for a in gt_bbox] + indices = [ + linear_sum_assignment(c.split(sizes, -1)[i].numpy()) + for i, c in enumerate(C) + ] + return [(paddle.to_tensor( + i, dtype=paddle.int64), paddle.to_tensor( + j, dtype=paddle.int64)) for i, j in indices] diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/position_encoding.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/position_encoding.py new file mode 100644 index 000000000..e54165918 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/position_encoding.py @@ -0,0 +1,108 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Modified from DETR (https://github.com/facebookresearch/detr) +# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import paddle +import paddle.nn as nn + +from ppdet.core.workspace import register, serializable + + +@register +@serializable +class PositionEmbedding(nn.Layer): + def __init__(self, + num_pos_feats=128, + temperature=10000, + normalize=True, + scale=None, + embed_type='sine', + num_embeddings=50, + offset=0.): + super(PositionEmbedding, self).__init__() + assert embed_type in ['sine', 'learned'] + + self.embed_type = embed_type + self.offset = offset + self.eps = 1e-6 + if self.embed_type == 'sine': + self.num_pos_feats = num_pos_feats + self.temperature = temperature + self.normalize = normalize + if scale is not None and normalize is False: + raise ValueError("normalize should be True if scale is passed") + if scale is None: + scale = 2 * math.pi + self.scale = scale + elif self.embed_type == 'learned': + self.row_embed = nn.Embedding(num_embeddings, num_pos_feats) + self.col_embed = nn.Embedding(num_embeddings, num_pos_feats) + else: + raise ValueError(f"not supported {self.embed_type}") + + def forward(self, mask): + """ + Args: + mask (Tensor): [B, H, W] + Returns: + pos (Tensor): [B, C, H, W] + """ + assert mask.dtype == paddle.bool + if self.embed_type == 'sine': + mask = mask.astype('float32') + y_embed = mask.cumsum(1, dtype='float32') + x_embed = mask.cumsum(2, dtype='float32') + if self.normalize: + y_embed = (y_embed + self.offset) / ( + y_embed[:, -1:, :] + self.eps) * self.scale + x_embed = (x_embed + self.offset) / ( + x_embed[:, :, -1:] + self.eps) * self.scale + + dim_t = 2 * (paddle.arange(self.num_pos_feats) // + 2).astype('float32') + dim_t = self.temperature**(dim_t / self.num_pos_feats) + + pos_x = x_embed.unsqueeze(-1) / dim_t + pos_y = y_embed.unsqueeze(-1) / dim_t + pos_x = paddle.stack( + (pos_x[:, :, :, 0::2].sin(), pos_x[:, :, :, 1::2].cos()), + axis=4).flatten(3) + pos_y = paddle.stack( + (pos_y[:, :, :, 0::2].sin(), pos_y[:, :, :, 1::2].cos()), + axis=4).flatten(3) + pos = paddle.concat((pos_y, pos_x), axis=3).transpose([0, 3, 1, 2]) + return pos + elif self.embed_type == 'learned': + h, w = mask.shape[-2:] + i = paddle.arange(w) + j = paddle.arange(h) + x_emb = self.col_embed(i) + y_emb = self.row_embed(j) + pos = paddle.concat( + [ + x_emb.unsqueeze(0).repeat(h, 1, 1), + y_emb.unsqueeze(1).repeat(1, w, 1), + ], + axis=-1).transpose([2, 0, 1]).unsqueeze(0).tile(mask.shape[0], + 1, 1, 1) + return pos + else: + raise ValueError(f"not supported {self.embed_type}") diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/utils.py new file mode 100644 index 000000000..414ada588 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/modeling/transformers/utils.py @@ -0,0 +1,109 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# Modified from DETR (https://github.com/facebookresearch/detr) +# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import copy +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ..bbox_utils import bbox_overlaps + +__all__ = [ + '_get_clones', 'bbox_overlaps', 'bbox_cxcywh_to_xyxy', + 'bbox_xyxy_to_cxcywh', 'sigmoid_focal_loss', 'inverse_sigmoid', + 'deformable_attention_core_func' +] + + +def _get_clones(module, N): + return nn.LayerList([copy.deepcopy(module) for _ in range(N)]) + + +def bbox_cxcywh_to_xyxy(x): + x_c, y_c, w, h = x.unbind(-1) + b = [(x_c - 0.5 * w), (y_c - 0.5 * h), (x_c + 0.5 * w), (y_c + 0.5 * h)] + return paddle.stack(b, axis=-1) + + +def bbox_xyxy_to_cxcywh(x): + x0, y0, x1, y1 = x.unbind(-1) + b = [(x0 + x1) / 2, (y0 + y1) / 2, (x1 - x0), (y1 - y0)] + return paddle.stack(b, axis=-1) + + +def sigmoid_focal_loss(logit, label, normalizer=1.0, alpha=0.25, gamma=2.0): + prob = F.sigmoid(logit) + ce_loss = F.binary_cross_entropy_with_logits(logit, label, reduction="none") + p_t = prob * label + (1 - prob) * (1 - label) + loss = ce_loss * ((1 - p_t)**gamma) + + if alpha >= 0: + alpha_t = alpha * label + (1 - alpha) * (1 - label) + loss = alpha_t * loss + return loss.mean(1).sum() / normalizer + + +def inverse_sigmoid(x, eps=1e-6): + x = x.clip(min=0., max=1.) + return paddle.log(x / (1 - x + eps) + eps) + + +def deformable_attention_core_func(value, value_spatial_shapes, + sampling_locations, attention_weights): + """ + Args: + value (Tensor): [bs, value_length, n_head, c] + value_spatial_shapes (Tensor): [n_levels, 2] + sampling_locations (Tensor): [bs, query_length, n_head, n_levels, n_points, 2] + attention_weights (Tensor): [bs, query_length, n_head, n_levels, n_points] + + Returns: + output (Tensor): [bs, Length_{query}, C] + """ + bs, Len_v, n_head, c = value.shape + _, Len_q, n_head, n_levels, n_points, _ = sampling_locations.shape + + value_list = value.split(value_spatial_shapes.prod(1).tolist(), axis=1) + sampling_grids = 2 * sampling_locations - 1 + sampling_value_list = [] + for level, (h, w) in enumerate(value_spatial_shapes.tolist()): + # N_, H_*W_, M_, D_ -> N_, H_*W_, M_*D_ -> N_, M_*D_, H_*W_ -> N_*M_, D_, H_, W_ + value_l_ = value_list[level].flatten(2).transpose( + [0, 2, 1]).reshape([bs * n_head, c, h, w]) + # N_, Lq_, M_, P_, 2 -> N_, M_, Lq_, P_, 2 -> N_*M_, Lq_, P_, 2 + sampling_grid_l_ = sampling_grids[:, :, :, level].transpose( + [0, 2, 1, 3, 4]).flatten(0, 1) + # N_*M_, D_, Lq_, P_ + sampling_value_l_ = F.grid_sample( + value_l_, + sampling_grid_l_, + mode='bilinear', + padding_mode='zeros', + align_corners=False) + sampling_value_list.append(sampling_value_l_) + # (N_, Lq_, M_, L_, P_) -> (N_, M_, Lq_, L_, P_) -> (N_*M_, 1, Lq_, L_*P_) + attention_weights = attention_weights.transpose([0, 2, 1, 3, 4]).reshape( + [bs * n_head, 1, Len_q, n_levels * n_points]) + output = (paddle.stack( + sampling_value_list, axis=-2).flatten(-2) * + attention_weights).sum(-1).reshape([bs, n_head * c, Len_q]) + + return output.transpose([0, 2, 1]) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/optimizer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/optimizer.py new file mode 100644 index 000000000..fcdcbd8d6 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/optimizer.py @@ -0,0 +1,333 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import math +import paddle +import paddle.nn as nn + +import paddle.optimizer as optimizer +import paddle.regularizer as regularizer + +from ppdet.core.workspace import register, serializable + +__all__ = ['LearningRate', 'OptimizerBuilder'] + +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@serializable +class CosineDecay(object): + """ + Cosine learning rate decay + + Args: + max_epochs (int): max epochs for the training process. + if you commbine cosine decay with warmup, it is recommended that + the max_iters is much larger than the warmup iter + """ + + def __init__(self, max_epochs=1000, use_warmup=True): + self.max_epochs = max_epochs + self.use_warmup = use_warmup + + def __call__(self, + base_lr=None, + boundary=None, + value=None, + step_per_epoch=None): + assert base_lr is not None, "either base LR or values should be provided" + + max_iters = self.max_epochs * int(step_per_epoch) + + if boundary is not None and value is not None and self.use_warmup: + for i in range(int(boundary[-1]), max_iters): + boundary.append(i) + + decayed_lr = base_lr * 0.5 * ( + math.cos(i * math.pi / max_iters) + 1) + value.append(decayed_lr) + return optimizer.lr.PiecewiseDecay(boundary, value) + + return optimizer.lr.CosineAnnealingDecay(base_lr, T_max=max_iters) + + +@serializable +class PiecewiseDecay(object): + """ + Multi step learning rate decay + + Args: + gamma (float | list): decay factor + milestones (list): steps at which to decay learning rate + """ + + def __init__(self, + gamma=[0.1, 0.01], + milestones=[8, 11], + values=None, + use_warmup=True): + super(PiecewiseDecay, self).__init__() + if type(gamma) is not list: + self.gamma = [] + for i in range(len(milestones)): + self.gamma.append(gamma / 10**i) + else: + self.gamma = gamma + self.milestones = milestones + self.values = values + self.use_warmup = use_warmup + + def __call__(self, + base_lr=None, + boundary=None, + value=None, + step_per_epoch=None): + if boundary is not None and self.use_warmup: + boundary.extend([int(step_per_epoch) * i for i in self.milestones]) + else: + # do not use LinearWarmup + boundary = [int(step_per_epoch) * i for i in self.milestones] + value = [base_lr] # during step[0, boundary[0]] is base_lr + + # self.values is setted directly in config + if self.values is not None: + assert len(self.milestones) + 1 == len(self.values) + return optimizer.lr.PiecewiseDecay(boundary, self.values) + + # value is computed by self.gamma + value = value if value is not None else [base_lr] + for i in self.gamma: + value.append(base_lr * i) + + return optimizer.lr.PiecewiseDecay(boundary, value) + + +@serializable +class LinearWarmup(object): + """ + Warm up learning rate linearly + + Args: + steps (int): warm up steps + start_factor (float): initial learning rate factor + """ + + def __init__(self, steps=500, start_factor=1. / 3): + super(LinearWarmup, self).__init__() + self.steps = steps + self.start_factor = start_factor + + def __call__(self, base_lr, step_per_epoch): + boundary = [] + value = [] + for i in range(self.steps + 1): + if self.steps > 0: + alpha = i / self.steps + factor = self.start_factor * (1 - alpha) + alpha + lr = base_lr * factor + value.append(lr) + if i > 0: + boundary.append(i) + return boundary, value + + +@serializable +class BurninWarmup(object): + """ + Warm up learning rate in burnin mode + Args: + steps (int): warm up steps + """ + + def __init__(self, steps=1000): + super(BurninWarmup, self).__init__() + self.steps = steps + + def __call__(self, base_lr, step_per_epoch): + boundary = [] + value = [] + burnin = min(self.steps, step_per_epoch) + for i in range(burnin + 1): + factor = (i * 1.0 / burnin)**4 + lr = base_lr * factor + value.append(lr) + if i > 0: + boundary.append(i) + return boundary, value + + +@register +class LearningRate(object): + """ + Learning Rate configuration + + Args: + base_lr (float): base learning rate + schedulers (list): learning rate schedulers + """ + __category__ = 'optim' + + def __init__(self, + base_lr=0.01, + schedulers=[PiecewiseDecay(), LinearWarmup()]): + super(LearningRate, self).__init__() + self.base_lr = base_lr + self.schedulers = schedulers + + def __call__(self, step_per_epoch): + assert len(self.schedulers) >= 1 + if not self.schedulers[0].use_warmup: + return self.schedulers[0](base_lr=self.base_lr, + step_per_epoch=step_per_epoch) + + # TODO: split warmup & decay + # warmup + boundary, value = self.schedulers[1](self.base_lr, step_per_epoch) + # decay + decay_lr = self.schedulers[0](self.base_lr, boundary, value, + step_per_epoch) + return decay_lr + + +@register +class OptimizerBuilder(): + """ + Build optimizer handles + Args: + regularizer (object): an `Regularizer` instance + optimizer (object): an `Optimizer` instance + """ + __category__ = 'optim' + + def __init__(self, + clip_grad_by_norm=None, + regularizer={'type': 'L2', + 'factor': .0001}, + optimizer={'type': 'Momentum', + 'momentum': .9}): + self.clip_grad_by_norm = clip_grad_by_norm + self.regularizer = regularizer + self.optimizer = optimizer + + def __call__(self, learning_rate, model=None): + if self.clip_grad_by_norm is not None: + grad_clip = nn.ClipGradByGlobalNorm( + clip_norm=self.clip_grad_by_norm) + else: + grad_clip = None + if self.regularizer and self.regularizer != 'None': + reg_type = self.regularizer['type'] + 'Decay' + reg_factor = self.regularizer['factor'] + regularization = getattr(regularizer, reg_type)(reg_factor) + else: + regularization = None + + optim_args = self.optimizer.copy() + optim_type = optim_args['type'] + del optim_args['type'] + if optim_type != 'AdamW': + optim_args['weight_decay'] = regularization + op = getattr(optimizer, optim_type) + + if 'without_weight_decay_params' in optim_args: + keys = optim_args['without_weight_decay_params'] + params = [{ + 'params': [ + p for n, p in model.named_parameters() + if any([k in n for k in keys]) + ], + 'weight_decay': 0. + }, { + 'params': [ + p for n, p in model.named_parameters() + if all([k not in n for k in keys]) + ] + }] + del optim_args['without_weight_decay_params'] + else: + params = model.parameters() + + return op(learning_rate=learning_rate, + parameters=params, + grad_clip=grad_clip, + **optim_args) + + +class ModelEMA(object): + """ + Exponential Weighted Average for Deep Neutal Networks + Args: + model (nn.Layer): Detector of model. + decay (int): The decay used for updating ema parameter. + Ema's parameter are updated with the formula: + `ema_param = decay * ema_param + (1 - decay) * cur_param`. + Defaults is 0.9998. + use_thres_step (bool): Whether set decay by thres_step or not + cycle_epoch (int): The epoch of interval to reset ema_param and + step. Defaults is -1, which means not reset. Its function is to + add a regular effect to ema, which is set according to experience + and is effective when the total training epoch is large. + """ + + def __init__(self, + model, + decay=0.9998, + use_thres_step=False, + cycle_epoch=-1): + self.step = 0 + self.epoch = 0 + self.decay = decay + self.state_dict = dict() + for k, v in model.state_dict().items(): + self.state_dict[k] = paddle.zeros_like(v) + self.use_thres_step = use_thres_step + self.cycle_epoch = cycle_epoch + + def reset(self): + self.step = 0 + self.epoch = 0 + for k, v in self.state_dict.items(): + self.state_dict[k] = paddle.zeros_like(v) + + def update(self, model): + if self.use_thres_step: + decay = min(self.decay, (1 + self.step) / (10 + self.step)) + else: + decay = self.decay + self._decay = decay + model_dict = model.state_dict() + for k, v in self.state_dict.items(): + v = decay * v + (1 - decay) * model_dict[k] + v.stop_gradient = True + self.state_dict[k] = v + self.step += 1 + + def apply(self): + if self.step == 0: + return self.state_dict + state_dict = dict() + for k, v in self.state_dict.items(): + v = v / (1 - self._decay**self.step) + v.stop_gradient = True + state_dict[k] = v + self.epoch += 1 + if self.cycle_epoch > 0 and self.epoch == self.cycle_epoch: + self.reset() + + return state_dict diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__init__.py new file mode 100644 index 000000000..dc22d0717 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__init__.py @@ -0,0 +1,82 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from . import prune +from . import quant +from . import distill +from . import unstructured_prune + +from .prune import * +from .quant import * +from .distill import * +from .unstructured_prune import * + +import yaml +from ppdet.core.workspace import load_config +from ppdet.utils.checkpoint import load_pretrain_weight + + +def build_slim_model(cfg, slim_cfg, mode='train'): + with open(slim_cfg) as f: + slim_load_cfg = yaml.load(f, Loader=yaml.Loader) + if mode != 'train' and slim_load_cfg['slim'] == 'Distill': + return cfg + + if slim_load_cfg['slim'] == 'Distill': + model = DistillModel(cfg, slim_cfg) + cfg['model'] = model + elif slim_load_cfg['slim'] == 'DistillPrune': + if mode == 'train': + model = DistillModel(cfg, slim_cfg) + pruner = create(cfg.pruner) + pruner(model.student_model) + else: + model = create(cfg.architecture) + weights = cfg.weights + load_config(slim_cfg) + pruner = create(cfg.pruner) + model = pruner(model) + load_pretrain_weight(model, weights) + cfg['model'] = model + cfg['slim_type'] = cfg.slim + elif slim_load_cfg['slim'] == 'PTQ': + model = create(cfg.architecture) + load_config(slim_cfg) + load_pretrain_weight(model, cfg.weights) + slim = create(cfg.slim) + cfg['slim_type'] = cfg.slim + cfg['model'] = slim(model) + cfg['slim'] = slim + elif slim_load_cfg['slim'] == 'UnstructuredPruner': + load_config(slim_cfg) + slim = create(cfg.slim) + cfg['slim_type'] = cfg.slim + cfg['slim'] = slim + cfg['unstructured_prune'] = True + else: + load_config(slim_cfg) + model = create(cfg.architecture) + if mode == 'train': + load_pretrain_weight(model, cfg.pretrain_weights) + slim = create(cfg.slim) + cfg['slim_type'] = cfg.slim + # TODO: fix quant export model in framework. + if mode == 'test' and slim_load_cfg['slim'] == 'QAT': + slim.quant_config['activation_preprocess_type'] = None + cfg['model'] = slim(model) + cfg['slim'] = slim + if mode != 'train': + load_pretrain_weight(cfg['model'], cfg.weights) + + return cfg diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..6efc3d48f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/distill.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/distill.cpython-37.pyc new file mode 100644 index 000000000..2bcf7f5ed Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/distill.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/prune.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/prune.cpython-37.pyc new file mode 100644 index 000000000..db85654e8 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/prune.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/quant.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/quant.cpython-37.pyc new file mode 100644 index 000000000..532a6a089 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/quant.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/unstructured_prune.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/unstructured_prune.cpython-37.pyc new file mode 100644 index 000000000..acdb77ed0 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/__pycache__/unstructured_prune.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/distill.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/distill.py new file mode 100644 index 000000000..b808553dd --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/distill.py @@ -0,0 +1,109 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +import paddle.nn as nn +import paddle.nn.functional as F + +from ppdet.core.workspace import register, create, load_config +from ppdet.modeling import ops +from ppdet.utils.checkpoint import load_pretrain_weight +from ppdet.utils.logger import setup_logger + +logger = setup_logger(__name__) + + +class DistillModel(nn.Layer): + def __init__(self, cfg, slim_cfg): + super(DistillModel, self).__init__() + + self.student_model = create(cfg.architecture) + logger.debug('Load student model pretrain_weights:{}'.format( + cfg.pretrain_weights)) + load_pretrain_weight(self.student_model, cfg.pretrain_weights) + + slim_cfg = load_config(slim_cfg) + self.teacher_model = create(slim_cfg.architecture) + self.distill_loss = create(slim_cfg.distill_loss) + logger.debug('Load teacher model pretrain_weights:{}'.format( + slim_cfg.pretrain_weights)) + load_pretrain_weight(self.teacher_model, slim_cfg.pretrain_weights) + + for param in self.teacher_model.parameters(): + param.trainable = False + + def parameters(self): + return self.student_model.parameters() + + def forward(self, inputs): + if self.training: + teacher_loss = self.teacher_model(inputs) + student_loss = self.student_model(inputs) + loss = self.distill_loss(self.teacher_model, self.student_model) + student_loss['distill_loss'] = loss + student_loss['teacher_loss'] = teacher_loss['loss'] + student_loss['loss'] += student_loss['distill_loss'] + return student_loss + else: + return self.student_model(inputs) + + +@register +class DistillYOLOv3Loss(nn.Layer): + def __init__(self, weight=1000): + super(DistillYOLOv3Loss, self).__init__() + self.weight = weight + + def obj_weighted_reg(self, sx, sy, sw, sh, tx, ty, tw, th, tobj): + loss_x = ops.sigmoid_cross_entropy_with_logits(sx, F.sigmoid(tx)) + loss_y = ops.sigmoid_cross_entropy_with_logits(sy, F.sigmoid(ty)) + loss_w = paddle.abs(sw - tw) + loss_h = paddle.abs(sh - th) + loss = paddle.add_n([loss_x, loss_y, loss_w, loss_h]) + weighted_loss = paddle.mean(loss * F.sigmoid(tobj)) + return weighted_loss + + def obj_weighted_cls(self, scls, tcls, tobj): + loss = ops.sigmoid_cross_entropy_with_logits(scls, F.sigmoid(tcls)) + weighted_loss = paddle.mean(paddle.multiply(loss, F.sigmoid(tobj))) + return weighted_loss + + def obj_loss(self, sobj, tobj): + obj_mask = paddle.cast(tobj > 0., dtype="float32") + obj_mask.stop_gradient = True + loss = paddle.mean( + ops.sigmoid_cross_entropy_with_logits(sobj, obj_mask)) + return loss + + def forward(self, teacher_model, student_model): + teacher_distill_pairs = teacher_model.yolo_head.loss.distill_pairs + student_distill_pairs = student_model.yolo_head.loss.distill_pairs + distill_reg_loss, distill_cls_loss, distill_obj_loss = [], [], [] + for s_pair, t_pair in zip(student_distill_pairs, teacher_distill_pairs): + distill_reg_loss.append( + self.obj_weighted_reg(s_pair[0], s_pair[1], s_pair[2], s_pair[ + 3], t_pair[0], t_pair[1], t_pair[2], t_pair[3], t_pair[4])) + distill_cls_loss.append( + self.obj_weighted_cls(s_pair[5], t_pair[5], t_pair[4])) + distill_obj_loss.append(self.obj_loss(s_pair[4], t_pair[4])) + distill_reg_loss = paddle.add_n(distill_reg_loss) + distill_cls_loss = paddle.add_n(distill_cls_loss) + distill_obj_loss = paddle.add_n(distill_obj_loss) + loss = (distill_reg_loss + distill_cls_loss + distill_obj_loss + ) * self.weight + return loss diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/prune.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/prune.py new file mode 100644 index 000000000..70d3de369 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/prune.py @@ -0,0 +1,85 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import paddle +from paddle.utils import try_import + +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +def print_prune_params(model): + model_dict = model.state_dict() + for key in model_dict.keys(): + weight_name = model_dict[key].name + logger.info('Parameter name: {}, shape: {}'.format( + weight_name, model_dict[key].shape)) + + +@register +@serializable +class Pruner(object): + def __init__(self, + criterion, + pruned_params, + pruned_ratios, + print_params=False): + super(Pruner, self).__init__() + assert criterion in ['l1_norm', 'fpgm'], \ + "unsupported prune criterion: {}".format(criterion) + self.criterion = criterion + self.pruned_params = pruned_params + self.pruned_ratios = pruned_ratios + self.print_params = print_params + + def __call__(self, model): + # FIXME: adapt to network graph when Training and inference are + # inconsistent, now only supports prune inference network graph. + model.eval() + paddleslim = try_import('paddleslim') + from paddleslim.analysis import dygraph_flops as flops + input_spec = [{ + "image": paddle.ones( + shape=[1, 3, 640, 640], dtype='float32'), + "im_shape": paddle.full( + [1, 2], 640, dtype='float32'), + "scale_factor": paddle.ones( + shape=[1, 2], dtype='float32') + }] + if self.print_params: + print_prune_params(model) + + ori_flops = flops(model, input_spec) / (1000**3) + logger.info("FLOPs before pruning: {}GFLOPs".format(ori_flops)) + if self.criterion == 'fpgm': + pruner = paddleslim.dygraph.FPGMFilterPruner(model, input_spec) + elif self.criterion == 'l1_norm': + pruner = paddleslim.dygraph.L1NormFilterPruner(model, input_spec) + + logger.info("pruned params: {}".format(self.pruned_params)) + pruned_ratios = [float(n) for n in self.pruned_ratios] + ratios = {} + for i, param in enumerate(self.pruned_params): + ratios[param] = pruned_ratios[i] + pruner.prune_vars(ratios, [0]) + pruned_flops = flops(model, input_spec) / (1000**3) + logger.info("FLOPs after pruning: {}GFLOPs; pruned ratio: {}".format( + pruned_flops, (ori_flops - pruned_flops) / ori_flops)) + + return model diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/quant.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/quant.py new file mode 100644 index 000000000..ab81127ae --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/quant.py @@ -0,0 +1,84 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from paddle.utils import try_import + +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class QAT(object): + def __init__(self, quant_config, print_model): + super(QAT, self).__init__() + self.quant_config = quant_config + self.print_model = print_model + + def __call__(self, model): + paddleslim = try_import('paddleslim') + self.quanter = paddleslim.dygraph.quant.QAT(config=self.quant_config) + if self.print_model: + logger.info("Model before quant:") + logger.info(model) + + self.quanter.quantize(model) + + if self.print_model: + logger.info("Quantized model:") + logger.info(model) + + return model + + def save_quantized_model(self, layer, path, input_spec=None, **config): + self.quanter.save_quantized_model( + model=layer, path=path, input_spec=input_spec, **config) + + +@register +@serializable +class PTQ(object): + def __init__(self, + ptq_config, + quant_batch_num=10, + output_dir='output_inference', + fuse=True, + fuse_list=None): + super(PTQ, self).__init__() + self.ptq_config = ptq_config + self.quant_batch_num = quant_batch_num + self.output_dir = output_dir + self.fuse = fuse + self.fuse_list = fuse_list + + def __call__(self, model): + paddleslim = try_import('paddleslim') + self.ptq = paddleslim.PTQ(**self.ptq_config) + model.eval() + quant_model = self.ptq.quantize( + model, fuse=self.fuse, fuse_list=self.fuse_list) + + return quant_model + + def save_quantized_model(self, + quant_model, + quantize_model_path, + input_spec=None): + self.ptq.save_quantized_model(quant_model, quantize_model_path, + input_spec) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/unstructured_prune.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/unstructured_prune.py new file mode 100644 index 000000000..1dc876a8c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/slim/unstructured_prune.py @@ -0,0 +1,66 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +from paddle.utils import try_import + +from ppdet.core.workspace import register, serializable +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + + +@register +@serializable +class UnstructuredPruner(object): + def __init__(self, + stable_epochs, + pruning_epochs, + tunning_epochs, + pruning_steps, + ratio, + initial_ratio, + prune_params_type=None): + self.stable_epochs = stable_epochs + self.pruning_epochs = pruning_epochs + self.tunning_epochs = tunning_epochs + self.ratio = ratio + self.prune_params_type = prune_params_type + self.initial_ratio = initial_ratio + self.pruning_steps = pruning_steps + + def __call__(self, model, steps_per_epoch, skip_params_func=None): + paddleslim = try_import('paddleslim') + from paddleslim import GMPUnstructuredPruner + configs = { + 'pruning_strategy': 'gmp', + 'stable_iterations': self.stable_epochs * steps_per_epoch, + 'pruning_iterations': self.pruning_epochs * steps_per_epoch, + 'tunning_iterations': self.tunning_epochs * steps_per_epoch, + 'resume_iteration': 0, + 'pruning_steps': self.pruning_steps, + 'initial_ratio': self.initial_ratio, + } + + pruner = GMPUnstructuredPruner( + model, + ratio=self.ratio, + skip_params_func=skip_params_func, + prune_params_type=self.prune_params_type, + local_sparsity=True, + configs=configs) + + return pruner diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__init__.py new file mode 100644 index 000000000..d0c32e260 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__init__.py @@ -0,0 +1,13 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..da3643808 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/check.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/check.cpython-37.pyc new file mode 100644 index 000000000..31e998962 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/check.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/checkpoint.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/checkpoint.cpython-37.pyc new file mode 100644 index 000000000..353a47169 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/checkpoint.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/cli.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/cli.cpython-37.pyc new file mode 100644 index 000000000..4c184bf00 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/cli.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/colormap.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/colormap.cpython-37.pyc new file mode 100644 index 000000000..b6e9a2c48 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/colormap.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/download.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/download.cpython-37.pyc new file mode 100644 index 000000000..6e2c53fa4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/download.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/logger.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/logger.cpython-37.pyc new file mode 100644 index 000000000..6b63f47ed Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/logger.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/profiler.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/profiler.cpython-37.pyc new file mode 100644 index 000000000..7a5b0b855 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/profiler.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/stats.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/stats.cpython-37.pyc new file mode 100644 index 000000000..087a97fcf Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/stats.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/visualizer.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/visualizer.cpython-37.pyc new file mode 100644 index 000000000..44fe18ba3 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/visualizer.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/voc_utils.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/voc_utils.cpython-37.pyc new file mode 100644 index 000000000..5082f3f63 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/__pycache__/voc_utils.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/check.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/check.py new file mode 100644 index 000000000..6c795b532 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/check.py @@ -0,0 +1,112 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import sys + +import paddle +import six +import paddle.version as fluid_version + +from .logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['check_gpu', 'check_npu', 'check_version', 'check_config'] + + +def check_npu(use_npu): + """ + Log error and exit when set use_npu=true in paddlepaddle + cpu/gpu/xpu version. + """ + err = "Config use_npu cannot be set as true while you are " \ + "using paddlepaddle cpu/gpu/xpu version ! \nPlease try: \n" \ + "\t1. Install paddlepaddle-npu to run model on NPU \n" \ + "\t2. Set use_npu as false in config file to run " \ + "model on CPU/GPU/XPU" + + try: + if use_npu and not paddle.is_compiled_with_npu(): + logger.error(err) + sys.exit(1) + except Exception as e: + pass + + +def check_gpu(use_gpu): + """ + Log error and exit when set use_gpu=true in paddlepaddle + cpu version. + """ + err = "Config use_gpu cannot be set as true while you are " \ + "using paddlepaddle cpu version ! \nPlease try: \n" \ + "\t1. Install paddlepaddle-gpu to run model on GPU \n" \ + "\t2. Set use_gpu as false in config file to run " \ + "model on CPU" + + try: + if use_gpu and not paddle.is_compiled_with_cuda(): + logger.error(err) + sys.exit(1) + except Exception as e: + pass + + +def check_version(version='2.0'): + """ + Log error and exit when the installed version of paddlepaddle is + not satisfied. + """ + err = "PaddlePaddle version {} or higher is required, " \ + "or a suitable develop version is satisfied as well. \n" \ + "Please make sure the version is good with your code.".format(version) + + version_installed = [ + fluid_version.major, fluid_version.minor, fluid_version.patch, + fluid_version.rc + ] + if version_installed == ['0', '0', '0', '0']: + return + version_split = version.split('.') + + length = min(len(version_installed), len(version_split)) + for i in six.moves.range(length): + if version_installed[i] > version_split[i]: + return + if version_installed[i] < version_split[i]: + raise Exception(err) + + +def check_config(cfg): + """ + Check the correctness of the configuration file. Log error and exit + when Config is not compliant. + """ + err = "'{}' not specified in config file. Please set it in config file." + check_list = ['architecture', 'num_classes'] + try: + for var in check_list: + if not var in cfg: + logger.error(err.format(var)) + sys.exit(1) + except Exception as e: + pass + + if 'log_iter' not in cfg: + cfg.log_iter = 20 + + return cfg diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/checkpoint.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/checkpoint.py new file mode 100644 index 000000000..b5aa84697 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/checkpoint.py @@ -0,0 +1,226 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +import errno +import os +import time +import numpy as np +import paddle +import paddle.nn as nn +from .download import get_weights_path + +from .logger import setup_logger +logger = setup_logger(__name__) + + +def is_url(path): + """ + Whether path is URL. + Args: + path (string): URL string or not. + """ + return path.startswith('http://') \ + or path.startswith('https://') \ + or path.startswith('ppdet://') + + +def _get_unique_endpoints(trainer_endpoints): + # Sorting is to avoid different environmental variables for each card + trainer_endpoints.sort() + ips = set() + unique_endpoints = set() + for endpoint in trainer_endpoints: + ip = endpoint.split(":")[0] + if ip in ips: + continue + ips.add(ip) + unique_endpoints.add(endpoint) + logger.info("unique_endpoints {}".format(unique_endpoints)) + return unique_endpoints + + +def _strip_postfix(path): + path, ext = os.path.splitext(path) + assert ext in ['', '.pdparams', '.pdopt', '.pdmodel'], \ + "Unknown postfix {} from weights".format(ext) + return path + + +def load_weight(model, weight, optimizer=None): + if is_url(weight): + weight = get_weights_path(weight) + + path = _strip_postfix(weight) + pdparam_path = path + '.pdparams' + if not os.path.exists(pdparam_path): + raise ValueError("Model pretrain path {} does not " + "exists.".format(pdparam_path)) + + param_state_dict = paddle.load(pdparam_path) + model_dict = model.state_dict() + model_weight = {} + incorrect_keys = 0 + + for key in model_dict.keys(): + if key in param_state_dict.keys(): + model_weight[key] = param_state_dict[key] + else: + logger.info('Unmatched key: {}'.format(key)) + incorrect_keys += 1 + + assert incorrect_keys == 0, "Load weight {} incorrectly, \ + {} keys unmatched, please check again.".format(weight, + incorrect_keys) + logger.info('Finish resuming model weights: {}'.format(pdparam_path)) + + model.set_dict(model_weight) + + last_epoch = 0 + if optimizer is not None and os.path.exists(path + '.pdopt'): + optim_state_dict = paddle.load(path + '.pdopt') + # to solve resume bug, will it be fixed in paddle 2.0 + for key in optimizer.state_dict().keys(): + if not key in optim_state_dict.keys(): + optim_state_dict[key] = optimizer.state_dict()[key] + if 'last_epoch' in optim_state_dict: + last_epoch = optim_state_dict.pop('last_epoch') + optimizer.set_state_dict(optim_state_dict) + + return last_epoch + + +def match_state_dict(model_state_dict, weight_state_dict): + """ + Match between the model state dict and pretrained weight state dict. + Return the matched state dict. + + The method supposes that all the names in pretrained weight state dict are + subclass of the names in models`, if the prefix 'backbone.' in pretrained weight + keys is stripped. And we could get the candidates for each model key. Then we + select the name with the longest matched size as the final match result. For + example, the model state dict has the name of + 'backbone.res2.res2a.branch2a.conv.weight' and the pretrained weight as + name of 'res2.res2a.branch2a.conv.weight' and 'branch2a.conv.weight'. We + match the 'res2.res2a.branch2a.conv.weight' to the model key. + """ + + model_keys = sorted(model_state_dict.keys()) + weight_keys = sorted(weight_state_dict.keys()) + + def match(a, b): + if a.startswith('backbone.res5'): + # In Faster RCNN, res5 pretrained weights have prefix of backbone, + # however, the corresponding model weights have difficult prefix, + # bbox_head. + b = b[9:] + return a == b or a.endswith("." + b) + + match_matrix = np.zeros([len(model_keys), len(weight_keys)]) + for i, m_k in enumerate(model_keys): + for j, w_k in enumerate(weight_keys): + if match(m_k, w_k): + match_matrix[i, j] = len(w_k) + max_id = match_matrix.argmax(1) + max_len = match_matrix.max(1) + max_id[max_len == 0] = -1 + not_load_weight_name = [] + for match_idx in range(len(max_id)): + if match_idx < len(weight_keys) and max_id[match_idx] == -1: + not_load_weight_name.append(weight_keys[match_idx]) + if len(not_load_weight_name) > 0: + logger.info('{} in pretrained weight is not used in the model, ' + 'and its will not be loaded'.format(not_load_weight_name)) + matched_keys = {} + result_state_dict = {} + for model_id, weight_id in enumerate(max_id): + if weight_id == -1: + continue + model_key = model_keys[model_id] + weight_key = weight_keys[weight_id] + weight_value = weight_state_dict[weight_key] + model_value_shape = list(model_state_dict[model_key].shape) + + if list(weight_value.shape) != model_value_shape: + logger.info( + 'The shape {} in pretrained weight {} is unmatched with ' + 'the shape {} in model {}. And the weight {} will not be ' + 'loaded'.format(weight_value.shape, weight_key, + model_value_shape, model_key, weight_key)) + continue + + assert model_key not in result_state_dict + result_state_dict[model_key] = weight_value + if weight_key in matched_keys: + raise ValueError('Ambiguity weight {} loaded, it matches at least ' + '{} and {} in the model'.format( + weight_key, model_key, matched_keys[ + weight_key])) + matched_keys[weight_key] = model_key + return result_state_dict + + +def load_pretrain_weight(model, pretrain_weight): + if is_url(pretrain_weight): + pretrain_weight = get_weights_path(pretrain_weight) + + path = _strip_postfix(pretrain_weight) + if not (os.path.isdir(path) or os.path.isfile(path) or + os.path.exists(path + '.pdparams')): + raise ValueError("Model pretrain path `{}` does not exists. " + "If you don't want to load pretrain model, " + "please delete `pretrain_weights` field in " + "config file.".format(path)) + + model_dict = model.state_dict() + + weights_path = path + '.pdparams' + param_state_dict = paddle.load(weights_path) + param_state_dict = match_state_dict(model_dict, param_state_dict) + + model.set_dict(param_state_dict) + logger.info('Finish loading model weights: {}'.format(weights_path)) + + +def save_model(model, optimizer, save_dir, save_name, last_epoch): + """ + save model into disk. + + Args: + model (paddle.nn.Layer): the Layer instalce to save parameters. + optimizer (paddle.optimizer.Optimizer): the Optimizer instance to + save optimizer states. + save_dir (str): the directory to be saved. + save_name (str): the path to be saved. + last_epoch (int): the epoch index. + """ + if paddle.distributed.get_rank() != 0: + return + if not os.path.exists(save_dir): + os.makedirs(save_dir) + save_path = os.path.join(save_dir, save_name) + if isinstance(model, nn.Layer): + paddle.save(model.state_dict(), save_path + ".pdparams") + else: + assert isinstance(model, + dict), 'model is not a instance of nn.layer or dict' + paddle.save(model, save_path + ".pdparams") + state_dict = optimizer.state_dict() + state_dict['last_epoch'] = last_epoch + paddle.save(state_dict, save_path + ".pdopt") + logger.info("Save checkpoint: {}".format(save_dir)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/cli.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/cli.py new file mode 100644 index 000000000..b8ba59d78 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/cli.py @@ -0,0 +1,151 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from argparse import ArgumentParser, RawDescriptionHelpFormatter + +import yaml +import re +from ppdet.core.workspace import get_registered_modules, dump_value + +__all__ = ['ColorTTY', 'ArgsParser'] + + +class ColorTTY(object): + def __init__(self): + super(ColorTTY, self).__init__() + self.colors = ['red', 'green', 'yellow', 'blue', 'magenta', 'cyan'] + + def __getattr__(self, attr): + if attr in self.colors: + color = self.colors.index(attr) + 31 + + def color_message(message): + return "[{}m{}".format(color, message) + + setattr(self, attr, color_message) + return color_message + + def bold(self, message): + return self.with_code('01', message) + + def with_code(self, code, message): + return "[{}m{}".format(code, message) + + +class ArgsParser(ArgumentParser): + def __init__(self): + super(ArgsParser, self).__init__( + formatter_class=RawDescriptionHelpFormatter) + self.add_argument("-c", "--config", help="configuration file to use") + self.add_argument( + "-o", "--opt", nargs='*', help="set configuration options") + + def parse_args(self, argv=None): + args = super(ArgsParser, self).parse_args(argv) + assert args.config is not None, \ + "Please specify --config=configure_file_path." + args.opt = self._parse_opt(args.opt) + return args + + def _parse_opt(self, opts): + config = {} + if not opts: + return config + for s in opts: + s = s.strip() + k, v = s.split('=', 1) + if '.' not in k: + config[k] = yaml.load(v, Loader=yaml.Loader) + else: + keys = k.split('.') + if keys[0] not in config: + config[keys[0]] = {} + cur = config[keys[0]] + for idx, key in enumerate(keys[1:]): + if idx == len(keys) - 2: + cur[key] = yaml.load(v, Loader=yaml.Loader) + else: + cur[key] = {} + cur = cur[key] + return config + + +def print_total_cfg(config): + modules = get_registered_modules() + color_tty = ColorTTY() + green = '___{}___'.format(color_tty.colors.index('green') + 31) + + styled = {} + for key in config.keys(): + if not config[key]: # empty schema + continue + + if key not in modules and not hasattr(config[key], '__dict__'): + styled[key] = config[key] + continue + elif key in modules: + module = modules[key] + else: + type_name = type(config[key]).__name__ + if type_name in modules: + module = modules[type_name].copy() + module.update({ + k: v + for k, v in config[key].__dict__.items() + if k in module.schema + }) + key += " ({})".format(type_name) + default = module.find_default_keys() + missing = module.find_missing_keys() + mismatch = module.find_mismatch_keys() + extra = module.find_extra_keys() + dep_missing = [] + for dep in module.inject: + if isinstance(module[dep], str) and module[dep] != '': + if module[dep] not in modules: # not a valid module + dep_missing.append(dep) + else: + dep_mod = modules[module[dep]] + # empty dict but mandatory + if not dep_mod and dep_mod.mandatory(): + dep_missing.append(dep) + override = list( + set(module.keys()) - set(default) - set(extra) - set(dep_missing)) + replacement = {} + for name in set(override + default + extra + mismatch + missing): + new_name = name + if name in missing: + value = "" + else: + value = module[name] + + if name in extra: + value = dump_value(value) + " " + elif name in mismatch: + value = dump_value(value) + " " + elif name in dep_missing: + value = dump_value(value) + " " + elif name in override and value != '': + mark = green + new_name = mark + name + replacement[new_name] = value + styled[key] = replacement + buffer = yaml.dump(styled, default_flow_style=False, default_style='') + buffer = (re.sub(r"", r"[31m[0m", buffer)) + buffer = (re.sub(r"", r"[33m[0m", buffer)) + buffer = (re.sub(r"", r"[31m[0m", buffer)) + buffer = (re.sub(r"", + r"[31m[0m", buffer)) + buffer = re.sub(r"___(\d+)___(.*?):", r"[\1m\2[0m:", buffer) + print(buffer) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/colormap.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/colormap.py new file mode 100644 index 000000000..a9cdbe891 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/colormap.py @@ -0,0 +1,58 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +import numpy as np + + +def colormap(rgb=False): + """ + Get colormap + + The code of this function is copied from https://github.com/facebookresearch/Detectron/blob/main/detectron/utils/colormap.py + """ + color_list = np.array([ + 0.000, 0.447, 0.741, 0.850, 0.325, 0.098, 0.929, 0.694, 0.125, 0.494, + 0.184, 0.556, 0.466, 0.674, 0.188, 0.301, 0.745, 0.933, 0.635, 0.078, + 0.184, 0.300, 0.300, 0.300, 0.600, 0.600, 0.600, 1.000, 0.000, 0.000, + 1.000, 0.500, 0.000, 0.749, 0.749, 0.000, 0.000, 1.000, 0.000, 0.000, + 0.000, 1.000, 0.667, 0.000, 1.000, 0.333, 0.333, 0.000, 0.333, 0.667, + 0.000, 0.333, 1.000, 0.000, 0.667, 0.333, 0.000, 0.667, 0.667, 0.000, + 0.667, 1.000, 0.000, 1.000, 0.333, 0.000, 1.000, 0.667, 0.000, 1.000, + 1.000, 0.000, 0.000, 0.333, 0.500, 0.000, 0.667, 0.500, 0.000, 1.000, + 0.500, 0.333, 0.000, 0.500, 0.333, 0.333, 0.500, 0.333, 0.667, 0.500, + 0.333, 1.000, 0.500, 0.667, 0.000, 0.500, 0.667, 0.333, 0.500, 0.667, + 0.667, 0.500, 0.667, 1.000, 0.500, 1.000, 0.000, 0.500, 1.000, 0.333, + 0.500, 1.000, 0.667, 0.500, 1.000, 1.000, 0.500, 0.000, 0.333, 1.000, + 0.000, 0.667, 1.000, 0.000, 1.000, 1.000, 0.333, 0.000, 1.000, 0.333, + 0.333, 1.000, 0.333, 0.667, 1.000, 0.333, 1.000, 1.000, 0.667, 0.000, + 1.000, 0.667, 0.333, 1.000, 0.667, 0.667, 1.000, 0.667, 1.000, 1.000, + 1.000, 0.000, 1.000, 1.000, 0.333, 1.000, 1.000, 0.667, 1.000, 0.167, + 0.000, 0.000, 0.333, 0.000, 0.000, 0.500, 0.000, 0.000, 0.667, 0.000, + 0.000, 0.833, 0.000, 0.000, 1.000, 0.000, 0.000, 0.000, 0.167, 0.000, + 0.000, 0.333, 0.000, 0.000, 0.500, 0.000, 0.000, 0.667, 0.000, 0.000, + 0.833, 0.000, 0.000, 1.000, 0.000, 0.000, 0.000, 0.167, 0.000, 0.000, + 0.333, 0.000, 0.000, 0.500, 0.000, 0.000, 0.667, 0.000, 0.000, 0.833, + 0.000, 0.000, 1.000, 0.000, 0.000, 0.000, 0.143, 0.143, 0.143, 0.286, + 0.286, 0.286, 0.429, 0.429, 0.429, 0.571, 0.571, 0.571, 0.714, 0.714, + 0.714, 0.857, 0.857, 0.857, 1.000, 1.000, 1.000 + ]).astype(np.float32) + color_list = color_list.reshape((-1, 3)) * 255 + if not rgb: + color_list = color_list[:, ::-1] + return color_list diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/download.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/download.py new file mode 100644 index 000000000..54c19c6cf --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/download.py @@ -0,0 +1,557 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import os.path as osp +import sys +import yaml +import time +import shutil +import requests +import tqdm +import hashlib +import base64 +import binascii +import tarfile +import zipfile + +from paddle.utils.download import _get_unique_endpoints +from ppdet.core.workspace import BASE_KEY +from .logger import setup_logger +from .voc_utils import create_list + +logger = setup_logger(__name__) + +__all__ = [ + 'get_weights_path', 'get_dataset_path', 'get_config_path', + 'download_dataset', 'create_voc_list' +] + +WEIGHTS_HOME = osp.expanduser("~/.cache/paddle/weights") +DATASET_HOME = osp.expanduser("~/.cache/paddle/dataset") +CONFIGS_HOME = osp.expanduser("~/.cache/paddle/configs") + +# dict of {dataset_name: (download_info, sub_dirs)} +# download info: [(url, md5sum)] +DATASETS = { + 'coco': ([ + ( + 'http://images.cocodataset.org/zips/train2017.zip', + 'cced6f7f71b7629ddf16f17bbcfab6b2', ), + ( + 'http://images.cocodataset.org/zips/val2017.zip', + '442b8da7639aecaf257c1dceb8ba8c80', ), + ( + 'http://images.cocodataset.org/annotations/annotations_trainval2017.zip', + 'f4bbac642086de4f52a3fdda2de5fa2c', ), + ], ["annotations", "train2017", "val2017"]), + 'voc': ([ + ( + 'http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar', + '6cd6e144f989b92b3379bac3b3de84fd', ), + ( + 'http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtrainval_06-Nov-2007.tar', + 'c52e279531787c972589f7e41ab4ae64', ), + ( + 'http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtest_06-Nov-2007.tar', + 'b6e924de25625d8de591ea690078ad9f', ), + ( + 'https://paddledet.bj.bcebos.com/data/label_list.txt', + '5ae5d62183cfb6f6d3ac109359d06a1b', ), + ], ["VOCdevkit/VOC2012", "VOCdevkit/VOC2007"]), + 'wider_face': ([ + ( + 'https://dataset.bj.bcebos.com/wider_face/WIDER_train.zip', + '3fedf70df600953d25982bcd13d91ba2', ), + ( + 'https://dataset.bj.bcebos.com/wider_face/WIDER_val.zip', + 'dfa7d7e790efa35df3788964cf0bbaea', ), + ( + 'https://dataset.bj.bcebos.com/wider_face/wider_face_split.zip', + 'a4a898d6193db4b9ef3260a68bad0dc7', ), + ], ["WIDER_train", "WIDER_val", "wider_face_split"]), + 'fruit': ([( + 'https://dataset.bj.bcebos.com/PaddleDetection_demo/fruit.tar', + 'baa8806617a54ccf3685fa7153388ae6', ), ], + ['Annotations', 'JPEGImages']), + 'roadsign_voc': ([( + 'https://paddlemodels.bj.bcebos.com/object_detection/roadsign_voc.tar', + '8d629c0f880dd8b48de9aeff44bf1f3e', ), ], ['annotations', 'images']), + 'roadsign_coco': ([( + 'https://paddlemodels.bj.bcebos.com/object_detection/roadsign_coco.tar', + '49ce5a9b5ad0d6266163cd01de4b018e', ), ], ['annotations', 'images']), + 'spine_coco': ([( + 'https://paddledet.bj.bcebos.com/data/spine_coco.tar', + '7ed69ae73f842cd2a8cf4f58dc3c5535', ), ], ['annotations', 'images']), + 'mot': (), + 'objects365': (), + 'coco_ce': ([( + 'https://paddledet.bj.bcebos.com/data/coco_ce.tar', + 'eadd1b79bc2f069f2744b1dd4e0c0329', ), ], []) +} + +DOWNLOAD_RETRY_LIMIT = 3 + +PPDET_WEIGHTS_DOWNLOAD_URL_PREFIX = 'https://paddledet.bj.bcebos.com/' + + +def parse_url(url): + url = url.replace("ppdet://", PPDET_WEIGHTS_DOWNLOAD_URL_PREFIX) + return url + + +def get_weights_path(url): + """Get weights path from WEIGHTS_HOME, if not exists, + download it from url. + """ + url = parse_url(url) + path, _ = get_path(url, WEIGHTS_HOME) + return path + + +def get_config_path(url): + """Get weights path from CONFIGS_HOME, if not exists, + download it from url. + """ + url = parse_url(url) + path = map_path(url, CONFIGS_HOME, path_depth=2) + if os.path.isfile(path): + return path + + # config file not found, try download + # 1. clear configs directory + if osp.isdir(CONFIGS_HOME): + shutil.rmtree(CONFIGS_HOME) + + # 2. get url + try: + from ppdet import __version__ as version + except ImportError: + version = None + + cfg_url = "ppdet://configs/{}/configs.tar".format(version) \ + if version else "ppdet://configs/configs.tar" + cfg_url = parse_url(cfg_url) + + # 3. download and decompress + cfg_fullname = _download_dist(cfg_url, osp.dirname(CONFIGS_HOME)) + _decompress_dist(cfg_fullname) + + # 4. check config file existing + if os.path.isfile(path): + return path + else: + logger.error("Get config {} failed after download, please contact us on " \ + "https://github.com/PaddlePaddle/PaddleDetection/issues".format(path)) + sys.exit(1) + + +def get_dataset_path(path, annotation, image_dir): + """ + If path exists, return path. + Otherwise, get dataset path from DATASET_HOME, if not exists, + download it. + """ + if _dataset_exists(path, annotation, image_dir): + return path + + logger.info("Dataset {} is not valid for reason above, try searching {} or " + "downloading dataset...".format( + osp.realpath(path), DATASET_HOME)) + + data_name = os.path.split(path.strip().lower())[-1] + for name, dataset in DATASETS.items(): + if data_name == name: + logger.debug("Parse dataset_dir {} as dataset " + "{}".format(path, name)) + if name == 'objects365': + raise NotImplementedError( + "Dataset {} is not valid for download automatically. " + "Please apply and download the dataset from " + "https://www.objects365.org/download.html".format(name)) + data_dir = osp.join(DATASET_HOME, name) + + if name == 'mot': + if osp.exists(path) or osp.exists(data_dir): + return data_dir + else: + raise NotImplementedError( + "Dataset {} is not valid for download automatically. " + "Please apply and download the dataset following docs/tutorials/PrepareMOTDataSet.md". + format(name)) + + if name == "spine_coco": + if _dataset_exists(data_dir, annotation, image_dir): + return data_dir + + # For voc, only check dir VOCdevkit/VOC2012, VOCdevkit/VOC2007 + if name in ['voc', 'fruit', 'roadsign_voc']: + exists = True + for sub_dir in dataset[1]: + check_dir = osp.join(data_dir, sub_dir) + if osp.exists(check_dir): + logger.info("Found {}".format(check_dir)) + else: + exists = False + if exists: + return data_dir + + # voc exist is checked above, voc is not exist here + check_exist = name != 'voc' and name != 'fruit' and name != 'roadsign_voc' + for url, md5sum in dataset[0]: + get_path(url, data_dir, md5sum, check_exist) + + # voc should create list after download + if name == 'voc': + create_voc_list(data_dir) + return data_dir + + # not match any dataset in DATASETS + raise ValueError( + "Dataset {} is not valid and cannot parse dataset type " + "'{}' for automaticly downloading, which only supports " + "'voc' , 'coco', 'wider_face', 'fruit', 'roadsign_voc' and 'mot' currently". + format(path, osp.split(path)[-1])) + + +def create_voc_list(data_dir, devkit_subdir='VOCdevkit'): + logger.debug("Create voc file list...") + devkit_dir = osp.join(data_dir, devkit_subdir) + years = ['2007', '2012'] + + # NOTE: since using auto download VOC + # dataset, VOC default label list should be used, + # do not generate label_list.txt here. For default + # label, see ../data/source/voc.py + create_list(devkit_dir, years, data_dir) + logger.debug("Create voc file list finished") + + +def map_path(url, root_dir, path_depth=1): + # parse path after download to decompress under root_dir + assert path_depth > 0, "path_depth should be a positive integer" + dirname = url + for _ in range(path_depth): + dirname = osp.dirname(dirname) + fpath = osp.relpath(url, dirname) + + zip_formats = ['.zip', '.tar', '.gz'] + for zip_format in zip_formats: + fpath = fpath.replace(zip_format, '') + return osp.join(root_dir, fpath) + + +def get_path(url, root_dir, md5sum=None, check_exist=True): + """ Download from given url to root_dir. + if file or directory specified by url is exists under + root_dir, return the path directly, otherwise download + from url and decompress it, return the path. + + url (str): download url + root_dir (str): root dir for downloading, it should be + WEIGHTS_HOME or DATASET_HOME + md5sum (str): md5 sum of download package + """ + # parse path after download to decompress under root_dir + fullpath = map_path(url, root_dir) + + # For same zip file, decompressed directory name different + # from zip file name, rename by following map + decompress_name_map = { + "VOCtrainval_11-May-2012": "VOCdevkit/VOC2012", + "VOCtrainval_06-Nov-2007": "VOCdevkit/VOC2007", + "VOCtest_06-Nov-2007": "VOCdevkit/VOC2007", + "annotations_trainval": "annotations" + } + for k, v in decompress_name_map.items(): + if fullpath.find(k) >= 0: + fullpath = osp.join(osp.split(fullpath)[0], v) + + if osp.exists(fullpath) and check_exist: + if not osp.isfile(fullpath) or \ + _check_exist_file_md5(fullpath, md5sum, url): + logger.debug("Found {}".format(fullpath)) + return fullpath, True + else: + os.remove(fullpath) + + fullname = _download_dist(url, root_dir, md5sum) + + # new weights format which postfix is 'pdparams' not + # need to decompress + if osp.splitext(fullname)[-1] not in ['.pdparams', '.yml']: + _decompress_dist(fullname) + + return fullpath, False + + +def download_dataset(path, dataset=None): + if dataset not in DATASETS.keys(): + logger.error("Unknown dataset {}, it should be " + "{}".format(dataset, DATASETS.keys())) + return + dataset_info = DATASETS[dataset][0] + for info in dataset_info: + get_path(info[0], path, info[1], False) + logger.debug("Download dataset {} finished.".format(dataset)) + + +def _dataset_exists(path, annotation, image_dir): + """ + Check if user define dataset exists + """ + if not osp.exists(path): + logger.warning("Config dataset_dir {} is not exits, " + "dataset config is not valid".format(path)) + return False + if annotation: + annotation_path = osp.join(path, annotation) + if not osp.isfile(annotation_path): + logger.warning("Config annotation {} is not a " + "file, dataset config is not " + "valid".format(annotation_path)) + return False + if image_dir: + image_path = osp.join(path, image_dir) + if not osp.isdir(image_path): + logger.warning("Config image_dir {} is not a " + "directory, dataset config is not " + "valid".format(image_path)) + return False + return True + + +def _download(url, path, md5sum=None): + """ + Download from url, save to path. + + url (str): download url + path (str): download to given path + """ + if not osp.exists(path): + os.makedirs(path) + + fname = osp.split(url)[-1] + fullname = osp.join(path, fname) + retry_cnt = 0 + + while not (osp.exists(fullname) and _check_exist_file_md5(fullname, md5sum, + url)): + if retry_cnt < DOWNLOAD_RETRY_LIMIT: + retry_cnt += 1 + else: + raise RuntimeError("Download from {} failed. " + "Retry limit reached".format(url)) + + logger.info("Downloading {} from {}".format(fname, url)) + + # NOTE: windows path join may incur \, which is invalid in url + if sys.platform == "win32": + url = url.replace('\\', '/') + + req = requests.get(url, stream=True) + if req.status_code != 200: + raise RuntimeError("Downloading from {} failed with code " + "{}!".format(url, req.status_code)) + + # For protecting download interupted, download to + # tmp_fullname firstly, move tmp_fullname to fullname + # after download finished + tmp_fullname = fullname + "_tmp" + total_size = req.headers.get('content-length') + with open(tmp_fullname, 'wb') as f: + if total_size: + for chunk in tqdm.tqdm( + req.iter_content(chunk_size=1024), + total=(int(total_size) + 1023) // 1024, + unit='KB'): + f.write(chunk) + else: + for chunk in req.iter_content(chunk_size=1024): + if chunk: + f.write(chunk) + shutil.move(tmp_fullname, fullname) + return fullname + + +def _download_dist(url, path, md5sum=None): + env = os.environ + if 'PADDLE_TRAINERS_NUM' in env and 'PADDLE_TRAINER_ID' in env: + trainer_id = int(env['PADDLE_TRAINER_ID']) + num_trainers = int(env['PADDLE_TRAINERS_NUM']) + if num_trainers <= 1: + return _download(url, path, md5sum) + else: + fname = osp.split(url)[-1] + fullname = osp.join(path, fname) + lock_path = fullname + '.download.lock' + + if not osp.isdir(path): + os.makedirs(path) + + if not osp.exists(fullname): + from paddle.distributed import ParallelEnv + unique_endpoints = _get_unique_endpoints(ParallelEnv() + .trainer_endpoints[:]) + with open(lock_path, 'w'): # touch + os.utime(lock_path, None) + if ParallelEnv().current_endpoint in unique_endpoints: + _download(url, path, md5sum) + os.remove(lock_path) + else: + while os.path.exists(lock_path): + time.sleep(0.5) + return fullname + else: + return _download(url, path, md5sum) + + +def _check_exist_file_md5(filename, md5sum, url): + # if md5sum is None, and file to check is weights file, + # read md5um from url and check, else check md5sum directly + return _md5check_from_url(filename, url) if md5sum is None \ + and filename.endswith('pdparams') \ + else _md5check(filename, md5sum) + + +def _md5check_from_url(filename, url): + # For weights in bcebos URLs, MD5 value is contained + # in request header as 'content_md5' + req = requests.get(url, stream=True) + content_md5 = req.headers.get('content-md5') + req.close() + if not content_md5 or _md5check( + filename, + binascii.hexlify(base64.b64decode(content_md5.strip('"'))).decode( + )): + return True + else: + return False + + +def _md5check(fullname, md5sum=None): + if md5sum is None: + return True + + logger.debug("File {} md5 checking...".format(fullname)) + md5 = hashlib.md5() + with open(fullname, 'rb') as f: + for chunk in iter(lambda: f.read(4096), b""): + md5.update(chunk) + calc_md5sum = md5.hexdigest() + + if calc_md5sum != md5sum: + logger.warning("File {} md5 check failed, {}(calc) != " + "{}(base)".format(fullname, calc_md5sum, md5sum)) + return False + return True + + +def _decompress(fname): + """ + Decompress for zip and tar file + """ + logger.info("Decompressing {}...".format(fname)) + + # For protecting decompressing interupted, + # decompress to fpath_tmp directory firstly, if decompress + # successed, move decompress files to fpath and delete + # fpath_tmp and remove download compress file. + fpath = osp.split(fname)[0] + fpath_tmp = osp.join(fpath, 'tmp') + if osp.isdir(fpath_tmp): + shutil.rmtree(fpath_tmp) + os.makedirs(fpath_tmp) + + if fname.find('tar') >= 0: + with tarfile.open(fname) as tf: + tf.extractall(path=fpath_tmp) + elif fname.find('zip') >= 0: + with zipfile.ZipFile(fname) as zf: + zf.extractall(path=fpath_tmp) + elif fname.find('.txt') >= 0: + return + else: + raise TypeError("Unsupport compress file type {}".format(fname)) + + for f in os.listdir(fpath_tmp): + src_dir = osp.join(fpath_tmp, f) + dst_dir = osp.join(fpath, f) + _move_and_merge_tree(src_dir, dst_dir) + + shutil.rmtree(fpath_tmp) + os.remove(fname) + + +def _decompress_dist(fname): + env = os.environ + if 'PADDLE_TRAINERS_NUM' in env and 'PADDLE_TRAINER_ID' in env: + trainer_id = int(env['PADDLE_TRAINER_ID']) + num_trainers = int(env['PADDLE_TRAINERS_NUM']) + if num_trainers <= 1: + _decompress(fname) + else: + lock_path = fname + '.decompress.lock' + from paddle.distributed import ParallelEnv + unique_endpoints = _get_unique_endpoints(ParallelEnv() + .trainer_endpoints[:]) + # NOTE(dkp): _decompress_dist always performed after + # _download_dist, in _download_dist sub-trainers is waiting + # for download lock file release with sleeping, if decompress + # prograss is very fast and finished with in the sleeping gap + # time, e.g in tiny dataset such as coco_ce, spine_coco, main + # trainer may finish decompress and release lock file, so we + # only craete lock file in main trainer and all sub-trainer + # wait 1s for main trainer to create lock file, for 1s is + # twice as sleeping gap, this waiting time can keep all + # trainer pipeline in order + # **change this if you have more elegent methods** + if ParallelEnv().current_endpoint in unique_endpoints: + with open(lock_path, 'w'): # touch + os.utime(lock_path, None) + _decompress(fname) + os.remove(lock_path) + else: + time.sleep(1) + while os.path.exists(lock_path): + time.sleep(0.5) + else: + _decompress(fname) + + +def _move_and_merge_tree(src, dst): + """ + Move src directory to dst, if dst is already exists, + merge src to dst + """ + if not osp.exists(dst): + shutil.move(src, dst) + elif osp.isfile(src): + shutil.move(src, dst) + else: + for fp in os.listdir(src): + src_fp = osp.join(src, fp) + dst_fp = osp.join(dst, fp) + if osp.isdir(src_fp): + if osp.isdir(dst_fp): + _move_and_merge_tree(src_fp, dst_fp) + else: + shutil.move(src_fp, dst_fp) + elif osp.isfile(src_fp) and \ + not osp.isfile(dst_fp): + shutil.move(src_fp, dst_fp) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/logger.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/logger.py new file mode 100644 index 000000000..51e296205 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/logger.py @@ -0,0 +1,70 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import os +import sys + +import paddle.distributed as dist + +__all__ = ['setup_logger'] + +logger_initialized = [] + + +def setup_logger(name="ppdet", output=None): + """ + Initialize logger and set its verbosity level to INFO. + Args: + output (str): a file name or a directory to save log. If None, will not save log file. + If ends with ".txt" or ".log", assumed to be a file name. + Otherwise, logs will be saved to `output/log.txt`. + name (str): the root module name of this logger + + Returns: + logging.Logger: a logger + """ + logger = logging.getLogger(name) + if name in logger_initialized: + return logger + + logger.setLevel(logging.INFO) + logger.propagate = False + + formatter = logging.Formatter( + "[%(asctime)s] %(name)s %(levelname)s: %(message)s", + datefmt="%m/%d %H:%M:%S") + # stdout logging: master only + local_rank = dist.get_rank() + if local_rank == 0: + ch = logging.StreamHandler(stream=sys.stdout) + ch.setLevel(logging.DEBUG) + ch.setFormatter(formatter) + logger.addHandler(ch) + + # file logging: all workers + if output is not None: + if output.endswith(".txt") or output.endswith(".log"): + filename = output + else: + filename = os.path.join(output, "log.txt") + if local_rank > 0: + filename = filename + ".rank{}".format(local_rank) + os.makedirs(os.path.dirname(filename)) + fh = logging.FileHandler(filename, mode='a') + fh.setLevel(logging.DEBUG) + fh.setFormatter(logging.Formatter()) + logger.addHandler(fh) + logger_initialized.append(name) + return logger diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/profiler.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/profiler.py new file mode 100644 index 000000000..cae3773fa --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/profiler.py @@ -0,0 +1,111 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import sys +import paddle + +# A global variable to record the number of calling times for profiler +# functions. It is used to specify the tracing range of training steps. +_profiler_step_id = 0 + +# A global variable to avoid parsing from string every time. +_profiler_options = None + + +class ProfilerOptions(object): + ''' + Use a string to initialize a ProfilerOptions. + The string should be in the format: "key1=value1;key2=value;key3=value3". + For example: + "profile_path=model.profile" + "batch_range=[50, 60]; profile_path=model.profile" + "batch_range=[50, 60]; tracer_option=OpDetail; profile_path=model.profile" + + ProfilerOptions supports following key-value pair: + batch_range - a integer list, e.g. [100, 110]. + state - a string, the optional values are 'CPU', 'GPU' or 'All'. + sorted_key - a string, the optional values are 'calls', 'total', + 'max', 'min' or 'ave. + tracer_option - a string, the optional values are 'Default', 'OpDetail', + 'AllOpDetail'. + profile_path - a string, the path to save the serialized profile data, + which can be used to generate a timeline. + exit_on_finished - a boolean. + ''' + + def __init__(self, options_str): + assert isinstance(options_str, str) + + self._options = { + 'batch_range': [10, 20], + 'state': 'All', + 'sorted_key': 'total', + 'tracer_option': 'Default', + 'profile_path': '/tmp/profile', + 'exit_on_finished': True + } + self._parse_from_string(options_str) + + def _parse_from_string(self, options_str): + for kv in options_str.replace(' ', '').split(';'): + key, value = kv.split('=') + if key == 'batch_range': + value_list = value.replace('[', '').replace(']', '').split(',') + value_list = list(map(int, value_list)) + if len(value_list) >= 2 and value_list[0] >= 0 and value_list[ + 1] > value_list[0]: + self._options[key] = value_list + elif key == 'exit_on_finished': + self._options[key] = value.lower() in ("yes", "true", "t", "1") + elif key in [ + 'state', 'sorted_key', 'tracer_option', 'profile_path' + ]: + self._options[key] = value + + def __getitem__(self, name): + if self._options.get(name, None) is None: + raise ValueError( + "ProfilerOptions does not have an option named %s." % name) + return self._options[name] + + +def add_profiler_step(options_str=None): + ''' + Enable the operator-level timing using PaddlePaddle's profiler. + The profiler uses a independent variable to count the profiler steps. + One call of this function is treated as a profiler step. + + Args: + profiler_options - a string to initialize the ProfilerOptions. + Default is None, and the profiler is disabled. + ''' + if options_str is None: + return + + global _profiler_step_id + global _profiler_options + + if _profiler_options is None: + _profiler_options = ProfilerOptions(options_str) + + if _profiler_step_id == _profiler_options['batch_range'][0]: + paddle.utils.profiler.start_profiler(_profiler_options['state'], + _profiler_options['tracer_option']) + elif _profiler_step_id == _profiler_options['batch_range'][1]: + paddle.utils.profiler.stop_profiler(_profiler_options['sorted_key'], + _profiler_options['profile_path']) + if _profiler_options['exit_on_finished']: + sys.exit(0) + + _profiler_step_id += 1 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/stats.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/stats.py new file mode 100644 index 000000000..4cd36d91c --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/stats.py @@ -0,0 +1,94 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import collections +import numpy as np + +__all__ = ['SmoothedValue', 'TrainingStats'] + + +class SmoothedValue(object): + """Track a series of values and provide access to smoothed values over a + window or the global series average. + """ + + def __init__(self, window_size=20, fmt=None): + if fmt is None: + fmt = "{median:.4f} ({avg:.4f})" + self.deque = collections.deque(maxlen=window_size) + self.fmt = fmt + self.total = 0. + self.count = 0 + + def update(self, value, n=1): + self.deque.append(value) + self.count += n + self.total += value * n + + @property + def median(self): + return np.median(self.deque) + + @property + def avg(self): + return np.mean(self.deque) + + @property + def max(self): + return np.max(self.deque) + + @property + def value(self): + return self.deque[-1] + + @property + def global_avg(self): + return self.total / self.count + + def __str__(self): + return self.fmt.format( + median=self.median, avg=self.avg, max=self.max, value=self.value) + + +class TrainingStats(object): + def __init__(self, window_size, delimiter=' '): + self.meters = None + self.window_size = window_size + self.delimiter = delimiter + + def update(self, stats): + if self.meters is None: + self.meters = { + k: SmoothedValue(self.window_size) + for k in stats.keys() + } + for k, v in self.meters.items(): + v.update(stats[k].numpy()) + + def get(self, extras=None): + stats = collections.OrderedDict() + if extras: + for k, v in extras.items(): + stats[k] = v + for k, v in self.meters.items(): + stats[k] = format(v.median, '.6f') + + return stats + + def log(self, extras=None): + d = self.get(extras) + strs = [] + for k, v in d.items(): + strs.append("{}: {}".format(k, str(v))) + return self.delimiter.join(strs) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/visualizer.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/visualizer.py new file mode 100644 index 000000000..fdfd966e2 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/visualizer.py @@ -0,0 +1,321 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +import numpy as np +from PIL import Image, ImageDraw +import cv2 +import math + +from .colormap import colormap +from ppdet.utils.logger import setup_logger +logger = setup_logger(__name__) + +__all__ = ['visualize_results'] + + +def visualize_results(image, + bbox_res, + mask_res, + segm_res, + keypoint_res, + im_id, + catid2name, + threshold=0.5): + """ + Visualize bbox and mask results + """ + if bbox_res is not None: + image = draw_bbox(image, im_id, catid2name, bbox_res, threshold) + if mask_res is not None: + image = draw_mask(image, im_id, mask_res, threshold) + if segm_res is not None: + image = draw_segm(image, im_id, catid2name, segm_res, threshold) + if keypoint_res is not None: + image = draw_pose(image, keypoint_res, threshold) + return image + + +def draw_mask(image, im_id, segms, threshold, alpha=0.7): + """ + Draw mask on image + """ + mask_color_id = 0 + w_ratio = .4 + color_list = colormap(rgb=True) + img_array = np.array(image).astype('float32') + for dt in np.array(segms): + if im_id != dt['image_id']: + continue + segm, score = dt['segmentation'], dt['score'] + if score < threshold: + continue + import pycocotools.mask as mask_util + mask = mask_util.decode(segm) * 255 + color_mask = color_list[mask_color_id % len(color_list), 0:3] + mask_color_id += 1 + for c in range(3): + color_mask[c] = color_mask[c] * (1 - w_ratio) + w_ratio * 255 + idx = np.nonzero(mask) + img_array[idx[0], idx[1], :] *= 1.0 - alpha + img_array[idx[0], idx[1], :] += alpha * color_mask + return Image.fromarray(img_array.astype('uint8')) + + +def draw_bbox(image, im_id, catid2name, bboxes, threshold): + """ + Draw bbox on image + """ + draw = ImageDraw.Draw(image) + + catid2color = {} + color_list = colormap(rgb=True)[:40] + for dt in np.array(bboxes): + if im_id != dt['image_id']: + continue + catid, bbox, score = dt['category_id'], dt['bbox'], dt['score'] + if score < threshold: + continue + + if catid not in catid2color: + idx = np.random.randint(len(color_list)) + catid2color[catid] = color_list[idx] + color = tuple(catid2color[catid]) + + # draw bbox + if len(bbox) == 4: + # draw bbox + xmin, ymin, w, h = bbox + xmax = xmin + w + ymax = ymin + h + draw.line( + [(xmin, ymin), (xmin, ymax), (xmax, ymax), (xmax, ymin), + (xmin, ymin)], + width=2, + fill=color) + elif len(bbox) == 8: + x1, y1, x2, y2, x3, y3, x4, y4 = bbox + draw.line( + [(x1, y1), (x2, y2), (x3, y3), (x4, y4), (x1, y1)], + width=2, + fill=color) + xmin = min(x1, x2, x3, x4) + ymin = min(y1, y2, y3, y4) + else: + logger.error('the shape of bbox must be [M, 4] or [M, 8]!') + + # draw label + text = "{} {:.2f}".format(catid2name[catid], score) + tw, th = draw.textsize(text) + draw.rectangle( + [(xmin + 1, ymin - th), (xmin + tw + 1, ymin)], fill=color) + draw.text((xmin + 1, ymin - th), text, fill=(255, 255, 255)) + + return image + + +def save_result(save_path, results, catid2name, threshold): + """ + save result as txt + """ + img_id = int(results["im_id"]) + with open(save_path, 'w') as f: + if "bbox_res" in results: + for dt in results["bbox_res"]: + catid, bbox, score = dt['category_id'], dt['bbox'], dt['score'] + if score < threshold: + continue + # each bbox result as a line + # for rbox: classname score x1 y1 x2 y2 x3 y3 x4 y4 + # for bbox: classname score x1 y1 w h + bbox_pred = '{} {} '.format(catid2name[catid], + score) + ' '.join( + [str(e) for e in bbox]) + f.write(bbox_pred + '\n') + elif "keypoint_res" in results: + for dt in results["keypoint_res"]: + kpts = dt['keypoints'] + scores = dt['score'] + keypoint_pred = [img_id, scores, kpts] + print(keypoint_pred, file=f) + else: + print("No valid results found, skip txt save") + + +def draw_segm(image, + im_id, + catid2name, + segms, + threshold, + alpha=0.7, + draw_box=True): + """ + Draw segmentation on image + """ + mask_color_id = 0 + w_ratio = .4 + color_list = colormap(rgb=True) + img_array = np.array(image).astype('float32') + for dt in np.array(segms): + if im_id != dt['image_id']: + continue + segm, score, catid = dt['segmentation'], dt['score'], dt['category_id'] + if score < threshold: + continue + import pycocotools.mask as mask_util + mask = mask_util.decode(segm) * 255 + color_mask = color_list[mask_color_id % len(color_list), 0:3] + mask_color_id += 1 + for c in range(3): + color_mask[c] = color_mask[c] * (1 - w_ratio) + w_ratio * 255 + idx = np.nonzero(mask) + img_array[idx[0], idx[1], :] *= 1.0 - alpha + img_array[idx[0], idx[1], :] += alpha * color_mask + + if not draw_box: + center_y, center_x = ndimage.measurements.center_of_mass(mask) + label_text = "{}".format(catid2name[catid]) + vis_pos = (max(int(center_x) - 10, 0), int(center_y)) + cv2.putText(img_array, label_text, vis_pos, + cv2.FONT_HERSHEY_COMPLEX, 0.3, (255, 255, 255)) + else: + mask = mask_util.decode(segm) * 255 + sum_x = np.sum(mask, axis=0) + x = np.where(sum_x > 0.5)[0] + sum_y = np.sum(mask, axis=1) + y = np.where(sum_y > 0.5)[0] + x0, x1, y0, y1 = x[0], x[-1], y[0], y[-1] + cv2.rectangle(img_array, (x0, y0), (x1, y1), + tuple(color_mask.astype('int32').tolist()), 1) + bbox_text = '%s %.2f' % (catid2name[catid], score) + t_size = cv2.getTextSize(bbox_text, 0, 0.3, thickness=1)[0] + cv2.rectangle(img_array, (x0, y0), (x0 + t_size[0], + y0 - t_size[1] - 3), + tuple(color_mask.astype('int32').tolist()), -1) + cv2.putText( + img_array, + bbox_text, (x0, y0 - 2), + cv2.FONT_HERSHEY_SIMPLEX, + 0.3, (0, 0, 0), + 1, + lineType=cv2.LINE_AA) + + return Image.fromarray(img_array.astype('uint8')) + + +def draw_pose(image, + results, + visual_thread=0.6, + save_name='pose.jpg', + save_dir='output', + returnimg=False, + ids=None): + try: + import matplotlib.pyplot as plt + import matplotlib + plt.switch_backend('agg') + except Exception as e: + logger.error('Matplotlib not found, please install matplotlib.' + 'for example: `pip install matplotlib`.') + raise e + + skeletons = np.array([item['keypoints'] for item in results]) + kpt_nums = 17 + if len(skeletons) > 0: + kpt_nums = int(skeletons.shape[1] / 3) + skeletons = skeletons.reshape(-1, kpt_nums, 3) + if kpt_nums == 17: #plot coco keypoint + EDGES = [(0, 1), (0, 2), (1, 3), (2, 4), (3, 5), (4, 6), (5, 7), (6, 8), + (7, 9), (8, 10), (5, 11), (6, 12), (11, 13), (12, 14), + (13, 15), (14, 16), (11, 12)] + else: #plot mpii keypoint + EDGES = [(0, 1), (1, 2), (3, 4), (4, 5), (2, 6), (3, 6), (6, 7), (7, 8), + (8, 9), (10, 11), (11, 12), (13, 14), (14, 15), (8, 12), + (8, 13)] + NUM_EDGES = len(EDGES) + + colors = [[255, 0, 0], [255, 85, 0], [255, 170, 0], [255, 255, 0], [170, 255, 0], [85, 255, 0], [0, 255, 0], \ + [0, 255, 85], [0, 255, 170], [0, 255, 255], [0, 170, 255], [0, 85, 255], [0, 0, 255], [85, 0, 255], \ + [170, 0, 255], [255, 0, 255], [255, 0, 170], [255, 0, 85]] + cmap = matplotlib.cm.get_cmap('hsv') + plt.figure() + + img = np.array(image).astype('float32') + + color_set = results['colors'] if 'colors' in results else None + + if 'bbox' in results and ids is None: + bboxs = results['bbox'] + for j, rect in enumerate(bboxs): + xmin, ymin, xmax, ymax = rect + color = colors[0] if color_set is None else colors[color_set[j] % + len(colors)] + cv2.rectangle(img, (xmin, ymin), (xmax, ymax), color, 1) + + canvas = img.copy() + for i in range(kpt_nums): + for j in range(len(skeletons)): + if skeletons[j][i, 2] < visual_thread: + continue + if ids is None: + color = colors[i] if color_set is None else colors[color_set[j] + % + len(colors)] + else: + color = get_color(ids[j]) + + cv2.circle( + canvas, + tuple(skeletons[j][i, 0:2].astype('int32')), + 2, + color, + thickness=-1) + + to_plot = cv2.addWeighted(img, 0.3, canvas, 0.7, 0) + fig = matplotlib.pyplot.gcf() + + stickwidth = 2 + + for i in range(NUM_EDGES): + for j in range(len(skeletons)): + edge = EDGES[i] + if skeletons[j][edge[0], 2] < visual_thread or skeletons[j][edge[ + 1], 2] < visual_thread: + continue + + cur_canvas = canvas.copy() + X = [skeletons[j][edge[0], 1], skeletons[j][edge[1], 1]] + Y = [skeletons[j][edge[0], 0], skeletons[j][edge[1], 0]] + mX = np.mean(X) + mY = np.mean(Y) + length = ((X[0] - X[1])**2 + (Y[0] - Y[1])**2)**0.5 + angle = math.degrees(math.atan2(X[0] - X[1], Y[0] - Y[1])) + polygon = cv2.ellipse2Poly((int(mY), int(mX)), + (int(length / 2), stickwidth), + int(angle), 0, 360, 1) + if ids is None: + color = colors[i] if color_set is None else colors[color_set[j] + % + len(colors)] + else: + color = get_color(ids[j]) + cv2.fillConvexPoly(cur_canvas, polygon, color) + canvas = cv2.addWeighted(canvas, 0.4, cur_canvas, 0.6, 0) + image = Image.fromarray(canvas.astype('uint8')) + plt.close() + return image diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/voc_utils.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/voc_utils.py new file mode 100644 index 000000000..cd6d9f90e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/ppdet/utils/voc_utils.py @@ -0,0 +1,86 @@ +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import os.path as osp +import re +import random + +__all__ = ['create_list'] + + +def create_list(devkit_dir, years, output_dir): + """ + create following list: + 1. trainval.txt + 2. test.txt + """ + trainval_list = [] + test_list = [] + for year in years: + trainval, test = _walk_voc_dir(devkit_dir, year, output_dir) + trainval_list.extend(trainval) + test_list.extend(test) + + random.shuffle(trainval_list) + with open(osp.join(output_dir, 'trainval.txt'), 'w') as ftrainval: + for item in trainval_list: + ftrainval.write(item[0] + ' ' + item[1] + '\n') + + with open(osp.join(output_dir, 'test.txt'), 'w') as fval: + ct = 0 + for item in test_list: + ct += 1 + fval.write(item[0] + ' ' + item[1] + '\n') + + +def _get_voc_dir(devkit_dir, year, type): + return osp.join(devkit_dir, 'VOC' + year, type) + + +def _walk_voc_dir(devkit_dir, year, output_dir): + filelist_dir = _get_voc_dir(devkit_dir, year, 'ImageSets/Main') + annotation_dir = _get_voc_dir(devkit_dir, year, 'Annotations') + img_dir = _get_voc_dir(devkit_dir, year, 'JPEGImages') + trainval_list = [] + test_list = [] + added = set() + + for _, _, files in os.walk(filelist_dir): + for fname in files: + img_ann_list = [] + if re.match(r'[a-z]+_trainval\.txt', fname): + img_ann_list = trainval_list + elif re.match(r'[a-z]+_test\.txt', fname): + img_ann_list = test_list + else: + continue + fpath = osp.join(filelist_dir, fname) + for line in open(fpath): + name_prefix = line.strip().split()[0] + if name_prefix in added: + continue + added.add(name_prefix) + ann_path = osp.join( + osp.relpath(annotation_dir, output_dir), + name_prefix + '.xml') + img_path = osp.join( + osp.relpath(img_dir, output_dir), name_prefix + '.jpg') + img_ann_list.append((img_path, ann_path)) + + return trainval_list, test_list diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__init__.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__init__.py new file mode 100644 index 000000000..e69de29bb diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/__init__.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 000000000..0f5e2594f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/__init__.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/det_preprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/det_preprocess.cpython-37.pyc new file mode 100644 index 000000000..ee6910208 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/det_preprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/postprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/postprocess.cpython-37.pyc new file mode 100644 index 000000000..fd7a26257 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/postprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_det.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_det.cpython-37.pyc new file mode 100644 index 000000000..882ceb3dc Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_det.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_rec.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_rec.cpython-37.pyc new file mode 100644 index 000000000..1e9a82a46 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/predict_rec.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/preprocess.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/preprocess.cpython-37.pyc new file mode 100644 index 000000000..bd2f03afd Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/__pycache__/preprocess.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/build_gallery.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/build_gallery.py new file mode 100644 index 000000000..7b69a04d7 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/build_gallery.py @@ -0,0 +1,214 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import faiss +import numpy as np +from tqdm import tqdm +import pickle + +from python.predict_rec import RecPredictor + +from utils import logger +from utils import config + + +def split_datafile(data_file, image_root, delimiter="\t"): + ''' + data_file: image path and info, which can be splitted by spacer + image_root: image path root + delimiter: delimiter + ''' + gallery_images = [] + gallery_docs = [] + with open(data_file, 'r', encoding='utf-8') as f: + lines = f.readlines() + for _, ori_line in enumerate(lines): + line = ori_line.strip().split(delimiter) + text_num = len(line) + assert text_num >= 2, f"line({ori_line}) must be splitted into at least 2 parts, but got {text_num}" + image_file = os.path.join(image_root, line[0]) + + gallery_images.append(image_file) + gallery_docs.append(ori_line.strip()) + + return gallery_images, gallery_docs + + +class GalleryBuilder(object): + def __init__(self, config): + + self.config = config + self.rec_predictor = RecPredictor(config) + assert 'IndexProcess' in config.keys(), "Index config not found ... " + self.build(config['IndexProcess']) + + def build(self, config): + ''' + build index from scratch + ''' + operation_method = config.get("index_operation", "new").lower() + + gallery_images, gallery_docs = split_datafile( + config['data_file'], config['image_root'], config['delimiter']) + + # when remove data in index, do not need extract fatures + if operation_method != "remove": + gallery_features = self._extract_features(gallery_images, config) + assert operation_method in [ + "new", "remove", "append" + ], "Only append, remove and new operation are supported" + + # vector.index: faiss index file + # id_map.pkl: use this file to map id to image_doc + if operation_method in ["remove", "append"]: + # if remove or append, vector.index and id_map.pkl must exist + assert os.path.join( + config["index_dir"], "vector.index" + ), "The vector.index dose not exist in {} when 'index_operation' is not None".format( + config["index_dir"]) + assert os.path.join( + config["index_dir"], "id_map.pkl" + ), "The id_map.pkl dose not exist in {} when 'index_operation' is not None".format( + config["index_dir"]) + index = faiss.read_index( + os.path.join(config["index_dir"], "vector.index")) + with open(os.path.join(config["index_dir"], "id_map.pkl"), + 'rb') as fd: + ids = pickle.load(fd) + assert index.ntotal == len(ids.keys( + )), "data number in index is not equal in in id_map" + else: + if not os.path.exists(config["index_dir"]): + os.makedirs(config["index_dir"], exist_ok=True) + index_method = config.get("index_method", "HNSW32") + + # if IVF method, cal ivf number automaticlly + if index_method == "IVF": + index_method = index_method + str( + min(int(len(gallery_images) // 8), 65536)) + ",Flat" + + # for binary index, add B at head of index_method + if config["dist_type"] == "hamming": + index_method = "B" + index_method + + #dist_type + dist_type = faiss.METRIC_INNER_PRODUCT if config[ + "dist_type"] == "IP" else faiss.METRIC_L2 + + #build index + if config["dist_type"] == "hamming": + index = faiss.index_binary_factory(config["embedding_size"], + index_method) + else: + index = faiss.index_factory(config["embedding_size"], + index_method, dist_type) + index = faiss.IndexIDMap2(index) + ids = {} + + if config["index_method"] == "HNSW32": + logger.warning( + "The HNSW32 method dose not support 'remove' operation") + + if operation_method != "remove": + # calculate id for new data + start_id = max(ids.keys()) + 1 if ids else 0 + ids_now = ( + np.arange(0, len(gallery_images)) + start_id).astype(np.int64) + + # only train when new index file + if operation_method == "new": + if config["dist_type"] == "hamming": + index.add(gallery_features) + else: + index.train(gallery_features) + + if not config["dist_type"] == "hamming": + index.add_with_ids(gallery_features, ids_now) + + for i, d in zip(list(ids_now), gallery_docs): + ids[i] = d + else: + if config["index_method"] == "HNSW32": + raise RuntimeError( + "The index_method: HNSW32 dose not support 'remove' operation" + ) + # remove ids in id_map, remove index data in faiss index + remove_ids = list( + filter(lambda k: ids.get(k) in gallery_docs, ids.keys())) + remove_ids = np.asarray(remove_ids) + index.remove_ids(remove_ids) + for k in remove_ids: + del ids[k] + + # store faiss index file and id_map file + if config["dist_type"] == "hamming": + faiss.write_index_binary( + index, os.path.join(config["index_dir"], "vector.index")) + else: + faiss.write_index( + index, os.path.join(config["index_dir"], "vector.index")) + + with open(os.path.join(config["index_dir"], "id_map.pkl"), 'wb') as fd: + pickle.dump(ids, fd) + + def _extract_features(self, gallery_images, config): + # extract gallery features + if config["dist_type"] == "hamming": + gallery_features = np.zeros( + [len(gallery_images), config['embedding_size'] // 8], + dtype=np.uint8) + else: + gallery_features = np.zeros( + [len(gallery_images), config['embedding_size']], + dtype=np.float32) + + #construct batch imgs and do inference + batch_size = config.get("batch_size", 32) + batch_img = [] + for i, image_file in enumerate(tqdm(gallery_images)): + img = cv2.imread(image_file) + if img is None: + logger.error("img empty, please check {}".format(image_file)) + exit() + img = img[:, :, ::-1] + batch_img.append(img) + + if (i + 1) % batch_size == 0: + rec_feat = self.rec_predictor.predict(batch_img) + gallery_features[i - batch_size + 1:i + 1, :] = rec_feat + batch_img = [] + + if len(batch_img) > 0: + rec_feat = self.rec_predictor.predict(batch_img) + gallery_features[-len(batch_img):, :] = rec_feat + batch_img = [] + + return gallery_features + + +def main(config): + GalleryBuilder(config) + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/det_preprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/det_preprocess.py new file mode 100644 index 000000000..65db32dc3 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/det_preprocess.py @@ -0,0 +1,216 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import cv2 +import numpy as np + + +def decode_image(im_file, im_info): + """read rgb image + Args: + im_file (str|np.ndarray): input can be image path or np.ndarray + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + if isinstance(im_file, str): + with open(im_file, 'rb') as f: + im_read = f.read() + data = np.frombuffer(im_read, dtype='uint8') + im = cv2.imdecode(data, 1) # BGR mode, but need RGB mode + im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB) + else: + im = im_file + im_info['im_shape'] = np.array(im.shape[:2], dtype=np.float32) + im_info['scale_factor'] = np.array([1., 1.], dtype=np.float32) + return im, im_info + + +class DetResize(object): + """resize image by target_size and max_size + Args: + target_size (int): the target size of image + keep_ratio (bool): whether keep_ratio or not, default true + interp (int): method of resize + """ + + def __init__( + self, + target_size, + keep_ratio=True, + interp=cv2.INTER_LINEAR, ): + if isinstance(target_size, int): + target_size = [target_size, target_size] + self.target_size = target_size + self.keep_ratio = keep_ratio + self.interp = interp + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + assert len(self.target_size) == 2 + assert self.target_size[0] > 0 and self.target_size[1] > 0 + im_channel = im.shape[2] + im_scale_y, im_scale_x = self.generate_scale(im) + # set image_shape + im_info['input_shape'][1] = int(im_scale_y * im.shape[0]) + im_info['input_shape'][2] = int(im_scale_x * im.shape[1]) + print(0000000000000000000000000000000000000000) + print(im) + print(im_scale_x,im_scale_y,cv2.INTER_LINEAR,self.interp) + im = cv2.resize( + im, + None, + None, + fx=im_scale_x, + fy=im_scale_y, + interpolation=self.interp) + print(im) + im_info['im_shape'] = np.array(im.shape[:2]).astype('float32') + im_info['scale_factor'] = np.array( + [im_scale_y, im_scale_x]).astype('float32') + return im, im_info + + def generate_scale(self, im): + """ + Args: + im (np.ndarray): image (np.ndarray) + Returns: + im_scale_x: the resize ratio of X + im_scale_y: the resize ratio of Y + """ + origin_shape = im.shape[:2] + im_c = im.shape[2] + if self.keep_ratio: + im_size_min = np.min(origin_shape) + im_size_max = np.max(origin_shape) + target_size_min = np.min(self.target_size) + target_size_max = np.max(self.target_size) + im_scale = float(target_size_min) / float(im_size_min) + if np.round(im_scale * im_size_max) > target_size_max: + im_scale = float(target_size_max) / float(im_size_max) + im_scale_x = im_scale + im_scale_y = im_scale + else: + resize_h, resize_w = self.target_size + im_scale_y = resize_h / float(origin_shape[0]) + im_scale_x = resize_w / float(origin_shape[1]) + return im_scale_y, im_scale_x + + +class DetNormalizeImage(object): + """normalize image + Args: + mean (list): im - mean + std (list): im / std + is_scale (bool): whether need im / 255 + is_channel_first (bool): if True: image shape is CHW, else: HWC + """ + + def __init__(self, mean, std, is_scale=True): + self.mean = mean + self.std = std + self.is_scale = is_scale + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + im = im.astype(np.float32, copy=False) + mean = np.array(self.mean)[np.newaxis, np.newaxis, :] + std = np.array(self.std)[np.newaxis, np.newaxis, :] + if self.is_scale: + im = im / 255.0 + print(im) + im -= mean + im /= std + return im, im_info + + +class DetPermute(object): + """permute image + Args: + to_bgr (bool): whether convert RGB to BGR + channel_first (bool): whether convert HWC to CHW + """ + + def __init__(self, ): + super().__init__() + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + #im = im.transpose((2, 0, 1)).copy() + print("detprossssssss") + print(im) + im = im.transpose((2, 0, 1)).copy() + print(im) + return im, im_info + + +class DetPadStride(object): + """ padding image for model with FPN , instead PadBatch(pad_to_stride, pad_gt) in original config + Args: + stride (bool): model with FPN need image shape % stride == 0 + """ + + def __init__(self, stride=0): + self.coarsest_stride = stride + + def __call__(self, im, im_info): + """ + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + Returns: + im (np.ndarray): processed image (np.ndarray) + im_info (dict): info of processed image + """ + coarsest_stride = self.coarsest_stride + if coarsest_stride <= 0: + return im, im_info + im_c, im_h, im_w = im.shape + pad_h = int(np.ceil(float(im_h) / coarsest_stride) * coarsest_stride) + pad_w = int(np.ceil(float(im_w) / coarsest_stride) * coarsest_stride) + padding_im = np.zeros((im_c, pad_h, pad_w), dtype=np.float32) + padding_im[:, :im_h, :im_w] = im + return padding_im, im_info + + +def det_preprocess(im, im_info, preprocess_ops): + for operator in preprocess_ops: + print(operator) + print(im) + print(666) + im, im_info = operator(im, im_info) + print(im) + return im, im_info diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/postprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/postprocess.py new file mode 100644 index 000000000..d26cbaa9a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/postprocess.py @@ -0,0 +1,161 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import copy +import shutil +from functools import partial +import importlib +import numpy as np +import paddle +import paddle.nn.functional as F + + +def build_postprocess(config): + if config is None: + return None + + mod = importlib.import_module(__name__) + config = copy.deepcopy(config) + + main_indicator = config.pop( + "main_indicator") if "main_indicator" in config else None + main_indicator = main_indicator if main_indicator else "" + + func_list = [] + for func in config: + func_list.append(getattr(mod, func)(**config[func])) + return PostProcesser(func_list, main_indicator) + + +class PostProcesser(object): + def __init__(self, func_list, main_indicator="Topk"): + self.func_list = func_list + self.main_indicator = main_indicator + + def __call__(self, x, image_file=None): + rtn = None + for func in self.func_list: + tmp = func(x, image_file) + if type(func).__name__ in self.main_indicator: + rtn = tmp + return rtn + + +class Topk(object): + def __init__(self, topk=1, class_id_map_file=None): + assert isinstance(topk, (int, )) + self.class_id_map = self.parse_class_id_map(class_id_map_file) + self.topk = topk + + def parse_class_id_map(self, class_id_map_file): + if class_id_map_file is None: + return None + + if not os.path.exists(class_id_map_file): + print( + "Warning: If want to use your own label_dict, please input legal path!\nOtherwise label_names will be empty!" + ) + return None + + try: + class_id_map = {} + with open(class_id_map_file, "r") as fin: + lines = fin.readlines() + for line in lines: + partition = line.split("\n")[0].partition(" ") + class_id_map[int(partition[0])] = str(partition[-1]) + except Exception as ex: + print(ex) + class_id_map = None + return class_id_map + + def __call__(self, x, file_names=None, multilabel=False): + if file_names is not None: + assert x.shape[0] == len(file_names) + y = [] + for idx, probs in enumerate(x): + index = probs.argsort(axis=0)[-self.topk:][::-1].astype( + "int32") if not multilabel else np.where( + probs >= 0.5)[0].astype("int32") + clas_id_list = [] + score_list = [] + label_name_list = [] + for i in index: + clas_id_list.append(i.item()) + score_list.append(probs[i].item()) + if self.class_id_map is not None: + label_name_list.append(self.class_id_map[i.item()]) + result = { + "class_ids": clas_id_list, + "scores": np.around( + score_list, decimals=5).tolist(), + } + if file_names is not None: + result["file_name"] = file_names[idx] + if label_name_list is not None: + result["label_names"] = label_name_list + y.append(result) + return y + + +class MultiLabelTopk(Topk): + def __init__(self, topk=1, class_id_map_file=None): + super().__init__() + + def __call__(self, x, file_names=None): + return super().__call__(x, file_names, multilabel=True) + + +class SavePreLabel(object): + def __init__(self, save_dir): + if save_dir is None: + raise Exception( + "Please specify save_dir if SavePreLabel specified.") + self.save_dir = partial(os.path.join, save_dir) + + def __call__(self, x, file_names=None): + if file_names is None: + return + assert x.shape[0] == len(file_names) + for idx, probs in enumerate(x): + index = probs.argsort(axis=0)[-1].astype("int32") + self.save(index, file_names[idx]) + + def save(self, id, image_file): + output_dir = self.save_dir(str(id)) + os.makedirs(output_dir, exist_ok=True) + shutil.copy(image_file, output_dir) + + +class Binarize(object): + def __init__(self, method="round"): + self.method = method + self.unit = np.array([[128, 64, 32, 16, 8, 4, 2, 1]]).T + + def __call__(self, x, file_names=None): + if self.method == "round": + x = np.round(x + 1).astype("uint8") - 1 + + if self.method == "sign": + x = ((np.sign(x) + 1) / 2).astype("uint8") + + embedding_size = x.shape[1] + assert embedding_size % 8 == 0, "The Binary index only support vectors with sizes multiple of 8" + + byte = np.zeros([x.shape[0], embedding_size // 8], dtype=np.uint8) + for i in range(embedding_size // 8): + byte[:, i:i + 1] = np.dot(x[:, i * 8:(i + 1) * 8], self.unit) + + return byte diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_cls.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_cls.py new file mode 100644 index 000000000..cdeb32e48 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_cls.py @@ -0,0 +1,140 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import numpy as np + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from python.preprocess import create_operators +from python.postprocess import build_postprocess + + +class ClsPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"]) + + self.preprocess_ops = [] + self.postprocess = None + if "PreProcess" in config: + if "transform_ops" in config["PreProcess"]: + self.preprocess_ops = create_operators(config["PreProcess"][ + "transform_ops"]) + if "PostProcess" in config: + self.postprocess = build_postprocess(config["PostProcess"]) + + # for whole_chain project to test each repo of paddle + self.benchmark = config["Global"].get("benchmark", False) + if self.benchmark: + import auto_log + import os + pid = os.getpid() + self.auto_logger = auto_log.AutoLogger( + model_name=config["Global"].get("model_name", "cls"), + model_precision='fp16' + if config["Global"]["use_fp16"] else 'fp32', + batch_size=config["Global"].get("batch_size", 1), + data_shape=[3, 224, 224], + save_path=config["Global"].get("save_log_path", + "./auto_log.log"), + inference_config=self.config, + pids=pid, + process_name=None, + gpu_ids=None, + time_keys=[ + 'preprocess_time', 'inference_time', 'postprocess_time' + ], + warmup=2) + + def predict(self, images): + input_names = self.paddle_predictor.get_input_names() + input_tensor = self.paddle_predictor.get_input_handle(input_names[0]) + + output_names = self.paddle_predictor.get_output_names() + output_tensor = self.paddle_predictor.get_output_handle(output_names[ + 0]) + if self.benchmark: + self.auto_logger.times.start() + if not isinstance(images, (list, )): + images = [images] + for idx in range(len(images)): + for ops in self.preprocess_ops: + images[idx] = ops(images[idx]) + image = np.array(images) + if self.benchmark: + self.auto_logger.times.stamp() + + input_tensor.copy_from_cpu(image) + self.paddle_predictor.run() + batch_output = output_tensor.copy_to_cpu() + if self.benchmark: + self.auto_logger.times.stamp() + if self.postprocess is not None: + batch_output = self.postprocess(batch_output) + if self.benchmark: + self.auto_logger.times.end(stamp=True) + return batch_output + + +def main(config): + cls_predictor = ClsPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + batch_imgs = [] + batch_names = [] + cnt = 0 + for idx, img_path in enumerate(image_list): + img = cv2.imread(img_path) + if img is None: + logger.warning( + "Image file failed to read and has been skipped. The path: {}". + format(img_path)) + else: + img = img[:, :, ::-1] + batch_imgs.append(img) + img_name = os.path.basename(img_path) + batch_names.append(img_name) + cnt += 1 + + if cnt % config["Global"]["batch_size"] == 0 or (idx + 1 + ) == len(image_list): + if len(batch_imgs) == 0: + continue + batch_results = cls_predictor.predict(batch_imgs) + for number, result_dict in enumerate(batch_results): + filename = batch_names[number] + clas_ids = result_dict["class_ids"] + scores_str = "[{}]".format(", ".join("{:.2f}".format( + r) for r in result_dict["scores"])) + label_names = result_dict["label_names"] + print("{}:\tclass id(s): {}, score(s): {}, label_name(s): {}". + format(filename, clas_ids, scores_str, label_names)) + batch_imgs = [] + batch_names = [] + if cls_predictor.benchmark: + cls_predictor.auto_logger.report() + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det.py new file mode 100644 index 000000000..0b9c25a5a --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det.py @@ -0,0 +1,195 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from det_preprocess import det_preprocess +from preprocess import create_operators +from utils.draw_bbox import draw_bbox_results + +import os +import argparse +import time +import yaml +import ast +from functools import reduce +import cv2 +import numpy as np +import paddle +import requests +import base64 +import json +class DetPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["det_inference_model_dir"]) + + self.preprocess_ops = create_operators(config["DetPreProcess"][ + "transform_ops"]) + self.config = config + + def preprocess(self, img): + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': np.array( + img.shape[:2], dtype=np.float32), + 'input_shape': self.config["Global"]["image_shape"], + "scale_factor": np.array( + [1., 1.], dtype=np.float32) + } + + im, im_info = det_preprocess(img, im_info, self.preprocess_ops) + print(111111111111111111111) + print(im) + inputs = self.create_inputs(im, im_info) + return inputs + + def create_inputs(self, im, im_info): + """generate input for different model type + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + model_arch (str): model type + Returns: + inputs (dict): input of model + """ + inputs = {} + inputs['image'] = np.array((im, )).astype('float32') + inputs['im_shape'] = np.array( + (im_info['im_shape'], )).astype('float32') + inputs['scale_factor'] = np.array( + (im_info['scale_factor'], )).astype('float32') + #print(inputs) + return inputs + + def parse_det_results(self, pred, threshold, label_list): + max_det_results = self.config["Global"]["max_det_results"] + keep_indexes = pred[:, 1].argsort()[::-1][:max_det_results] + results = [] + for idx in keep_indexes: + single_res = pred[idx] + class_id = int(single_res[0]) + score = single_res[1] + bbox = single_res[2:] + if score < threshold: + continue + label_name = label_list[class_id] + ''' + results.append({ + "class_id": class_id, + "score": score, + "bbox": bbox, + "label_name": label_name, + })''' + results.append({ + "bbox": bbox, + "rec_docs": "background", + "rec_scores": score, + }) + return results + + def predict(self, image, threshold=0.5, run_benchmark=False): + ''' + Args: + image (str/np.ndarray): path of image/ np.ndarray read by cv2ps + threshold (float): threshold of predicted box' score + Returns: + results (dict): include 'boxes': np.ndarray: shape:[N,6], N: number of box, + matix element:[class, score, x_min, y_min, x_max, y_max] + MaskRCNN's results include 'masks': np.ndarray: + shape: [N, im_h, im_w] + ''' + inputs = self.preprocess(image) + print(str(inputs)) + np_boxes = None + input_names = self.paddle_predictor.get_input_names() + print(input_names) + for i in range(len(input_names)): + input_tensor = self.paddle_predictor.get_input_handle(input_names[ + i]) + input_tensor.copy_from_cpu(inputs[input_names[i]]) + print("!!!!!!!",inputs[input_names[i]]) + t1 = time.time() + print(self.paddle_predictor.run()) + output_names = self.paddle_predictor.get_output_names() + boxes_tensor = self.paddle_predictor.get_output_handle(output_names[0]) + + np_boxes = boxes_tensor.copy_to_cpu() + t2 = time.time() + + print("Inference: {} ms per batch image".format((t2 - t1) * 1000.0)) + + # do not perform postprocess in benchmark mode + results = [] + if reduce(lambda x, y: x * y, np_boxes.shape) < 6: + print('[WARNNING] No object detected.') + results = np.array([]) + else: + results = np_boxes + + results = self.parse_det_results(results, + self.config["Global"]["threshold"], + self.config["Global"]["labe_list"]) + return results + + +def main(config): + det_predictor = DetPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = det_predictor.predict(img) + print(output) + draw_bbox_results(img, output, image_file) + + return image_file,output + +def cv2_to_base64_img(img): + data = cv2.imencode('.jpg', img)[1] + return base64.b64encode(data.tostring()).decode('utf8') + +def solve_output(output,image_file): + print(image_file) + img = cv2.imread(image_file) + + for bbox in output: + left,top,right,bottom = int(bbox["bbox"][0]),int(bbox["bbox"][1]),int(bbox["bbox"][2]),int(bbox["bbox"][3]) + print(left,top,right,bottom) + img_crop = img[top:bottom,left:right] + url = "http://123.157.241.94:36807/ppyolo_mbv3/prediction" + img2 = {"key": ["image"], "value": [cv2_to_base64_img(img_crop)]} + r = requests.post(url=url,data=json.dumps(img2), timeout=5) + r = r.json() + print(r) + result = eval(r['value'][0])[0] + cv2.putText(img,str(round(float(result["scores"][0]),2)),(left,top+30), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) + cv2.putText(img,str(result["label_names"][0]),(left,top+60), cv2.FONT_HERSHEY_SIMPLEX,1.2,(0,255,0),2) + cv2.rectangle(img,(left ,top),(right,bottom), (0, 0, 255), 2) + cv2.imwrite("./output/ppyolo_result" + image_file[image_file.rfind("/"):],img) +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + image_file,output = main(config) + #solve_output(output,image_file) + diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det_bak.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det_bak.py new file mode 100644 index 000000000..323d65ab1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_det_bak.py @@ -0,0 +1,167 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from det_preprocess import det_preprocess +from preprocess import create_operators +from utils.draw_bbox import draw_bbox_results + +import os +import argparse +import time +import yaml +import ast +from functools import reduce +import cv2 +import numpy as np +import paddle + + +class DetPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["det_inference_model_dir"]) + + self.preprocess_ops = create_operators(config["DetPreProcess"][ + "transform_ops"]) + self.config = config + + def preprocess(self, img): + im_info = { + 'scale_factor': np.array( + [1., 1.], dtype=np.float32), + 'im_shape': np.array( + img.shape[:2], dtype=np.float32), + 'input_shape': self.config["Global"]["image_shape"], + "scale_factor": np.array( + [1., 1.], dtype=np.float32) + } + im, im_info = det_preprocess(img, im_info, self.preprocess_ops) + inputs = self.create_inputs(im, im_info) + return inputs + + def create_inputs(self, im, im_info): + """generate input for different model type + Args: + im (np.ndarray): image (np.ndarray) + im_info (dict): info of image + model_arch (str): model type + Returns: + inputs (dict): input of model + """ + inputs = {} + inputs['image'] = np.array((im, )).astype('float32') + inputs['im_shape'] = np.array( + (im_info['im_shape'], )).astype('float32') + inputs['scale_factor'] = np.array( + (im_info['scale_factor'], )).astype('float32') + print(inputs) + return inputs + + def parse_det_results(self, pred, threshold, label_list): + max_det_results = self.config["Global"]["max_det_results"] + keep_indexes = pred[:, 1].argsort()[::-1][:max_det_results] + results = [] + for idx in keep_indexes: + single_res = pred[idx] + class_id = int(single_res[0]) + score = single_res[1] + bbox = single_res[2:] + if score < threshold: + continue + label_name = label_list[class_id] + ''' + results.append({ + "class_id": class_id, + "score": score, + "bbox": bbox, + "label_name": label_name, + })''' + results.append({ + "bbox": bbox, + "rec_docs": "background", + "rec_scores": score, + }) + return results + + def predict(self, image, threshold=0.5, run_benchmark=False): + ''' + Args: + image (str/np.ndarray): path of image/ np.ndarray read by cv2 + threshold (float): threshold of predicted box' score + Returns: + results (dict): include 'boxes': np.ndarray: shape:[N,6], N: number of box, + matix element:[class, score, x_min, y_min, x_max, y_max] + MaskRCNN's results include 'masks': np.ndarray: + shape: [N, im_h, im_w] + ''' + inputs = self.preprocess(image) + np_boxes = None + input_names = self.paddle_predictor.get_input_names() + + for i in range(len(input_names)): + input_tensor = self.paddle_predictor.get_input_handle(input_names[ + i]) + input_tensor.copy_from_cpu(inputs[input_names[i]]) + + t1 = time.time() + self.paddle_predictor.run() + output_names = self.paddle_predictor.get_output_names() + boxes_tensor = self.paddle_predictor.get_output_handle(output_names[0]) + np_boxes = boxes_tensor.copy_to_cpu() + t2 = time.time() + + print("Inference: {} ms per batch image".format((t2 - t1) * 1000.0)) + + # do not perform postprocess in benchmark mode + results = [] + if reduce(lambda x, y: x * y, np_boxes.shape) < 6: + print('[WARNNING] No object detected.') + results = np.array([]) + else: + results = np_boxes + + results = self.parse_det_results(results, + self.config["Global"]["threshold"], + self.config["Global"]["labe_list"]) + return results + + +def main(config): + det_predictor = DetPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = det_predictor.predict(img) + print(output) + draw_bbox_results(img, output, image_file) + print(output) + + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_rec.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_rec.py new file mode 100644 index 000000000..d41c513f8 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_rec.py @@ -0,0 +1,105 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import cv2 +import numpy as np + +from utils import logger +from utils import config +from utils.predictor import Predictor +from utils.get_image_list import get_image_list +from preprocess import create_operators +from postprocess import build_postprocess + + +class RecPredictor(Predictor): + def __init__(self, config): + super().__init__(config["Global"], + config["Global"]["rec_inference_model_dir"]) + self.preprocess_ops = create_operators(config["RecPreProcess"][ + "transform_ops"]) + self.postprocess = build_postprocess(config["RecPostProcess"]) + + def predict(self, images, feature_normalize=True): + input_names = self.paddle_predictor.get_input_names() + input_tensor = self.paddle_predictor.get_input_handle(input_names[0]) + + output_names = self.paddle_predictor.get_output_names() + output_tensor = self.paddle_predictor.get_output_handle(output_names[ + 0]) + + if not isinstance(images, (list, )): + images = [images] + for idx in range(len(images)): + for ops in self.preprocess_ops: + images[idx] = ops(images[idx]) + image = np.array(images) + + input_tensor.copy_from_cpu(image) + self.paddle_predictor.run() + batch_output = output_tensor.copy_to_cpu() + + if feature_normalize: + feas_norm = np.sqrt( + np.sum(np.square(batch_output), axis=1, keepdims=True)) + batch_output = np.divide(batch_output, feas_norm) + + if self.postprocess is not None: + batch_output = self.postprocess(batch_output) + return batch_output + + +def main(config): + rec_predictor = RecPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + batch_imgs = [] + batch_names = [] + cnt = 0 + for idx, img_path in enumerate(image_list): + img = cv2.imread(img_path) + if img is None: + logger.warning( + "Image file failed to read and has been skipped. The path: {}". + format(img_path)) + else: + img = img[:, :, ::-1] + batch_imgs.append(img) + img_name = os.path.basename(img_path) + batch_names.append(img_name) + cnt += 1 + + if cnt % config["Global"]["batch_size"] == 0 or (idx + 1) == len(image_list): + if len(batch_imgs) == 0: + continue + + batch_results = rec_predictor.predict(batch_imgs) + for number, result_dict in enumerate(batch_results): + filename = batch_names[number] + print("{}:\t{}".format(filename, result_dict)) + batch_imgs = [] + batch_names = [] + + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_system.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_system.py new file mode 100644 index 000000000..fb2d66a53 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/predict_system.py @@ -0,0 +1,145 @@ +# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import sys + +__dir__ = os.path.dirname(os.path.abspath(__file__)) +sys.path.append(os.path.abspath(os.path.join(__dir__, '../'))) + +import copy +import cv2 +import numpy as np +import faiss +import pickle + +from python.predict_rec import RecPredictor +from python.predict_det import DetPredictor + +from utils import logger +from utils import config +from utils.get_image_list import get_image_list +from utils.draw_bbox import draw_bbox_results + + +class SystemPredictor(object): + def __init__(self, config): + + self.config = config + self.rec_predictor = RecPredictor(config) + self.det_predictor = DetPredictor(config) + + assert 'IndexProcess' in config.keys(), "Index config not found ... " + self.return_k = self.config['IndexProcess']['return_k'] + + index_dir = self.config["IndexProcess"]["index_dir"] + assert os.path.exists(os.path.join( + index_dir, "vector.index")), "vector.index not found ..." + assert os.path.exists(os.path.join( + index_dir, "id_map.pkl")), "id_map.pkl not found ... " + + if config['IndexProcess'].get("binary_index", False): + self.Searcher = faiss.read_index_binary( + os.path.join(index_dir, "vector.index")) + else: + self.Searcher = faiss.read_index( + os.path.join(index_dir, "vector.index")) + + with open(os.path.join(index_dir, "id_map.pkl"), "rb") as fd: + self.id_map = pickle.load(fd) + + def append_self(self, results, shape): + results.append({ + "class_id": 0, + "score": 1.0, + "bbox": + np.array([0, 0, shape[1], shape[0]]), # xmin, ymin, xmax, ymax + "label_name": "foreground", + }) + return results + + def nms_to_rec_results(self, results, thresh=0.1): + filtered_results = [] + x1 = np.array([r["bbox"][0] for r in results]).astype("float32") + y1 = np.array([r["bbox"][1] for r in results]).astype("float32") + x2 = np.array([r["bbox"][2] for r in results]).astype("float32") + y2 = np.array([r["bbox"][3] for r in results]).astype("float32") + scores = np.array([r["rec_scores"] for r in results]) + + areas = (x2 - x1 + 1) * (y2 - y1 + 1) + order = scores.argsort()[::-1] + while order.size > 0: + i = order[0] + xx1 = np.maximum(x1[i], x1[order[1:]]) + yy1 = np.maximum(y1[i], y1[order[1:]]) + xx2 = np.minimum(x2[i], x2[order[1:]]) + yy2 = np.minimum(y2[i], y2[order[1:]]) + + w = np.maximum(0.0, xx2 - xx1 + 1) + h = np.maximum(0.0, yy2 - yy1 + 1) + inter = w * h + ovr = inter / (areas[i] + areas[order[1:]] - inter) + inds = np.where(ovr <= thresh)[0] + order = order[inds + 1] + filtered_results.append(results[i]) + + return filtered_results + + def predict(self, img): + output = [] + # st1: get all detection results + results = self.det_predictor.predict(img) + + # st2: add the whole image for recognition to improve recall + results = self.append_self(results, img.shape) + + # st3: recognition process, use score_thres to ensure accuracy + for result in results: + preds = {} + xmin, ymin, xmax, ymax = result["bbox"].astype("int") + crop_img = img[ymin:ymax, xmin:xmax, :].copy() + rec_results = self.rec_predictor.predict(crop_img) + preds["bbox"] = [xmin, ymin, xmax, ymax] + scores, docs = self.Searcher.search(rec_results, self.return_k) + + # just top-1 result will be returned for the final + if scores[0][0] >= self.config["IndexProcess"]["score_thres"]: + preds["rec_docs"] = self.id_map[docs[0][0]].split()[1] + preds["rec_scores"] = scores[0][0] + output.append(preds) + + # st5: nms to the final results to avoid fetching duplicate results + output = self.nms_to_rec_results( + output, self.config["Global"]["rec_nms_thresold"]) + + return output + + +def main(config): + system_predictor = SystemPredictor(config) + image_list = get_image_list(config["Global"]["infer_imgs"]) + + assert config["Global"]["batch_size"] == 1 + for idx, image_file in enumerate(image_list): + img = cv2.imread(image_file)[:, :, ::-1] + output = system_predictor.predict(img) + print(image_file) + draw_bbox_results(img, output, image_file) + print(output) + return + + +if __name__ == "__main__": + args = config.parse_args() + config = config.get_config(args.config, overrides=args.override, show=True) + main(config) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/preprocess.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/preprocess.py new file mode 100644 index 000000000..1da32ad6e --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/python/preprocess.py @@ -0,0 +1,337 @@ +""" +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +""" + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function +from __future__ import unicode_literals + +from functools import partial +import six +import math +import random +import cv2 +import numpy as np +import importlib +from PIL import Image + +from python.det_preprocess import DetNormalizeImage, DetPadStride, DetPermute, DetResize + + +def create_operators(params): + """ + create operators based on the config + + Args: + params(list): a dict list, used to create some operators + """ + assert isinstance(params, list), ('operator config should be a list') + mod = importlib.import_module(__name__) + ops = [] + for operator in params: + assert isinstance(operator, + dict) and len(operator) == 1, "yaml format error" + op_name = list(operator)[0] + param = {} if operator[op_name] is None else operator[op_name] + op = getattr(mod, op_name)(**param) + ops.append(op) + + return ops + + +class UnifiedResize(object): + def __init__(self, interpolation=None, backend="cv2"): + _cv2_interp_from_str = { + 'nearest': cv2.INTER_NEAREST, + 'bilinear': cv2.INTER_LINEAR, + 'area': cv2.INTER_AREA, + 'bicubic': cv2.INTER_CUBIC, + 'lanczos': cv2.INTER_LANCZOS4 + } + _pil_interp_from_str = { + 'nearest': Image.NEAREST, + 'bilinear': Image.BILINEAR, + 'bicubic': Image.BICUBIC, + 'box': Image.BOX, + 'lanczos': Image.LANCZOS, + 'hamming': Image.HAMMING + } + + def _pil_resize(src, size, resample): + pil_img = Image.fromarray(src) + pil_img = pil_img.resize(size, resample) + return np.asarray(pil_img) + + if backend.lower() == "cv2": + if isinstance(interpolation, str): + interpolation = _cv2_interp_from_str[interpolation.lower()] + # compatible with opencv < version 4.4.0 + elif interpolation is None: + interpolation = cv2.INTER_LINEAR + self.resize_func = partial(cv2.resize, interpolation=interpolation) + elif backend.lower() == "pil": + if isinstance(interpolation, str): + interpolation = _pil_interp_from_str[interpolation.lower()] + self.resize_func = partial(_pil_resize, resample=interpolation) + else: + logger.warning( + f"The backend of Resize only support \"cv2\" or \"PIL\". \"f{backend}\" is unavailable. Use \"cv2\" instead." + ) + self.resize_func = cv2.resize + + def __call__(self, src, size): + return self.resize_func(src, size) + + +class OperatorParamError(ValueError): + """ OperatorParamError + """ + pass + + +class DecodeImage(object): + """ decode image """ + + def __init__(self, to_rgb=True, to_np=False, channel_first=False): + self.to_rgb = to_rgb + self.to_np = to_np # to numpy + self.channel_first = channel_first # only enabled when to_np is True + + def __call__(self, img): + if six.PY2: + assert type(img) is str and len( + img) > 0, "invalid input 'img' in DecodeImage" + else: + assert type(img) is bytes and len( + img) > 0, "invalid input 'img' in DecodeImage" + data = np.frombuffer(img, dtype='uint8') + img = cv2.imdecode(data, 1) + if self.to_rgb: + assert img.shape[2] == 3, 'invalid shape of image[%s]' % ( + img.shape) + img = img[:, :, ::-1] + + if self.channel_first: + img = img.transpose((2, 0, 1)) + + return img + + +class ResizeImage(object): + """ resize image """ + + def __init__(self, + size=None, + resize_short=None, + interpolation=None, + backend="cv2"): + if resize_short is not None and resize_short > 0: + self.resize_short = resize_short + self.w = None + self.h = None + elif size is not None: + self.resize_short = None + self.w = size if type(size) is int else size[0] + self.h = size if type(size) is int else size[1] + else: + raise OperatorParamError("invalid params for ReisizeImage for '\ + 'both 'size' and 'resize_short' are None") + + self._resize_func = UnifiedResize( + interpolation=interpolation, backend=backend) + + def __call__(self, img): + img_h, img_w = img.shape[:2] + if self.resize_short is not None: + percent = float(self.resize_short) / min(img_w, img_h) + w = int(round(img_w * percent)) + h = int(round(img_h * percent)) + else: + w = self.w + h = self.h + return self._resize_func(img, (w, h)) + + +class CropImage(object): + """ crop image """ + + def __init__(self, size): + if type(size) is int: + self.size = (size, size) + else: + self.size = size # (h, w) + + def __call__(self, img): + w, h = self.size + img_h, img_w = img.shape[:2] + + if img_h < h or img_w < w: + raise Exception( + f"The size({h}, {w}) of CropImage must be greater than size({img_h}, {img_w}) of image. Please check image original size and size of ResizeImage if used." + ) + + w_start = (img_w - w) // 2 + h_start = (img_h - h) // 2 + + w_end = w_start + w + h_end = h_start + h + return img[h_start:h_end, w_start:w_end, :] + + +class RandCropImage(object): + """ random crop image """ + + def __init__(self, + size, + scale=None, + ratio=None, + interpolation=None, + backend="cv2"): + if type(size) is int: + self.size = (size, size) # (h, w) + else: + self.size = size + + self.scale = [0.08, 1.0] if scale is None else scale + self.ratio = [3. / 4., 4. / 3.] if ratio is None else ratio + + self._resize_func = UnifiedResize( + interpolation=interpolation, backend=backend) + + def __call__(self, img): + size = self.size + scale = self.scale + ratio = self.ratio + + aspect_ratio = math.sqrt(random.uniform(*ratio)) + w = 1. * aspect_ratio + h = 1. / aspect_ratio + + img_h, img_w = img.shape[:2] + + bound = min((float(img_w) / img_h) / (w**2), + (float(img_h) / img_w) / (h**2)) + scale_max = min(scale[1], bound) + scale_min = min(scale[0], bound) + + target_area = img_w * img_h * random.uniform(scale_min, scale_max) + target_size = math.sqrt(target_area) + w = int(target_size * w) + h = int(target_size * h) + + i = random.randint(0, img_w - w) + j = random.randint(0, img_h - h) + + img = img[j:j + h, i:i + w, :] + + return self._resize_func(img, size) + + +class RandFlipImage(object): + """ random flip image + flip_code: + 1: Flipped Horizontally + 0: Flipped Vertically + -1: Flipped Horizontally & Vertically + """ + + def __init__(self, flip_code=1): + assert flip_code in [-1, 0, 1 + ], "flip_code should be a value in [-1, 0, 1]" + self.flip_code = flip_code + + def __call__(self, img): + if random.randint(0, 1) == 1: + return cv2.flip(img, self.flip_code) + else: + return img + + +class AutoAugment(object): + def __init__(self): + self.policy = ImageNetPolicy() + + def __call__(self, img): + from PIL import Image + img = np.ascontiguousarray(img) + img = Image.fromarray(img) + img = self.policy(img) + img = np.asarray(img) + + +class NormalizeImage(object): + """ normalize image such as substract mean, divide std + """ + + def __init__(self, + scale=None, + mean=None, + std=None, + order='chw', + output_fp16=False, + channel_num=3): + if isinstance(scale, str): + scale = eval(scale) + assert channel_num in [ + 3, 4 + ], "channel number of input image should be set to 3 or 4." + self.channel_num = channel_num + self.output_dtype = 'float16' if output_fp16 else 'float32' + self.scale = np.float32(scale if scale is not None else 1.0 / 255.0) + self.order = order + mean = mean if mean is not None else [0.485, 0.456, 0.406] + std = std if std is not None else [0.229, 0.224, 0.225] + + shape = (3, 1, 1) if self.order == 'chw' else (1, 1, 3) + self.mean = np.array(mean).reshape(shape).astype('float32') + self.std = np.array(std).reshape(shape).astype('float32') + + def __call__(self, img): + from PIL import Image + if isinstance(img, Image.Image): + img = np.array(img) + + assert isinstance(img, + np.ndarray), "invalid input 'img' in NormalizeImage" + + img = (img.astype('float32') * self.scale - self.mean) / self.std + + if self.channel_num == 4: + img_h = img.shape[1] if self.order == 'chw' else img.shape[0] + img_w = img.shape[2] if self.order == 'chw' else img.shape[1] + pad_zeros = np.zeros( + (1, img_h, img_w)) if self.order == 'chw' else np.zeros( + (img_h, img_w, 1)) + img = (np.concatenate( + (img, pad_zeros), axis=0) + if self.order == 'chw' else np.concatenate( + (img, pad_zeros), axis=2)) + return img.astype(self.output_dtype) + + +class ToCHWImage(object): + """ convert hwc image to chw image + """ + + def __init__(self): + pass + + def __call__(self, img): + from PIL import Image + if isinstance(img, Image.Image): + img = np.array(img) + + return img.transpose((2, 0, 1)) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/bbox.json b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/bbox.json new file mode 100644 index 000000000..07f30b1d5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/bbox.json @@ -0,0 +1 @@ +[{"image_id": 0, "category_id": 1, "bbox": [695.3732299804688, 251.02064514160156, 267.14727783203125, 667.5905609130859], "score": 0.873725414276123}, {"image_id": 0, "category_id": 1, "bbox": [313.9666748046875, 371.2845458984375, 484.68304443359375, 648.7493286132812], "score": 0.7265506982803345}, {"image_id": 0, "category_id": 1, "bbox": [677.9193115234375, 361.2143859863281, 318.40789794921875, 701.8808288574219], "score": 0.6433115005493164}, {"image_id": 0, "category_id": 1, "bbox": [578.6394653320312, 274.9395751953125, 366.99652099609375, 684.5674438476562], "score": 0.3449091911315918}, {"image_id": 0, "category_id": 1, "bbox": [360.52294921875, 197.74574279785156, 515.4684448242188, 735.8594207763672], "score": 0.265125036239624}, {"image_id": 0, "category_id": 1, "bbox": [70.241455078125, 31.89822006225586, 796.9187622070312, 981.4816627502441], "score": 0.21352604031562805}, {"image_id": 0, "category_id": 1, "bbox": [496.6165466308594, 357.5012512207031, 427.0873718261719, 716.1687927246094], "score": 0.21161292493343353}, {"image_id": 0, "category_id": 1, "bbox": [641.3889770507812, 138.3158721923828, 381.14117431640625, 685.3566131591797], "score": 0.19399303197860718}, {"image_id": 0, "category_id": 1, "bbox": [204.57801818847656, 45.5566291809082, 355.13804626464844, 607.0048942565918], "score": 0.18370762467384338}, {"image_id": 0, "category_id": 1, "bbox": [395.5518493652344, 472.3749694824219, 451.5275573730469, 607.6250305175781], "score": 0.18051570653915405}, {"image_id": 0, "category_id": 1, "bbox": [1625.5928955078125, 0.0, 294.4071044921875, 623.5017700195312], "score": 0.15865056216716766}, {"image_id": 0, "category_id": 1, "bbox": [259.0799865722656, 157.78518676757812, 347.6471252441406, 600.0338439941406], "score": 0.12434157729148865}, {"image_id": 0, "category_id": 1, "bbox": [321.93572998046875, 157.0794219970703, 410.17913818359375, 611.7620086669922], "score": 0.11471202969551086}, {"image_id": 0, "category_id": 1, "bbox": [785.7514038085938, 248.96939086914062, 397.41314697265625, 666.6106872558594], "score": 0.11342783272266388}, {"image_id": 0, "category_id": 1, "bbox": [303.8069152832031, 11.689882278442383, 404.1588439941406, 592.6624126434326], "score": 0.09945282340049744}, {"image_id": 0, "category_id": 1, "bbox": [217.24996948242188, 146.66268920898438, 833.2827453613281, 933.3373107910156], "score": 0.09419939666986465}, {"image_id": 0, "category_id": 1, "bbox": [414.4038391113281, 13.798330307006836, 827.4375915527344, 995.2766208648682], "score": 0.09250863641500473}, {"image_id": 0, "category_id": 1, "bbox": [719.5376586914062, 514.0806884765625, 393.53778076171875, 565.9193115234375], "score": 0.09161505103111267}, {"image_id": 0, "category_id": 1, "bbox": [192.4891815185547, 406.003173828125, 520.6764068603516, 673.996826171875], "score": 0.09044816344976425}, {"image_id": 0, "category_id": 1, "bbox": [669.56103515625, 42.59006118774414, 309.156982421875, 644.9569358825684], "score": 0.08217358589172363}, {"image_id": 0, "category_id": 1, "bbox": [364.2251892089844, 342.280029296875, 907.6244201660156, 737.719970703125], "score": 0.0715600848197937}, {"image_id": 0, "category_id": 1, "bbox": [620.1963500976562, 20.660905838012695, 860.1359252929688, 970.1545238494873], "score": 0.07122403383255005}, {"image_id": 0, "category_id": 1, "bbox": [113.67265319824219, 312.0962829589844, 485.3422393798828, 697.7134094238281], "score": 0.06512273102998734}, {"image_id": 0, "category_id": 1, "bbox": [1213.353515625, 19.728063583374023, 706.646484375, 966.699182510376], "score": 0.06291121244430542}, {"image_id": 0, "category_id": 1, "bbox": [0.0, 271.63079833984375, 748.4136352539062, 808.3692016601562], "score": 0.05891818925738335}, {"image_id": 0, "category_id": 1, "bbox": [803.0306396484375, 150.83236694335938, 859.263427734375, 929.1676330566406], "score": 0.05692801997065544}, {"image_id": 0, "category_id": 1, "bbox": [205.8365020751953, 0.0, 435.48240661621094, 491.6878967285156], "score": 0.056504569947719574}, {"image_id": 0, "category_id": 1, "bbox": [423.4618225097656, 8.994043350219727, 364.2948913574219, 588.2376461029053], "score": 0.056325770914554596}, {"image_id": 0, "category_id": 1, "bbox": [181.5787811279297, 189.3105926513672, 355.9820098876953, 625.7866363525391], "score": 0.05588925629854202}, {"image_id": 0, "category_id": 1, "bbox": [533.9801025390625, 1.8528099060058594, 351.68658447265625, 602.8824195861816], "score": 0.05168253183364868}, {"image_id": 0, "category_id": 1, "bbox": [0.0, 45.22894287109375, 598.8795166015625, 931.0927124023438], "score": 0.05018442124128342}, {"image_id": 0, "category_id": 1, "bbox": [1343.640380859375, 248.69747924804688, 576.359619140625, 831.3025207519531], "score": 0.0436907596886158}, {"image_id": 0, "category_id": 1, "bbox": [556.1782836914062, 88.61674499511719, 387.64990234375, 659.6730499267578], "score": 0.04353850334882736}, {"image_id": 0, "category_id": 1, "bbox": [1513.3153076171875, 0.0, 384.036376953125, 486.5316162109375], "score": 0.04245037958025932}, {"image_id": 0, "category_id": 1, "bbox": [82.62589263916016, 99.5450439453125, 404.5263900756836, 588.563232421875], "score": 0.04022052511572838}, {"image_id": 0, "category_id": 2, "bbox": [450.3915710449219, 676.380126953125, 578.2577209472656, 403.619873046875], "score": 0.22677572071552277}, {"image_id": 0, "category_id": 2, "bbox": [351.4678649902344, 732.3275756835938, 543.7572937011719, 347.67242431640625], "score": 0.1537037044763565}, {"image_id": 0, "category_id": 2, "bbox": [707.0989379882812, 579.64892578125, 436.22406005859375, 500.35107421875], "score": 0.12268988788127899}, {"image_id": 0, "category_id": 2, "bbox": [564.0088500976562, 772.2525634765625, 551.3733520507812, 307.7474365234375], "score": 0.11803540587425232}, {"image_id": 0, "category_id": 2, "bbox": [787.9828491210938, 719.3270874023438, 377.94049072265625, 360.67291259765625], "score": 0.08660945296287537}, {"image_id": 0, "category_id": 2, "bbox": [359.7429504394531, 555.33984375, 416.2214050292969, 524.66015625], "score": 0.08527899533510208}, {"image_id": 0, "category_id": 2, "bbox": [303.93731689453125, 330.9964294433594, 495.6619873046875, 621.1058044433594], "score": 0.08165193349123001}, {"image_id": 0, "category_id": 2, "bbox": [262.8908386230469, 647.728271484375, 441.9308166503906, 432.271728515625], "score": 0.07838243246078491}, {"image_id": 0, "category_id": 2, "bbox": [884.8604736328125, 661.5791015625, 431.5517578125, 418.4208984375], "score": 0.06516268104314804}, {"image_id": 0, "category_id": 2, "bbox": [813.291748046875, 463.5531311035156, 364.7635498046875, 615.6419372558594], "score": 0.06085878983139992}, {"image_id": 0, "category_id": 2, "bbox": [661.9994506835938, 425.6100158691406, 345.1868896484375, 654.3899841308594], "score": 0.0573558434844017}, {"image_id": 0, "category_id": 2, "bbox": [204.57801818847656, 45.5566291809082, 355.13804626464844, 607.0048942565918], "score": 0.05367649346590042}, {"image_id": 0, "category_id": 2, "bbox": [998.26416015625, 600.0878295898438, 400.3265380859375, 479.91217041015625], "score": 0.05328734591603279}, {"image_id": 0, "category_id": 2, "bbox": [695.3732299804688, 251.02064514160156, 267.14727783203125, 667.5905609130859], "score": 0.05168259143829346}, {"image_id": 0, "category_id": 2, "bbox": [273.9808044433594, 195.56089782714844, 389.7720642089844, 651.4178009033203], "score": 0.050418075174093246}, {"image_id": 0, "category_id": 2, "bbox": [115.82988739013672, 647.270751953125, 480.43604278564453, 432.729248046875], "score": 0.04801110923290253}, {"image_id": 0, "category_id": 2, "bbox": [1087.3267822265625, 669.7846069335938, 422.6895751953125, 410.21539306640625], "score": 0.044634222984313965}, {"image_id": 0, "category_id": 2, "bbox": [1304.338623046875, 82.7868423461914, 328.2535400390625, 659.7090072631836], "score": 0.0437178835272789}, {"image_id": 0, "category_id": 2, "bbox": [916.3257446289062, 417.77728271484375, 377.30450439453125, 611.5194702148438], "score": 0.04322515428066254}, {"image_id": 0, "category_id": 2, "bbox": [307.6468505859375, 48.68049621582031, 405.5543212890625, 608.9121551513672], "score": 0.040602296590805054}, {"image_id": 0, "category_id": 2, "bbox": [1426.37451171875, 0.0, 402.5771484375, 477.365966796875], "score": 0.03979233279824257}, {"image_id": 0, "category_id": 2, "bbox": [1396.8446044921875, 95.05376434326172, 346.6268310546875, 641.4352493286133], "score": 0.03934914618730545}, {"image_id": 0, "category_id": 2, "bbox": [1215.671630859375, 529.39892578125, 391.1053466796875, 550.60107421875], "score": 0.03912525624036789}, {"image_id": 0, "category_id": 2, "bbox": [419.9827880859375, 56.105613708496094, 375.37811279296875, 597.7714004516602], "score": 0.038809966295957565}, {"image_id": 0, "category_id": 2, "bbox": [1116.685791015625, 424.75189208984375, 387.8294677734375, 592.2509155273438], "score": 0.03874225914478302}, {"image_id": 0, "category_id": 2, "bbox": [513.9990234375, 422.6842346191406, 400.6072998046875, 657.3157653808594], "score": 0.03870948404073715}, {"image_id": 0, "category_id": 2, "bbox": [1216.1732177734375, 23.961257934570312, 367.170654296875, 669.3480072021484], "score": 0.037472717463970184}, {"image_id": 0, "category_id": 2, "bbox": [1317.4234619140625, 361.7864990234375, 340.16748046875, 620.159912109375], "score": 0.03717796504497528}, {"image_id": 0, "category_id": 2, "bbox": [72.07478332519531, 155.77610778808594, 407.47047424316406, 592.4249420166016], "score": 0.03708164766430855}, {"image_id": 0, "category_id": 3, "bbox": [364.1284484863281, 406.3480224609375, 510.1944885253906, 673.6519775390625], "score": 0.26016080379486084}, {"image_id": 0, "category_id": 3, "bbox": [677.9193115234375, 361.2143859863281, 318.40789794921875, 701.8808288574219], "score": 0.25869840383529663}, {"image_id": 0, "category_id": 3, "bbox": [583.8218383789062, 564.3338623046875, 503.92864990234375, 515.6661376953125], "score": 0.25320538878440857}, {"image_id": 0, "category_id": 3, "bbox": [695.3732299804688, 251.02064514160156, 267.14727783203125, 667.5905609130859], "score": 0.23515181243419647}, {"image_id": 0, "category_id": 3, "bbox": [360.52294921875, 197.74574279785156, 515.4684448242188, 735.8594207763672], "score": 0.2219495177268982}, {"image_id": 0, "category_id": 3, "bbox": [684.4028930664062, 702.7177124023438, 479.45318603515625, 377.28228759765625], "score": 0.17938347160816193}, {"image_id": 0, "category_id": 3, "bbox": [205.68795776367188, 81.5010757446289, 336.9908752441406, 613.2951278686523], "score": 0.15753011405467987}, {"image_id": 0, "category_id": 3, "bbox": [818.2278442382812, 523.0381469726562, 356.09674072265625, 556.9618530273438], "score": 0.14738178253173828}, {"image_id": 0, "category_id": 3, "bbox": [353.4762268066406, 673.176513671875, 542.4141540527344, 406.823486328125], "score": 0.14062419533729553}, {"image_id": 0, "category_id": 3, "bbox": [449.11468505859375, 733.4881591796875, 576.4956665039062, 346.5118408203125], "score": 0.11796186864376068}, {"image_id": 0, "category_id": 3, "bbox": [308.56683349609375, 75.7350082397461, 414.03912353515625, 639.7704238891602], "score": 0.11490284651517868}, {"image_id": 0, "category_id": 3, "bbox": [205.1191864013672, 0.0, 428.0331573486328, 540.8389892578125], "score": 0.10311776399612427}, {"image_id": 0, "category_id": 3, "bbox": [277.3680725097656, 357.999755859375, 437.1716003417969, 613.5983276367188], "score": 0.09784413874149323}, {"image_id": 0, "category_id": 3, "bbox": [259.0799865722656, 157.78518676757812, 347.6471252441406, 600.0338439941406], "score": 0.09211573749780655}, {"image_id": 0, "category_id": 3, "bbox": [519.8496704101562, 307.5931396484375, 414.776123046875, 714.5018310546875], "score": 0.09025998413562775}, {"image_id": 0, "category_id": 3, "bbox": [360.65911865234375, 460.4512634277344, 909.9977416992188, 619.5487365722656], "score": 0.08487977832555771}, {"image_id": 0, "category_id": 3, "bbox": [79.99319458007812, 139.39523315429688, 786.0184631347656, 940.6047668457031], "score": 0.07553756237030029}, {"image_id": 0, "category_id": 3, "bbox": [419.9827880859375, 56.105613708496094, 375.37811279296875, 597.7714004516602], "score": 0.07363048195838928}, {"image_id": 0, "category_id": 3, "bbox": [904.127197265625, 529.7086181640625, 395.068359375, 550.2913818359375], "score": 0.06731098145246506}, {"image_id": 0, "category_id": 3, "bbox": [236.01333618164062, 33.08015823364258, 785.3453063964844, 987.9305839538574], "score": 0.06729743629693985}, {"image_id": 0, "category_id": 3, "bbox": [268.31854248046875, 556.6338500976562, 426.06280517578125, 523.3661499023438], "score": 0.059684500098228455}, {"image_id": 0, "category_id": 3, "bbox": [796.9890747070312, 324.95416259765625, 390.37554931640625, 675.0439453125], "score": 0.05858004465699196}, {"image_id": 0, "category_id": 3, "bbox": [1513.0430908203125, 0.0, 406.9569091796875, 228.5352325439453], "score": 0.05848925933241844}, {"image_id": 0, "category_id": 3, "bbox": [82.62589263916016, 99.5450439453125, 404.5263900756836, 588.563232421875], "score": 0.056956395506858826}, {"image_id": 0, "category_id": 3, "bbox": [304.0814514160156, 0.0, 421.3404846191406, 501.8050231933594], "score": 0.0562148354947567}, {"image_id": 0, "category_id": 3, "bbox": [533.9801025390625, 1.8528099060058594, 351.68658447265625, 602.8824195861816], "score": 0.051280952990055084}, {"image_id": 0, "category_id": 3, "bbox": [1697.5096435546875, 0.0, 222.4903564453125, 193.1752471923828], "score": 0.049951519817113876}, {"image_id": 0, "category_id": 3, "bbox": [669.56103515625, 42.59006118774414, 309.156982421875, 644.9569358825684], "score": 0.04696698486804962}, {"image_id": 0, "category_id": 3, "bbox": [1668.965087890625, 41.29924774169922, 251.034912109375, 234.64386749267578], "score": 0.04610244184732437}, {"image_id": 0, "category_id": 3, "bbox": [571.380615234375, 246.01126098632812, 896.9088134765625, 833.9887390136719], "score": 0.046007189899683}, {"image_id": 0, "category_id": 3, "bbox": [641.3889770507812, 138.3158721923828, 381.14117431640625, 685.3566131591797], "score": 0.044205036014318466}, {"image_id": 0, "category_id": 3, "bbox": [1625.5928955078125, 0.0, 294.4071044921875, 623.5017700195312], "score": 0.04209528863430023}, {"image_id": 0, "category_id": 3, "bbox": [104.86660766601562, 260.8401794433594, 496.3695983886719, 713.7294616699219], "score": 0.04085254669189453}, {"image_id": 0, "category_id": 3, "bbox": [414.4038391113281, 13.798330307006836, 827.4375915527344, 995.2766208648682], "score": 0.040834248065948486}, {"image_id": 0, "category_id": 3, "bbox": [1304.338623046875, 82.7868423461914, 328.2535400390625, 659.7090072631836], "score": 0.03985943645238876}, {"image_id": 0, "category_id": 3, "bbox": [1513.3153076171875, 0.0, 384.036376953125, 486.5316162109375], "score": 0.03837663680315018}, {"image_id": 1, "category_id": 1, "bbox": [997.5174560546875, 213.14662170410156, 364.3343505859375, 641.0175628662109], "score": 0.24449113011360168}, {"image_id": 1, "category_id": 1, "bbox": [227.72842407226562, 121.88896942138672, 342.5745544433594, 564.0383987426758], "score": 0.1826373040676117}, {"image_id": 1, "category_id": 1, "bbox": [982.564453125, 139.61932373046875, 373.0555419921875, 516.285888671875], "score": 0.16992300748825073}, {"image_id": 1, "category_id": 1, "bbox": [1631.2984619140625, 0.0, 274.699462890625, 555.2535400390625], "score": 0.10783612728118896}, {"image_id": 1, "category_id": 1, "bbox": [357.509033203125, 115.10198974609375, 960.323486328125, 964.8980102539062], "score": 0.09964483976364136}, {"image_id": 1, "category_id": 1, "bbox": [35.93037414550781, 35.0949821472168, 843.0650482177734, 961.9981575012207], "score": 0.0948261097073555}, {"image_id": 1, "category_id": 1, "bbox": [229.6660919189453, 0.0, 382.66831970214844, 524.5968017578125], "score": 0.08264536410570145}, {"image_id": 1, "category_id": 1, "bbox": [1199.9945068359375, 7.507335662841797, 720.0054931640625, 983.9323616027832], "score": 0.07683487236499786}, {"image_id": 1, "category_id": 1, "bbox": [568.7382202148438, 238.6720733642578, 919.2339477539062, 841.3279266357422], "score": 0.07614229619503021}, {"image_id": 1, "category_id": 1, "bbox": [147.3486785888672, 232.01626586914062, 986.4595489501953, 847.9837341308594], "score": 0.06581543385982513}, {"image_id": 1, "category_id": 1, "bbox": [307.1380615234375, 0.0, 395.32794189453125, 577.035400390625], "score": 0.06514111906290054}, {"image_id": 1, "category_id": 1, "bbox": [808.4256591796875, 155.9962921142578, 474.565185546875, 654.7679901123047], "score": 0.061964504420757294}, {"image_id": 1, "category_id": 1, "bbox": [0.0, 40.31918716430664, 568.2688598632812, 945.4715843200684], "score": 0.05491221696138382}, {"image_id": 1, "category_id": 1, "bbox": [218.0932159423828, 149.05030822753906, 432.78709411621094, 697.3251800537109], "score": 0.052795279771089554}, {"image_id": 1, "category_id": 1, "bbox": [1171.8597412109375, 339.8847961425781, 748.1402587890625, 740.1152038574219], "score": 0.05180462822318077}, {"image_id": 1, "category_id": 1, "bbox": [947.7489624023438, 0.0, 373.21234130859375, 559.6314697265625], "score": 0.050812363624572754}, {"image_id": 1, "category_id": 1, "bbox": [47.953216552734375, 0.0, 850.6510314941406, 645.6210327148438], "score": 0.050588712096214294}, {"image_id": 1, "category_id": 1, "bbox": [808.0147705078125, 123.57368469238281, 812.243408203125, 956.4263153076172], "score": 0.04943518340587616}, {"image_id": 1, "category_id": 1, "bbox": [0.0, 252.17372131347656, 738.1876831054688, 827.8262786865234], "score": 0.04770356044173241}, {"image_id": 1, "category_id": 1, "bbox": [1071.857177734375, 276.0357971191406, 408.15771484375, 651.8440856933594], "score": 0.046735167503356934}, {"image_id": 1, "category_id": 1, "bbox": [944.5446166992188, 336.05474853515625, 380.69586181640625, 642.3316650390625], "score": 0.04438221827149391}, {"image_id": 1, "category_id": 1, "bbox": [991.9332885742188, 126.97108459472656, 862.9631958007812, 953.0289154052734], "score": 0.04360035061836243}, {"image_id": 1, "category_id": 1, "bbox": [1034.800537109375, 0.0, 498.264404296875, 652.00048828125], "score": 0.04268684610724449}, {"image_id": 1, "category_id": 1, "bbox": [1698.17626953125, 0.0, 221.82373046875, 197.60653686523438], "score": 0.04033979773521423}, {"image_id": 1, "category_id": 1, "bbox": [184.4237823486328, 0.0, 954.5077362060547, 546.6664428710938], "score": 0.039156991988420486}, {"image_id": 1, "category_id": 1, "bbox": [359.7218017578125, 0.0, 982.869873046875, 644.6243896484375], "score": 0.03688075393438339}, {"image_id": 1, "category_id": 1, "bbox": [0.0, 0.0, 578.293701171875, 544.5452270507812], "score": 0.03655608743429184}, {"image_id": 1, "category_id": 1, "bbox": [980.3106079101562, 0.0, 925.4635620117188, 647.032958984375], "score": 0.034414708614349365}, {"image_id": 1, "category_id": 1, "bbox": [1008.9415893554688, 408.375, 340.67498779296875, 237.55377197265625], "score": 0.03395402431488037}, {"image_id": 1, "category_id": 1, "bbox": [764.9998168945312, 453.47576904296875, 926.6813354492188, 626.5242309570312], "score": 0.033227451145648956}, {"image_id": 1, "category_id": 2, "bbox": [973.8045043945312, 134.927001953125, 392.18572998046875, 515.348388671875], "score": 0.2480056881904602}, {"image_id": 1, "category_id": 2, "bbox": [1033.345458984375, 193.55758666992188, 474.0589599609375, 502.2799377441406], "score": 0.12044714391231537}, {"image_id": 1, "category_id": 2, "bbox": [997.5174560546875, 213.14662170410156, 364.3343505859375, 641.0175628662109], "score": 0.10539223253726959}, {"image_id": 1, "category_id": 2, "bbox": [720.1891479492188, 55.10577392578125, 444.19879150390625, 588.2930908203125], "score": 0.06655608117580414}, {"image_id": 1, "category_id": 2, "bbox": [947.7489624023438, 0.0, 373.21234130859375, 559.6314697265625], "score": 0.06620338559150696}, {"image_id": 1, "category_id": 2, "bbox": [944.5446166992188, 336.05474853515625, 380.69586181640625, 642.3316650390625], "score": 0.05874659866094589}, {"image_id": 1, "category_id": 2, "bbox": [1213.3853759765625, 15.75413703918457, 356.4039306640625, 569.0481090545654], "score": 0.05729030817747116}, {"image_id": 1, "category_id": 2, "bbox": [728.943115234375, 204.8000946044922, 436.90478515625, 587.9622955322266], "score": 0.056557174772024155}, {"image_id": 1, "category_id": 2, "bbox": [837.0780029296875, 285.71533203125, 426.9288330078125, 633.6383666992188], "score": 0.05022319778800011}, {"image_id": 1, "category_id": 2, "bbox": [1006.3452758789062, 482.0842590332031, 350.26129150390625, 594.3320007324219], "score": 0.04891819506883621}, {"image_id": 1, "category_id": 2, "bbox": [1102.243408203125, 342.4518127441406, 371.1197509765625, 634.0279235839844], "score": 0.048709992319345474}, {"image_id": 1, "category_id": 2, "bbox": [822.7955932617188, 485.13818359375, 428.04119873046875, 590.968505859375], "score": 0.04719912260770798}, {"image_id": 1, "category_id": 2, "bbox": [1185.6767578125, 247.70567321777344, 413.442626953125, 614.7293853759766], "score": 0.04611574485898018}, {"image_id": 1, "category_id": 2, "bbox": [1312.8875732421875, 0.0, 401.6728515625, 478.9422302246094], "score": 0.0458284355700016}, {"image_id": 1, "category_id": 2, "bbox": [1078.21728515625, 0.0, 401.2777099609375, 556.3494262695312], "score": 0.04388556629419327}, {"image_id": 1, "category_id": 2, "bbox": [1112.9974365234375, 543.3936157226562, 376.166748046875, 536.6063842773438], "score": 0.04236176237463951}, {"image_id": 1, "category_id": 2, "bbox": [1393.505126953125, 0.0, 426.3736572265625, 538.716552734375], "score": 0.041083235293626785}, {"image_id": 1, "category_id": 2, "bbox": [209.38162231445312, 13.68343734741211, 375.6072692871094, 580.5582618713379], "score": 0.040439072996377945}, {"image_id": 1, "category_id": 2, "bbox": [638.6920166015625, 106.63209533691406, 390.194091796875, 589.7396697998047], "score": 0.040002308785915375}, {"image_id": 1, "category_id": 2, "bbox": [1293.2567138671875, 100.03126525878906, 381.57763671875, 611.4940032958984], "score": 0.03915474936366081}, {"image_id": 1, "category_id": 2, "bbox": [808.0147705078125, 123.57368469238281, 812.243408203125, 956.4263153076172], "score": 0.038055650889873505}, {"image_id": 1, "category_id": 2, "bbox": [1281.4400634765625, 595.01318359375, 434.9503173828125, 484.98681640625], "score": 0.037920162081718445}, {"image_id": 1, "category_id": 2, "bbox": [806.5025024414062, 0.0, 473.18572998046875, 627.6235961914062], "score": 0.03787766024470329}, {"image_id": 1, "category_id": 2, "bbox": [158.92922973632812, 118.40994262695312, 950.0995788574219, 961.5900573730469], "score": 0.03619258478283882}, {"image_id": 1, "category_id": 2, "bbox": [231.0110321044922, 149.81060791015625, 344.5010528564453, 588.2882080078125], "score": 0.03588823601603508}, {"image_id": 1, "category_id": 2, "bbox": [619.6505126953125, 24.263837814331055, 830.1717529296875, 972.438310623169], "score": 0.03542332351207733}, {"image_id": 1, "category_id": 2, "bbox": [360.0400085449219, 226.43418884277344, 963.6678771972656, 853.5658111572266], "score": 0.03445490449666977}, {"image_id": 1, "category_id": 2, "bbox": [15.535308837890625, 282.0261535644531, 452.07696533203125, 572.8584899902344], "score": 0.034069325774908066}, {"image_id": 1, "category_id": 2, "bbox": [1520.5927734375, 0.0, 377.05078125, 488.8376159667969], "score": 0.03389637544751167}, {"image_id": 1, "category_id": 2, "bbox": [783.0108642578125, 0.0, 922.7584228515625, 736.8048095703125], "score": 0.0338481180369854}, {"image_id": 1, "category_id": 2, "bbox": [982.9711303710938, 240.32254028320312, 883.5609741210938, 839.6774597167969], "score": 0.033812154084444046}, {"image_id": 1, "category_id": 2, "bbox": [1214.23583984375, 393.50433349609375, 397.9107666015625, 656.9534301757812], "score": 0.03371719643473625}, {"image_id": 1, "category_id": 2, "bbox": [704.4246826171875, 370.22894287109375, 470.9154052734375, 606.06640625], "score": 0.033109813928604126}, {"image_id": 1, "category_id": 2, "bbox": [518.7343139648438, 103.01958465576172, 408.92376708984375, 601.5965042114258], "score": 0.03308727219700813}, {"image_id": 1, "category_id": 2, "bbox": [996.6253662109375, 0.0, 416.7921142578125, 429.11865234375], "score": 0.03264719247817993}, {"image_id": 1, "category_id": 3, "bbox": [982.564453125, 139.61932373046875, 373.0555419921875, 516.285888671875], "score": 0.7264499664306641}, {"image_id": 1, "category_id": 3, "bbox": [997.5174560546875, 213.14662170410156, 364.3343505859375, 641.0175628662109], "score": 0.4768505096435547}, {"image_id": 1, "category_id": 3, "bbox": [1033.345458984375, 193.55758666992188, 474.0589599609375, 502.2799377441406], "score": 0.26717087626457214}, {"image_id": 1, "category_id": 3, "bbox": [808.4256591796875, 155.9962921142578, 474.565185546875, 654.7679901123047], "score": 0.1510302722454071}, {"image_id": 1, "category_id": 3, "bbox": [223.36300659179688, 74.33831787109375, 344.9846496582031, 563.6932373046875], "score": 0.13036808371543884}, {"image_id": 1, "category_id": 3, "bbox": [947.7489624023438, 0.0, 373.21234130859375, 559.6314697265625], "score": 0.12277168035507202}, {"image_id": 1, "category_id": 3, "bbox": [1071.857177734375, 276.0357971191406, 408.15771484375, 651.8440856933594], "score": 0.1194649338722229}, {"image_id": 1, "category_id": 3, "bbox": [943.7728881835938, 0.0, 537.1013793945312, 660.8310546875], "score": 0.1142682284116745}, {"image_id": 1, "category_id": 3, "bbox": [944.5446166992188, 336.05474853515625, 380.69586181640625, 642.3316650390625], "score": 0.09109840542078018}, {"image_id": 1, "category_id": 3, "bbox": [806.5025024414062, 0.0, 473.18572998046875, 627.6235961914062], "score": 0.08420402556657791}, {"image_id": 1, "category_id": 3, "bbox": [597.389404296875, 121.8159408569336, 852.0494384765625, 958.1840591430664], "score": 0.08250249177217484}, {"image_id": 1, "category_id": 3, "bbox": [1199.9945068359375, 7.507335662841797, 720.0054931640625, 983.9323616027832], "score": 0.07666439563035965}, {"image_id": 1, "category_id": 3, "bbox": [1136.9561767578125, 55.35620880126953, 506.6279296875, 679.4648971557617], "score": 0.07268732786178589}, {"image_id": 1, "category_id": 3, "bbox": [307.1380615234375, 0.0, 395.32794189453125, 577.035400390625], "score": 0.0674259215593338}, {"image_id": 1, "category_id": 3, "bbox": [1649.48876953125, 40.35232925415039, 270.51123046875, 242.96892929077148], "score": 0.06582880765199661}, {"image_id": 1, "category_id": 3, "bbox": [1698.17626953125, 0.0, 221.82373046875, 197.60653686523438], "score": 0.06534602493047714}, {"image_id": 1, "category_id": 3, "bbox": [1516.6143798828125, 0.0, 403.3856201171875, 230.1896209716797], "score": 0.06522159278392792}, {"image_id": 1, "category_id": 3, "bbox": [158.65740966796875, 150.12693786621094, 437.06036376953125, 690.1106719970703], "score": 0.060804471373558044}, {"image_id": 1, "category_id": 3, "bbox": [807.333251953125, 234.8459930419922, 827.429443359375, 845.1540069580078], "score": 0.06033357232809067}, {"image_id": 1, "category_id": 3, "bbox": [4.9527740478515625, 243.22787475585938, 928.4971160888672, 836.7721252441406], "score": 0.05078185349702835}, {"image_id": 1, "category_id": 3, "bbox": [329.77490234375, 326.3771667480469, 1026.0186767578125, 753.6228332519531], "score": 0.049885157495737076}, {"image_id": 1, "category_id": 3, "bbox": [991.9332885742188, 126.97108459472656, 862.9631958007812, 953.0289154052734], "score": 0.04851303622126579}, {"image_id": 1, "category_id": 3, "bbox": [980.3106079101562, 0.0, 925.4635620117188, 647.032958984375], "score": 0.04800134897232056}, {"image_id": 1, "category_id": 3, "bbox": [1338.8642578125, 233.53536987304688, 581.1357421875, 846.4646301269531], "score": 0.04772469028830528}, {"image_id": 1, "category_id": 3, "bbox": [1213.3853759765625, 15.75413703918457, 356.4039306640625, 569.0481090545654], "score": 0.04643239453434944}, {"image_id": 1, "category_id": 3, "bbox": [800.8641967773438, 0.0, 852.9357299804688, 849.7445678710938], "score": 0.044755831360816956}, {"image_id": 1, "category_id": 3, "bbox": [312.36627197265625, 167.0087890625, 401.69873046875, 661.0322265625], "score": 0.041519712656736374}, {"image_id": 1, "category_id": 3, "bbox": [400.1049499511719, 48.6071891784668, 419.9554748535156, 610.5923957824707], "score": 0.0405220091342926}, {"image_id": 1, "category_id": 3, "bbox": [718.2431030273438, 98.81644439697266, 450.33905029296875, 598.4150009155273], "score": 0.038510944694280624}, {"image_id": 1, "category_id": 3, "bbox": [116.00665283203125, 449.1982116699219, 1051.5519409179688, 630.8017883300781], "score": 0.037095338106155396}, {"image_id": 1, "category_id": 3, "bbox": [181.21389770507812, 25.111913681030273, 926.2496032714844, 974.7529544830322], "score": 0.035761523991823196}, {"image_id": 1, "category_id": 3, "bbox": [1394.4649658203125, 0.0, 428.072021484375, 483.1687316894531], "score": 0.034289196133613586}, {"image_id": 1, "category_id": 3, "bbox": [1638.4244384765625, 0.0, 267.3065185546875, 489.2379455566406], "score": 0.034216392785310745}, {"image_id": 1, "category_id": 3, "bbox": [343.0436096191406, 0.0, 1001.8591003417969, 851.460205078125], "score": 0.03389797732234001}, {"image_id": 1, "category_id": 3, "bbox": [36.36460876464844, 119.03919982910156, 466.03038024902344, 653.4481048583984], "score": 0.03331182524561882}, {"image_id": 2, "category_id": 1, "bbox": [744.51220703125, 19.65835189819336, 205.94293212890625, 430.02743911743164], "score": 0.6258186101913452}, {"image_id": 2, "category_id": 1, "bbox": [341.05206298828125, 255.701904296875, 218.2022705078125, 387.58868408203125], "score": 0.6193803548812866}, {"image_id": 2, "category_id": 1, "bbox": [783.372314453125, 424.6482849121094, 224.4444580078125, 295.3517150878906], "score": 0.5141648054122925}, {"image_id": 2, "category_id": 1, "bbox": [726.4847412109375, 93.18611907958984, 169.7232666015625, 394.22974395751953], "score": 0.41604265570640564}, {"image_id": 2, "category_id": 1, "bbox": [335.10595703125, 370.3768310546875, 223.8814697265625, 349.6231689453125], "score": 0.3340129554271698}, {"image_id": 2, "category_id": 1, "bbox": [717.7142333984375, 0.0, 181.5628662109375, 393.0101013183594], "score": 0.33154189586639404}, {"image_id": 2, "category_id": 1, "bbox": [779.9652099609375, 274.44085693359375, 284.675048828125, 438.3267822265625], "score": 0.31457409262657166}, {"image_id": 2, "category_id": 1, "bbox": [846.1083374023438, 313.611328125, 277.40692138671875, 406.388671875], "score": 0.29272767901420593}, {"image_id": 2, "category_id": 1, "bbox": [776.2566528320312, 0.0, 186.19512939453125, 338.6823425292969], "score": 0.2866514027118683}, {"image_id": 2, "category_id": 1, "bbox": [817.074462890625, 26.227901458740234, 210.306640625, 416.2046546936035], "score": 0.27876150608062744}, {"image_id": 2, "category_id": 1, "bbox": [873.1625366210938, 182.67300415039062, 263.01251220703125, 454.3901672363281], "score": 0.26855337619781494}, {"image_id": 2, "category_id": 1, "bbox": [734.4992065429688, 119.64443969726562, 224.48321533203125, 421.4815368652344], "score": 0.24493303894996643}, {"image_id": 2, "category_id": 1, "bbox": [606.072509765625, 438.74627685546875, 218.1585693359375, 281.25372314453125], "score": 0.23254962265491486}, {"image_id": 2, "category_id": 1, "bbox": [920.0265502929688, 267.436279296875, 231.66461181640625, 436.240234375], "score": 0.20196624100208282}, {"image_id": 2, "category_id": 1, "bbox": [652.8048706054688, 61.643943786621094, 231.64385986328125, 397.25260162353516], "score": 0.18838302791118622}, {"image_id": 2, "category_id": 1, "bbox": [705.2552490234375, 395.81103515625, 270.5015869140625, 324.18896484375], "score": 0.16972988843917847}, {"image_id": 2, "category_id": 1, "bbox": [253.3760986328125, 301.7941589355469, 290.6611328125, 355.4490661621094], "score": 0.159967303276062}, {"image_id": 2, "category_id": 1, "bbox": [321.4034729003906, 172.58836364746094, 295.1708068847656, 481.1648712158203], "score": 0.13419145345687866}, {"image_id": 2, "category_id": 1, "bbox": [813.7197875976562, 161.0009002685547, 239.02044677734375, 432.31727600097656], "score": 0.11243642121553421}, {"image_id": 2, "category_id": 1, "bbox": [807.4182739257812, 0.0, 215.60589599609375, 300.4808044433594], "score": 0.11205413937568665}, {"image_id": 2, "category_id": 1, "bbox": [859.1271362304688, 21.510337829589844, 237.39678955078125, 423.57401275634766], "score": 0.1014789417386055}, {"image_id": 2, "category_id": 1, "bbox": [756.2540283203125, 203.2908935546875, 220.2879638671875, 415.2799072265625], "score": 0.09805745631456375}, {"image_id": 2, "category_id": 1, "bbox": [1073.6773681640625, 0.0, 205.17919921875, 371.9908447265625], "score": 0.0948367491364479}, {"image_id": 2, "category_id": 1, "bbox": [485.6324462890625, 504.1072082519531, 289.45184326171875, 215.89279174804688], "score": 0.08554235845804214}, {"image_id": 2, "category_id": 1, "bbox": [26.012725830078125, 91.69560241699219, 570.5726623535156, 628.3043975830078], "score": 0.080867238342762}, {"image_id": 2, "category_id": 1, "bbox": [642.5072021484375, 0.0, 219.867431640625, 347.63055419921875], "score": 0.08052774518728256}, {"image_id": 2, "category_id": 1, "bbox": [689.7314453125, 195.74658203125, 231.28759765625, 428.3529052734375], "score": 0.080379419028759}, {"image_id": 2, "category_id": 1, "bbox": [620.2318115234375, 149.4236297607422, 261.87451171875, 442.38990783691406], "score": 0.07949261367321014}, {"image_id": 2, "category_id": 1, "bbox": [133.427490234375, 19.396190643310547, 602.2239379882812, 636.1353645324707], "score": 0.0761333554983139}, {"image_id": 2, "category_id": 1, "bbox": [292.21710205078125, 93.14957427978516, 550.6807250976562, 626.8504257202148], "score": 0.07333367317914963}, {"image_id": 2, "category_id": 1, "bbox": [777.72705078125, 169.46307373046875, 502.27294921875, 550.5369262695312], "score": 0.06907118856906891}, {"image_id": 2, "category_id": 1, "bbox": [917.220458984375, 0.0, 256.2156982421875, 412.54730224609375], "score": 0.0612383596599102}, {"image_id": 2, "category_id": 1, "bbox": [932.5823974609375, 149.875, 219.9869384765625, 403.67120361328125], "score": 0.06098754703998566}, {"image_id": 2, "category_id": 1, "bbox": [352.563232421875, 139.30145263671875, 226.39105224609375, 420.54901123046875], "score": 0.060007575899362564}, {"image_id": 2, "category_id": 1, "bbox": [678.4630126953125, 278.8869323730469, 276.070556640625, 420.6309509277344], "score": 0.05950196459889412}, {"image_id": 2, "category_id": 1, "bbox": [460.831787109375, 337.7149658203125, 298.4686279296875, 343.38970947265625], "score": 0.05949684605002403}, {"image_id": 2, "category_id": 1, "bbox": [998.7368774414062, 0.0, 245.53350830078125, 412.25396728515625], "score": 0.05697806552052498}, {"image_id": 2, "category_id": 2, "bbox": [460.831787109375, 337.7149658203125, 298.4686279296875, 343.38970947265625], "score": 0.2763864994049072}, {"image_id": 2, "category_id": 2, "bbox": [528.7192993164062, 380.21295166015625, 252.83746337890625, 339.78704833984375], "score": 0.19620339572429657}, {"image_id": 2, "category_id": 2, "bbox": [399.87469482421875, 411.403564453125, 328.1947021484375, 308.596435546875], "score": 0.16476958990097046}, {"image_id": 2, "category_id": 2, "bbox": [332.7411804199219, 402.03375244140625, 260.7944641113281, 317.96624755859375], "score": 0.13649119436740875}, {"image_id": 2, "category_id": 2, "bbox": [583.6277465820312, 337.2882995605469, 288.97125244140625, 357.6206970214844], "score": 0.12643887102603912}, {"image_id": 2, "category_id": 2, "bbox": [517.05029296875, 255.56263732910156, 268.694091796875, 385.2583465576172], "score": 0.11980830878019333}, {"image_id": 2, "category_id": 2, "bbox": [337.965087890625, 301.36932373046875, 227.52069091796875, 358.75146484375], "score": 0.10761857777833939}, {"image_id": 2, "category_id": 2, "bbox": [695.6056518554688, 323.1462097167969, 274.38934326171875, 390.1117248535156], "score": 0.102951280772686}, {"image_id": 2, "category_id": 2, "bbox": [136.2960662841797, 406.1953125, 331.1933135986328, 313.8046875], "score": 0.09216462075710297}, {"image_id": 2, "category_id": 2, "bbox": [216.53497314453125, 378.3658447265625, 321.48394775390625, 341.6341552734375], "score": 0.09048698842525482}, {"image_id": 2, "category_id": 2, "bbox": [612.4758911132812, 403.2018127441406, 222.2435302734375, 316.7981872558594], "score": 0.08795756101608276}, {"image_id": 2, "category_id": 2, "bbox": [765.1884765625, 273.8833312988281, 259.88916015625, 436.8657531738281], "score": 0.08420597016811371}, {"image_id": 2, "category_id": 2, "bbox": [783.372314453125, 424.6482849121094, 224.4444580078125, 295.3517150878906], "score": 0.07613956928253174}, {"image_id": 2, "category_id": 2, "bbox": [458.3592834472656, 219.12957763671875, 286.9651794433594, 390.527587890625], "score": 0.0755750983953476}, {"image_id": 2, "category_id": 2, "bbox": [722.8893432617188, 127.14606475830078, 178.783935546875, 403.04442596435547], "score": 0.07539104670286179}, {"image_id": 2, "category_id": 2, "bbox": [396.8189697265625, 244.10755920410156, 283.2181396484375, 410.03013610839844], "score": 0.06955207884311676}, {"image_id": 2, "category_id": 2, "bbox": [609.1813354492188, 197.85885620117188, 257.57708740234375, 427.9960021972656], "score": 0.06862004101276398}, {"image_id": 2, "category_id": 2, "bbox": [583.3659057617188, 479.3427734375, 219.020751953125, 240.6572265625], "score": 0.06362133473157883}, {"image_id": 2, "category_id": 2, "bbox": [739.1348876953125, 75.46237182617188, 216.69390869140625, 425.19586181640625], "score": 0.06258812546730042}, {"image_id": 2, "category_id": 2, "bbox": [689.7314453125, 195.74658203125, 231.28759765625, 428.3529052734375], "score": 0.062394216656684875}, {"image_id": 2, "category_id": 2, "bbox": [548.1318359375, 168.33888244628906, 257.3387451171875, 412.57603454589844], "score": 0.05868319049477577}, {"image_id": 2, "category_id": 2, "bbox": [811.650634765625, 189.78045654296875, 256.252685546875, 440.336669921875], "score": 0.05751308426260948}, {"image_id": 2, "category_id": 2, "bbox": [812.8318481445312, 48.27541732788086, 218.65264892578125, 413.82223892211914], "score": 0.057065609842538834}, {"image_id": 2, "category_id": 3, "bbox": [458.23187255859375, 301.9195251464844, 303.54986572265625, 349.7483215332031], "score": 0.4590396583080292}, {"image_id": 2, "category_id": 3, "bbox": [337.965087890625, 301.36932373046875, 227.52069091796875, 358.75146484375], "score": 0.3770824670791626}, {"image_id": 2, "category_id": 3, "bbox": [421.9063720703125, 379.9833068847656, 300.52996826171875, 338.5633850097656], "score": 0.27733495831489563}, {"image_id": 2, "category_id": 3, "bbox": [739.1348876953125, 75.46237182617188, 216.69390869140625, 425.19586181640625], "score": 0.2621413767337799}, {"image_id": 2, "category_id": 3, "bbox": [528.7192993164062, 380.21295166015625, 252.83746337890625, 339.78704833984375], "score": 0.2607411742210388}, {"image_id": 2, "category_id": 3, "bbox": [784.1431274414062, 394.77001953125, 226.4840087890625, 325.22998046875], "score": 0.18814820051193237}, {"image_id": 2, "category_id": 3, "bbox": [612.4758911132812, 403.2018127441406, 222.2435302734375, 316.7981872558594], "score": 0.18639683723449707}, {"image_id": 2, "category_id": 3, "bbox": [302.52392578125, 381.92193603515625, 278.96337890625, 338.07806396484375], "score": 0.18557214736938477}, {"image_id": 2, "category_id": 3, "bbox": [583.6277465820312, 337.2882995605469, 288.97125244140625, 357.6206970214844], "score": 0.18001800775527954}, {"image_id": 2, "category_id": 3, "bbox": [765.1612548828125, 0.0, 187.5819091796875, 393.3279724121094], "score": 0.17456629872322083}, {"image_id": 2, "category_id": 3, "bbox": [703.5507202148438, 361.93634033203125, 273.1512451171875, 358.06365966796875], "score": 0.1600247323513031}, {"image_id": 2, "category_id": 3, "bbox": [722.8893432617188, 127.14606475830078, 178.783935546875, 403.04442596435547], "score": 0.1579897403717041}, {"image_id": 2, "category_id": 3, "bbox": [793.8466186523438, 223.62258911132812, 278.86956787109375, 451.4606018066406], "score": 0.15615512430667877}, {"image_id": 2, "category_id": 3, "bbox": [651.6652221679688, 99.53595733642578, 220.79681396484375, 385.6624984741211], "score": 0.137779101729393}, {"image_id": 2, "category_id": 3, "bbox": [812.8318481445312, 48.27541732788086, 218.65264892578125, 413.82223892211914], "score": 0.13507284224033356}, {"image_id": 2, "category_id": 3, "bbox": [717.7142333984375, 0.0, 181.5628662109375, 393.0101013183594], "score": 0.12757351994514465}, {"image_id": 2, "category_id": 3, "bbox": [252.07528686523438, 338.7651062011719, 283.1451110839844, 347.1487731933594], "score": 0.1192394495010376}, {"image_id": 2, "category_id": 3, "bbox": [583.3659057617188, 479.3427734375, 219.020751953125, 240.6572265625], "score": 0.10770971328020096}, {"image_id": 2, "category_id": 3, "bbox": [321.4034729003906, 172.58836364746094, 295.1708068847656, 481.1648712158203], "score": 0.10290965437889099}, {"image_id": 2, "category_id": 3, "bbox": [846.1083374023438, 313.611328125, 277.40692138671875, 406.388671875], "score": 0.10273727774620056}, {"image_id": 2, "category_id": 3, "bbox": [873.1625366210938, 182.67300415039062, 263.01251220703125, 454.3901672363281], "score": 0.09928280115127563}, {"image_id": 2, "category_id": 3, "bbox": [819.218994140625, 0.0, 197.50787353515625, 328.0443115234375], "score": 0.09798740595579147}, {"image_id": 2, "category_id": 3, "bbox": [756.2540283203125, 203.2908935546875, 220.2879638671875, 415.2799072265625], "score": 0.09763888269662857}, {"image_id": 2, "category_id": 3, "bbox": [689.7314453125, 195.74658203125, 231.28759765625, 428.3529052734375], "score": 0.08941485732793808}, {"image_id": 2, "category_id": 3, "bbox": [529.9564819335938, 216.22195434570312, 271.16717529296875, 394.2934265136719], "score": 0.08671518415212631}, {"image_id": 2, "category_id": 3, "bbox": [672.0599975585938, 167.43064880371094, 539.4906616210938, 552.5693511962891], "score": 0.08390206098556519}, {"image_id": 2, "category_id": 3, "bbox": [873.9005126953125, 103.83332824707031, 227.7401123046875, 394.8602752685547], "score": 0.08202643692493439}, {"image_id": 2, "category_id": 3, "bbox": [920.0265502929688, 267.436279296875, 231.66461181640625, 436.240234375], "score": 0.08083553612232208}, {"image_id": 2, "category_id": 3, "bbox": [384.2223205566406, 166.7992401123047, 591.0961608886719, 553.2007598876953], "score": 0.0731722041964531}, {"image_id": 2, "category_id": 3, "bbox": [620.2318115234375, 149.4236297607422, 261.87451171875, 442.38990783691406], "score": 0.07241438329219818}, {"image_id": 2, "category_id": 3, "bbox": [396.8189697265625, 244.10755920410156, 283.2181396484375, 410.03013610839844], "score": 0.07126547396183014}, {"image_id": 2, "category_id": 3, "bbox": [277.08489990234375, 246.34165954589844, 579.9214477539062, 473.65834045410156], "score": 0.06578364968299866}, {"image_id": 2, "category_id": 3, "bbox": [642.5072021484375, 0.0, 219.867431640625, 347.63055419921875], "score": 0.06564173847436905}, {"image_id": 2, "category_id": 3, "bbox": [858.1854248046875, 0.0, 236.5518798828125, 417.0265197753906], "score": 0.06508669257164001}, {"image_id": 2, "category_id": 3, "bbox": [922.9540405273438, 74.82980346679688, 236.89178466796875, 396.58624267578125], "score": 0.06488502770662308}, {"image_id": 2, "category_id": 3, "bbox": [685.9042358398438, 0.0, 532.8472290039062, 499.79718017578125], "score": 0.0637088492512703}, {"image_id": 2, "category_id": 3, "bbox": [521.4393310546875, 234.0424041748047, 593.0108642578125, 485.9575958251953], "score": 0.06309568136930466}, {"image_id": 2, "category_id": 3, "bbox": [257.2774353027344, 21.257274627685547, 605.9231262207031, 635.1350593566895], "score": 0.06047413498163223}, {"image_id": 2, "category_id": 3, "bbox": [144.00750732421875, 91.28055572509766, 588.2901000976562, 628.7194442749023], "score": 0.06018701568245888}, {"image_id": 2, "category_id": 3, "bbox": [0.0, 253.0356903076172, 624.9921875, 466.9643096923828], "score": 0.058970093727111816}, {"image_id": 3, "category_id": 1, "bbox": [1062.6109619140625, 172.09158325195312, 267.5137939453125, 690.2594909667969], "score": 0.8554974794387817}, {"image_id": 3, "category_id": 1, "bbox": [1044.677490234375, 304.13751220703125, 328.4405517578125, 673.0938720703125], "score": 0.22548407316207886}, {"image_id": 3, "category_id": 1, "bbox": [1067.3922119140625, 105.67932891845703, 410.9188232421875, 663.9231491088867], "score": 0.18820108473300934}, {"image_id": 3, "category_id": 1, "bbox": [942.4746704101562, 95.7232666015625, 406.83685302734375, 671.319091796875], "score": 0.17742173373699188}, {"image_id": 3, "category_id": 1, "bbox": [211.43357849121094, 63.336402893066406, 371.26515197753906, 566.8272933959961], "score": 0.13464315235614777}, {"image_id": 3, "category_id": 1, "bbox": [1071.1988525390625, 248.0361328125, 435.7906494140625, 693.4053955078125], "score": 0.1281583160161972}, {"image_id": 3, "category_id": 1, "bbox": [15.644989013671875, 26.96539306640625, 861.0198669433594, 971.7730102539062], "score": 0.12508565187454224}, {"image_id": 3, "category_id": 1, "bbox": [754.4990844726562, 338.10528564453125, 416.88629150390625, 548.2113647460938], "score": 0.10627594590187073}, {"image_id": 3, "category_id": 1, "bbox": [855.8370361328125, 139.90440368652344, 376.65869140625, 656.0648345947266], "score": 0.09000253677368164}, {"image_id": 3, "category_id": 1, "bbox": [752.0570068359375, 492.2317199707031, 427.544677734375, 524.1719665527344], "score": 0.08174709230661392}, {"image_id": 3, "category_id": 1, "bbox": [379.1195068359375, 12.600322723388672, 904.7841796875, 976.2365913391113], "score": 0.06506496667861938}, {"image_id": 3, "category_id": 1, "bbox": [859.6172485351562, 306.58062744140625, 490.55670166015625, 691.5777587890625], "score": 0.05971480533480644}, {"image_id": 3, "category_id": 1, "bbox": [40.517669677734375, 0.0, 841.0033264160156, 644.2609252929688], "score": 0.059572476893663406}, {"image_id": 3, "category_id": 1, "bbox": [1612.1806640625, 0.0, 307.8193359375, 632.8601684570312], "score": 0.056635987013578415}, {"image_id": 3, "category_id": 1, "bbox": [1199.2998046875, 2.7812232971191406, 720.7001953125, 969.9078025817871], "score": 0.05443067476153374}, {"image_id": 3, "category_id": 1, "bbox": [1044.2452392578125, 0.0, 338.06689453125, 630.6046752929688], "score": 0.0515710785984993}, {"image_id": 3, "category_id": 1, "bbox": [164.13111877441406, 124.5389175415039, 958.2756195068359, 955.4610824584961], "score": 0.05086848884820938}, {"image_id": 3, "category_id": 1, "bbox": [871.9336547851562, 16.370933532714844, 712.7164916992188, 969.6228408813477], "score": 0.05083303898572922}, {"image_id": 3, "category_id": 1, "bbox": [225.689208984375, 167.04391479492188, 386.56988525390625, 668.6869201660156], "score": 0.04855210706591606}, {"image_id": 3, "category_id": 1, "bbox": [312.9552917480469, 449.56884765625, 1014.3861389160156, 630.43115234375], "score": 0.046587973833084106}, {"image_id": 3, "category_id": 1, "bbox": [0.0, 366.6127624511719, 970.5386352539062, 713.3872375488281], "score": 0.043125469237565994}, {"image_id": 3, "category_id": 1, "bbox": [584.5629272460938, 134.6425018310547, 870.8963012695312, 945.3574981689453], "score": 0.04302931949496269}, {"image_id": 3, "category_id": 1, "bbox": [206.25791931152344, 0.0, 922.0660552978516, 542.7752685546875], "score": 0.04221544787287712}, {"image_id": 3, "category_id": 1, "bbox": [284.2922058105469, 30.980749130249023, 409.1413879394531, 540.564172744751], "score": 0.04201867803931236}, {"image_id": 3, "category_id": 1, "bbox": [0.0, 133.40328979492188, 543.7913818359375, 946.5967102050781], "score": 0.040996551513671875}, {"image_id": 3, "category_id": 1, "bbox": [774.9108276367188, 163.31649780273438, 356.80828857421875, 605.0807800292969], "score": 0.04084865003824234}, {"image_id": 3, "category_id": 1, "bbox": [1169.1866455078125, 175.12033081054688, 405.1151123046875, 683.1728210449219], "score": 0.03840429335832596}, {"image_id": 3, "category_id": 1, "bbox": [0.0, 0.0, 579.2350463867188, 541.5942993164062], "score": 0.03822336345911026}, {"image_id": 3, "category_id": 1, "bbox": [942.8778076171875, 240.502685546875, 964.590576171875, 839.497314453125], "score": 0.03770189359784126}, {"image_id": 3, "category_id": 2, "bbox": [752.0570068359375, 492.2317199707031, 427.544677734375, 524.1719665527344], "score": 0.3003515899181366}, {"image_id": 3, "category_id": 2, "bbox": [754.4990844726562, 338.10528564453125, 416.88629150390625, 548.2113647460938], "score": 0.233663409948349}, {"image_id": 3, "category_id": 2, "bbox": [653.6954956054688, 408.7784729003906, 446.77386474609375, 501.6832580566406], "score": 0.22367450594902039}, {"image_id": 3, "category_id": 2, "bbox": [774.9108276367188, 163.31649780273438, 356.80828857421875, 605.0807800292969], "score": 0.0981830283999443}, {"image_id": 3, "category_id": 2, "bbox": [652.7385864257812, 207.93833923339844, 391.58514404296875, 607.3859405517578], "score": 0.0954592376947403}, {"image_id": 3, "category_id": 2, "bbox": [856.4178466796875, 0.0, 382.7547607421875, 505.3421936035156], "score": 0.07242642343044281}, {"image_id": 3, "category_id": 2, "bbox": [855.8370361328125, 139.90440368652344, 376.65869140625, 656.0648345947266], "score": 0.07181432098150253}, {"image_id": 3, "category_id": 2, "bbox": [525.6502685546875, 257.5973815917969, 434.15716552734375, 606.4940490722656], "score": 0.07046037167310715}, {"image_id": 3, "category_id": 2, "bbox": [755.4590454101562, 9.1924409866333, 340.84844970703125, 572.8677396774292], "score": 0.06798525899648666}, {"image_id": 3, "category_id": 2, "bbox": [975.8546142578125, 203.83355712890625, 368.1129150390625, 680.2666625976562], "score": 0.06325004249811172}, {"image_id": 3, "category_id": 2, "bbox": [645.7936401367188, 54.5694694519043, 361.14892578125, 590.0721321105957], "score": 0.05978944152593613}, {"image_id": 3, "category_id": 2, "bbox": [997.7792358398438, 0.0, 397.32757568359375, 454.37225341796875], "score": 0.054236311465501785}, {"image_id": 3, "category_id": 2, "bbox": [939.0239868164062, 0.0, 357.60491943359375, 613.39453125], "score": 0.053379639983177185}, {"image_id": 3, "category_id": 2, "bbox": [1073.0240478515625, 157.5142822265625, 251.18603515625, 657.6571044921875], "score": 0.05317959934473038}, {"image_id": 3, "category_id": 2, "bbox": [506.24615478515625, 450.9496765136719, 481.31195068359375, 616.5881652832031], "score": 0.05315014347434044}, {"image_id": 3, "category_id": 2, "bbox": [1327.133056640625, 0.0, 403.162109375, 473.2433166503906], "score": 0.044893115758895874}, {"image_id": 3, "category_id": 2, "bbox": [868.4386596679688, 513.032470703125, 472.65374755859375, 566.967529296875], "score": 0.044380977749824524}, {"image_id": 3, "category_id": 2, "bbox": [1278.5374755859375, 606.5696411132812, 425.0609130859375, 473.43035888671875], "score": 0.043142929673194885}, {"image_id": 3, "category_id": 2, "bbox": [195.42141723632812, 9.756786346435547, 910.6348571777344, 986.5281867980957], "score": 0.04299960285425186}, {"image_id": 3, "category_id": 2, "bbox": [985.7695922851562, 448.445556640625, 428.54547119140625, 623.4085693359375], "score": 0.042913708835840225}, {"image_id": 3, "category_id": 2, "bbox": [425.92767333984375, 375.4584045410156, 431.82537841796875, 574.7204895019531], "score": 0.040645796805620193}, {"image_id": 3, "category_id": 2, "bbox": [584.5629272460938, 134.6425018310547, 870.8963012695312, 945.3574981689453], "score": 0.0405254028737545}, {"image_id": 3, "category_id": 2, "bbox": [889.6776733398438, 0.0, 465.28961181640625, 358.43231201171875], "score": 0.04014848545193672}, {"image_id": 3, "category_id": 2, "bbox": [1097.4705810546875, 0.0, 350.6376953125, 546.5389404296875], "score": 0.03943585231900215}, {"image_id": 3, "category_id": 2, "bbox": [1195.40625, 12.213595390319824, 372.856201171875, 579.0114412307739], "score": 0.03930164873600006}, {"image_id": 3, "category_id": 2, "bbox": [517.7697143554688, 97.83709716796875, 413.25732421875, 607.6661376953125], "score": 0.03894736245274544}, {"image_id": 3, "category_id": 2, "bbox": [871.9336547851562, 16.370933532714844, 712.7164916992188, 969.6228408813477], "score": 0.038895394653081894}, {"image_id": 3, "category_id": 2, "bbox": [1100.6619873046875, 515.8238525390625, 396.0205078125, 564.1761474609375], "score": 0.03860515356063843}, {"image_id": 3, "category_id": 3, "bbox": [758.8390502929688, 401.0691833496094, 410.55584716796875, 512.3815612792969], "score": 0.6097580790519714}, {"image_id": 3, "category_id": 3, "bbox": [759.6553955078125, 203.74600219726562, 385.493896484375, 620.5393981933594], "score": 0.32060912251472473}, {"image_id": 3, "category_id": 3, "bbox": [1053.79736328125, 249.32789611816406, 309.4051513671875, 684.8350067138672], "score": 0.2741156220436096}, {"image_id": 3, "category_id": 3, "bbox": [651.7236938476562, 458.0031433105469, 453.90423583984375, 494.6220397949219], "score": 0.26093780994415283}, {"image_id": 3, "category_id": 3, "bbox": [719.1403198242188, 483.4475402832031, 462.17132568359375, 596.5524597167969], "score": 0.24497590959072113}, {"image_id": 3, "category_id": 3, "bbox": [646.70361328125, 251.95274353027344, 415.8240966796875, 624.9683990478516], "score": 0.23766449093818665}, {"image_id": 3, "category_id": 3, "bbox": [859.6172485351562, 306.58062744140625, 490.55670166015625, 691.5777587890625], "score": 0.18282315135002136}, {"image_id": 3, "category_id": 3, "bbox": [973.3248291015625, 148.60418701171875, 370.6070556640625, 659.4913940429688], "score": 0.1543910801410675}, {"image_id": 3, "category_id": 3, "bbox": [871.9336547851562, 16.370933532714844, 712.7164916992188, 969.6228408813477], "score": 0.11219005286693573}, {"image_id": 3, "category_id": 3, "bbox": [476.3395690917969, 319.9029235839844, 508.9530334472656, 666.6101379394531], "score": 0.09965826570987701}, {"image_id": 3, "category_id": 3, "bbox": [332.31024169921875, 242.50025939941406, 978.8953247070312, 837.4997406005859], "score": 0.09948483109474182}, {"image_id": 3, "category_id": 3, "bbox": [204.70193481445312, 18.419668197631836, 384.5010070800781, 565.8228244781494], "score": 0.09832261502742767}, {"image_id": 3, "category_id": 3, "bbox": [608.7565307617188, 27.79786491394043, 829.2695922851562, 949.5771350860596], "score": 0.095442995429039}, {"image_id": 3, "category_id": 3, "bbox": [1055.3665771484375, 160.5343780517578, 400.16845703125, 651.6037445068359], "score": 0.09204820543527603}, {"image_id": 3, "category_id": 3, "bbox": [855.8370361328125, 139.90440368652344, 376.65869140625, 656.0648345947266], "score": 0.0878898948431015}, {"image_id": 3, "category_id": 3, "bbox": [859.98486328125, 16.704797744750977, 351.4844970703125, 561.8995723724365], "score": 0.07988101989030838}, {"image_id": 3, "category_id": 3, "bbox": [1199.2998046875, 2.7812232971191406, 720.7001953125, 969.9078025817871], "score": 0.0781468003988266}, {"image_id": 3, "category_id": 3, "bbox": [564.3340454101562, 339.4085388183594, 915.3651733398438, 740.5914611816406], "score": 0.07565648853778839}, {"image_id": 3, "category_id": 3, "bbox": [764.0009765625, 57.73131561279297, 340.64453125, 590.4665603637695], "score": 0.07268774509429932}, {"image_id": 3, "category_id": 3, "bbox": [924.7733154296875, 0.0, 388.41015625, 509.7742919921875], "score": 0.07242055982351303}, {"image_id": 3, "category_id": 3, "bbox": [0.0, 257.3575439453125, 948.650634765625, 822.6424560546875], "score": 0.06593562662601471}, {"image_id": 3, "category_id": 3, "bbox": [195.42141723632812, 9.756786346435547, 910.6348571777344, 986.5281867980957], "score": 0.061988331377506256}, {"image_id": 3, "category_id": 3, "bbox": [215.5074005126953, 122.37738037109375, 376.42942810058594, 644.8145751953125], "score": 0.06160365417599678}, {"image_id": 3, "category_id": 3, "bbox": [1090.6600341796875, 311.5863342285156, 391.56884765625, 666.8712463378906], "score": 0.06122758239507675}, {"image_id": 3, "category_id": 3, "bbox": [649.1865844726562, 100.15503692626953, 355.96728515625, 612.296989440918], "score": 0.05818498134613037}, {"image_id": 3, "category_id": 3, "bbox": [985.7695922851562, 448.445556640625, 428.54547119140625, 623.4085693359375], "score": 0.057218585163354874}, {"image_id": 3, "category_id": 3, "bbox": [958.5966796875, 117.79381561279297, 931.6561279296875, 943.087043762207], "score": 0.05542192608118057}, {"image_id": 3, "category_id": 3, "bbox": [882.8435668945312, 563.7335815429688, 464.60565185546875, 516.2664184570312], "score": 0.0543479286134243}, {"image_id": 3, "category_id": 3, "bbox": [1097.4705810546875, 0.0, 350.6376953125, 546.5389404296875], "score": 0.054037921130657196}, {"image_id": 3, "category_id": 3, "bbox": [1047.592529296875, 37.14024353027344, 315.6435546875, 666.1573028564453], "score": 0.05339280143380165}, {"image_id": 3, "category_id": 3, "bbox": [1593.1241455078125, 0.0, 326.8758544921875, 171.67945861816406], "score": 0.05163172632455826}, {"image_id": 3, "category_id": 3, "bbox": [360.89776611328125, 0.0, 936.7034301757812, 866.6369018554688], "score": 0.05028960108757019}, {"image_id": 3, "category_id": 3, "bbox": [384.5453186035156, 0.0, 470.0379333496094, 621.4329833984375], "score": 0.050009556114673615}, {"image_id": 3, "category_id": 3, "bbox": [1747.4293212890625, 0.0, 172.5706787109375, 190.30010986328125], "score": 0.048559874296188354}, {"image_id": 3, "category_id": 3, "bbox": [1337.5218505859375, 240.56752014160156, 582.4781494140625, 839.4324798583984], "score": 0.04853980615735054}, {"image_id": 3, "category_id": 3, "bbox": [1664.9505615234375, 40.913291931152344, 255.0494384765625, 235.99219512939453], "score": 0.04805126413702965}, {"image_id": 3, "category_id": 3, "bbox": [978.0254516601562, 0.0, 935.7778930664062, 636.2633666992188], "score": 0.04743540287017822}, {"image_id": 3, "category_id": 3, "bbox": [295.2467041015625, 0.0, 453.291015625, 541.2123413085938], "score": 0.04569484293460846}, {"image_id": 3, "category_id": 3, "bbox": [525.8992309570312, 200.7103271484375, 428.49810791015625, 610.9923095703125], "score": 0.04564596712589264}, {"image_id": 3, "category_id": 3, "bbox": [993.9578857421875, 0.0, 415.4808349609375, 408.9975280761719], "score": 0.04514865204691887}, {"image_id": 3, "category_id": 3, "bbox": [119.15414428710938, 458.3590393066406, 1029.3442687988281, 621.6409606933594], "score": 0.04494065046310425}, {"image_id": 3, "category_id": 3, "bbox": [513.8644409179688, 51.16743087768555, 417.33184814453125, 591.2509651184082], "score": 0.04119905084371567}, {"image_id": 3, "category_id": 3, "bbox": [761.5701904296875, 451.7342834472656, 912.9019775390625, 628.2657165527344], "score": 0.039075352251529694}, {"image_id": 4, "category_id": 1, "bbox": [397.45013427734375, 331.7752380371094, 320.796630859375, 372.2411193847656], "score": 0.28497105836868286}, {"image_id": 4, "category_id": 1, "bbox": [376.37060546875, 416.31146240234375, 284.68511962890625, 303.68853759765625], "score": 0.1885378509759903}, {"image_id": 4, "category_id": 1, "bbox": [24.439422607421875, 6.773895263671875, 572.1087951660156, 670.6602478027344], "score": 0.17092974483966827}, {"image_id": 4, "category_id": 1, "bbox": [445.58050537109375, 414.1954650878906, 334.97515869140625, 305.8045349121094], "score": 0.13983874022960663}, {"image_id": 4, "category_id": 1, "bbox": [232.28192138671875, 0.0, 666.2379760742188, 668.6141967773438], "score": 0.10380109399557114}, {"image_id": 4, "category_id": 1, "bbox": [176.77444458007812, 43.659324645996094, 260.2023620605469, 375.76309967041016], "score": 0.1037050411105156}, {"image_id": 4, "category_id": 1, "bbox": [531.2096557617188, 14.432842254638672, 575.9177856445312, 650.7794380187988], "score": 0.08547429740428925}, {"image_id": 4, "category_id": 1, "bbox": [910.6246948242188, 171.77479553222656, 192.69866943359375, 381.67869567871094], "score": 0.07813899219036102}, {"image_id": 4, "category_id": 1, "bbox": [115.0670166015625, 70.14557647705078, 609.45947265625, 649.8544235229492], "score": 0.07679840177297592}, {"image_id": 4, "category_id": 1, "bbox": [0.0, 10.541553497314453, 382.60296630859375, 651.5828971862793], "score": 0.06780929118394852}, {"image_id": 4, "category_id": 1, "bbox": [520.8143310546875, 272.1015930175781, 276.12982177734375, 430.3283386230469], "score": 0.06716139614582062}, {"image_id": 4, "category_id": 1, "bbox": [390.17010498046875, 73.32344055175781, 626.562744140625, 646.6765594482422], "score": 0.06582247465848923}, {"image_id": 4, "category_id": 1, "bbox": [652.3070678710938, 0.0, 570.4338989257812, 593.7958984375], "score": 0.05452652648091316}, {"image_id": 4, "category_id": 1, "bbox": [447.46588134765625, 241.58740234375, 296.29461669921875, 425.5943603515625], "score": 0.04758226126432419}, {"image_id": 4, "category_id": 1, "bbox": [894.305908203125, 61.47008514404297, 223.75146484375, 431.90503692626953], "score": 0.04694928228855133}, {"image_id": 4, "category_id": 1, "bbox": [799.7202758789062, 16.609336853027344, 480.27972412109375, 655.8966445922852], "score": 0.046636469662189484}, {"image_id": 4, "category_id": 1, "bbox": [0.0, 164.54600524902344, 495.96441650390625, 555.4539947509766], "score": 0.04421522095799446}, {"image_id": 4, "category_id": 1, "bbox": [129.14520263671875, 0.0, 268.1044006347656, 364.01470947265625], "score": 0.043574798852205276}, {"image_id": 4, "category_id": 1, "bbox": [581.5690307617188, 340.9226379394531, 304.24249267578125, 379.0773620605469], "score": 0.04293534904718399}, {"image_id": 4, "category_id": 1, "bbox": [614.4529418945312, 156.3250732421875, 665.5470581054688, 563.6749267578125], "score": 0.04131922125816345}, {"image_id": 4, "category_id": 1, "bbox": [891.5611572265625, 165.38406372070312, 388.4388427734375, 554.6159362792969], "score": 0.040074389427900314}, {"image_id": 4, "category_id": 1, "bbox": [0.0, 313.2903747558594, 649.7593994140625, 406.7096252441406], "score": 0.03984108939766884}, {"image_id": 4, "category_id": 1, "bbox": [362.428955078125, 243.91925048828125, 273.8861083984375, 419.33087158203125], "score": 0.03846558555960655}, {"image_id": 4, "category_id": 1, "bbox": [836.0636596679688, 168.64820861816406, 236.83526611328125, 388.6175994873047], "score": 0.038367822766304016}, {"image_id": 4, "category_id": 1, "bbox": [359.7510986328125, 310.008544921875, 636.627197265625, 409.991455078125], "score": 0.03755444660782814}, {"image_id": 4, "category_id": 1, "bbox": [0.0, 0.0, 625.0577392578125, 418.72918701171875], "score": 0.035537876188755035}, {"image_id": 4, "category_id": 1, "bbox": [886.0164184570312, 227.14430236816406, 247.46343994140625, 432.3501434326172], "score": 0.03456771746277809}, {"image_id": 4, "category_id": 1, "bbox": [197.64126586914062, 0.0, 252.26589965820312, 289.5471496582031], "score": 0.03391050547361374}, {"image_id": 4, "category_id": 1, "bbox": [4.626350402832031, 0.0, 196.50164031982422, 316.6710510253906], "score": 0.03320614993572235}, {"image_id": 4, "category_id": 1, "bbox": [926.2615356445312, 0.0, 249.37567138671875, 413.1147766113281], "score": 0.032749176025390625}, {"image_id": 4, "category_id": 2, "bbox": [377.0150146484375, 441.8638000488281, 356.01873779296875, 278.1361999511719], "score": 0.1847490817308426}, {"image_id": 4, "category_id": 2, "bbox": [518.9544677734375, 501.8125305175781, 276.9503173828125, 218.18746948242188], "score": 0.08543698489665985}, {"image_id": 4, "category_id": 2, "bbox": [576.572509765625, 471.4356994628906, 318.5218505859375, 248.56430053710938], "score": 0.0722944587469101}, {"image_id": 4, "category_id": 2, "bbox": [653.3884887695312, 433.4109802246094, 312.9586181640625, 286.5890197753906], "score": 0.06941265612840652}, {"image_id": 4, "category_id": 2, "bbox": [376.3510437011719, 274.1206970214844, 269.1332092285156, 424.9900207519531], "score": 0.05690854415297508}, {"image_id": 4, "category_id": 2, "bbox": [589.2485961914062, 323.2144775390625, 284.6270751953125, 393.0078125], "score": 0.056753985583782196}, {"image_id": 4, "category_id": 2, "bbox": [733.3068237304688, 357.3998718261719, 263.5528564453125, 362.6001281738281], "score": 0.05355953797698021}, {"image_id": 4, "category_id": 2, "bbox": [207.04470825195312, 441.85577392578125, 307.4269104003906, 278.14422607421875], "score": 0.049918774515390396}, {"image_id": 4, "category_id": 2, "bbox": [170.00930786132812, 0.0, 265.4787292480469, 365.0776062011719], "score": 0.04930106922984123}, {"image_id": 4, "category_id": 2, "bbox": [712.827880859375, 483.6827087402344, 304.56011962890625, 236.31729125976562], "score": 0.049154140055179596}, {"image_id": 4, "category_id": 2, "bbox": [282.4036865234375, 456.0691223144531, 305.145263671875, 263.9308776855469], "score": 0.04835902899503708}, {"image_id": 4, "category_id": 2, "bbox": [802.904296875, 441.2373046875, 270.303466796875, 278.7626953125], "score": 0.04729650914669037}, {"image_id": 4, "category_id": 2, "bbox": [281.1540222167969, 285.8037414550781, 299.9507751464844, 398.7033386230469], "score": 0.0444076806306839}, {"image_id": 4, "category_id": 2, "bbox": [24.439422607421875, 6.773895263671875, 572.1087951660156, 670.6602478027344], "score": 0.04322801157832146}, {"image_id": 4, "category_id": 2, "bbox": [520.8143310546875, 272.1015930175781, 276.12982177734375, 430.3283386230469], "score": 0.04288966581225395}, {"image_id": 4, "category_id": 2, "bbox": [858.801025390625, 395.4917297363281, 295.848876953125, 324.5082702636719], "score": 0.042575303465127945}, {"image_id": 4, "category_id": 2, "bbox": [115.0670166015625, 70.14557647705078, 609.45947265625, 649.8544235229492], "score": 0.04233325272798538}, {"image_id": 4, "category_id": 2, "bbox": [447.46588134765625, 241.58740234375, 296.29461669921875, 425.5943603515625], "score": 0.04075048863887787}, {"image_id": 4, "category_id": 2, "bbox": [666.6970825195312, 283.4668884277344, 275.7899169921875, 388.2985534667969], "score": 0.0395934134721756}, {"image_id": 4, "category_id": 2, "bbox": [148.91253662109375, 325.29266357421875, 291.58331298828125, 390.6903076171875], "score": 0.03882993012666702}, {"image_id": 4, "category_id": 2, "bbox": [908.2291870117188, 118.75697326660156, 193.99029541015625, 380.1521453857422], "score": 0.038248952478170395}, {"image_id": 4, "category_id": 2, "bbox": [487.989013671875, 0.0, 283.19122314453125, 272.9339599609375], "score": 0.03801370784640312}, {"image_id": 4, "category_id": 2, "bbox": [143.1012420654297, 176.53311157226562, 306.8593292236328, 407.0924987792969], "score": 0.037674326449632645}, {"image_id": 4, "category_id": 2, "bbox": [738.9717407226562, 249.3201904296875, 260.5611572265625, 391.15972900390625], "score": 0.03731111064553261}, {"image_id": 4, "category_id": 2, "bbox": [203.1391143798828, 244.4378662109375, 314.8472137451172, 416.3216552734375], "score": 0.037274062633514404}, {"image_id": 4, "category_id": 2, "bbox": [190.6995086669922, 69.39990997314453, 276.6146697998047, 384.0929183959961], "score": 0.036336854100227356}, {"image_id": 4, "category_id": 2, "bbox": [387.3960876464844, 148.62301635742188, 624.8207092285156, 571.3769836425781], "score": 0.034882862120866776}, {"image_id": 4, "category_id": 2, "bbox": [91.60383605957031, 211.22630310058594, 283.0103302001953, 412.54335021972656], "score": 0.034308817237615585}, {"image_id": 4, "category_id": 2, "bbox": [808.9228515625, 278.013427734375, 252.779541015625, 402.91943359375], "score": 0.03400701656937599}, {"image_id": 4, "category_id": 2, "bbox": [914.099609375, 172.1251220703125, 239.283935546875, 383.98223876953125], "score": 0.03395265340805054}, {"image_id": 4, "category_id": 2, "bbox": [831.378173828125, 106.09955596923828, 241.36767578125, 396.61702728271484], "score": 0.03381285071372986}, {"image_id": 4, "category_id": 2, "bbox": [337.2003173828125, 0.0, 290.8519287109375, 277.9066467285156], "score": 0.0337570495903492}, {"image_id": 4, "category_id": 2, "bbox": [117.33758544921875, 0.0, 643.1915283203125, 368.38494873046875], "score": 0.03323587775230408}, {"image_id": 4, "category_id": 2, "bbox": [277.93719482421875, 178.6481170654297, 296.04034423828125, 403.57460021972656], "score": 0.03302469104528427}, {"image_id": 4, "category_id": 2, "bbox": [201.95736694335938, 0.0, 260.64837646484375, 247.5636444091797], "score": 0.032673660665750504}, {"image_id": 4, "category_id": 3, "bbox": [393.51251220703125, 375.6250305175781, 326.43798828125, 344.3749694824219], "score": 0.36567801237106323}, {"image_id": 4, "category_id": 3, "bbox": [452.724609375, 482.6803283691406, 329.5482177734375, 237.31967163085938], "score": 0.2727847695350647}, {"image_id": 4, "category_id": 3, "bbox": [518.72509765625, 413.3856506347656, 307.924560546875, 306.6143493652344], "score": 0.1634632647037506}, {"image_id": 4, "category_id": 3, "bbox": [581.5690307617188, 340.9226379394531, 304.24249267578125, 379.0773620605469], "score": 0.12450331449508667}, {"image_id": 4, "category_id": 3, "bbox": [430.58770751953125, 242.2676544189453, 351.80267333984375, 477.7323455810547], "score": 0.12397114932537079}, {"image_id": 4, "category_id": 3, "bbox": [357.0220947265625, 483.7095642089844, 301.26727294921875, 236.29043579101562], "score": 0.12064994126558304}, {"image_id": 4, "category_id": 3, "bbox": [910.6246948242188, 171.77479553222656, 192.69866943359375, 381.67869567871094], "score": 0.12026355415582657}, {"image_id": 4, "category_id": 3, "bbox": [361.70684814453125, 233.12232971191406, 656.2803344726562, 486.87767028808594], "score": 0.08650375157594681}, {"image_id": 4, "category_id": 3, "bbox": [176.21185302734375, 13.021408081054688, 255.90704345703125, 375.21839904785156], "score": 0.08559748530387878}, {"image_id": 4, "category_id": 3, "bbox": [894.305908203125, 61.47008514404297, 223.75146484375, 431.90503692626953], "score": 0.08531910926103592}, {"image_id": 4, "category_id": 3, "bbox": [243.307861328125, 148.9058380126953, 647.037109375, 571.0941619873047], "score": 0.07947652786970139}, {"image_id": 4, "category_id": 3, "bbox": [658.0391845703125, 360.10845947265625, 291.13458251953125, 359.89154052734375], "score": 0.07744626700878143}, {"image_id": 4, "category_id": 3, "bbox": [20.744720458984375, 77.56388854980469, 574.2724914550781, 642.4361114501953], "score": 0.07439250499010086}, {"image_id": 4, "category_id": 3, "bbox": [84.84759521484375, 237.20913696289062, 695.0452270507812, 482.7908630371094], "score": 0.06842564791440964}, {"image_id": 4, "category_id": 3, "bbox": [836.0636596679688, 168.64820861816406, 236.83526611328125, 388.6175994873047], "score": 0.06791317462921143}, {"image_id": 4, "category_id": 3, "bbox": [376.3510437011719, 274.1206970214844, 269.1332092285156, 424.9900207519531], "score": 0.06650400906801224}, {"image_id": 4, "category_id": 3, "bbox": [578.517578125, 516.388916015625, 324.10345458984375, 203.611083984375], "score": 0.06137565150856972}, {"image_id": 4, "category_id": 3, "bbox": [390.4422607421875, 2.4640274047851562, 634.010009765625, 661.6906967163086], "score": 0.056865230202674866}, {"image_id": 4, "category_id": 3, "bbox": [113.1173095703125, 0.0, 617.25439453125, 679.6934814453125], "score": 0.056435950100421906}, {"image_id": 4, "category_id": 3, "bbox": [729.5507202148438, 397.22991943359375, 269.9395751953125, 322.77008056640625], "score": 0.055190324783325195}, {"image_id": 4, "category_id": 3, "bbox": [886.0164184570312, 227.14430236816406, 247.46343994140625, 432.3501434326172], "score": 0.05418068915605545}, {"image_id": 4, "category_id": 3, "bbox": [504.15887451171875, 152.8347625732422, 638.1753540039062, 567.1652374267578], "score": 0.051819853484630585}, {"image_id": 4, "category_id": 3, "bbox": [630.5555419921875, 10.142784118652344, 610.983154296875, 652.5460586547852], "score": 0.04585684463381767}, {"image_id": 4, "category_id": 3, "bbox": [261.17547607421875, 323.62603759765625, 333.8204345703125, 396.37396240234375], "score": 0.04497351124882698}, {"image_id": 4, "category_id": 3, "bbox": [810.75634765625, 317.15380859375, 243.022216796875, 399.53863525390625], "score": 0.04169793799519539}, {"image_id": 4, "category_id": 3, "bbox": [263.6995849609375, 387.8171081542969, 624.1729736328125, 332.1828918457031], "score": 0.04066556319594383}, {"image_id": 4, "category_id": 3, "bbox": [743.3009033203125, 286.3894958496094, 252.87646484375, 383.6214294433594], "score": 0.04034033417701721}, {"image_id": 4, "category_id": 3, "bbox": [0.0, 83.92012786865234, 369.17962646484375, 636.0798721313477], "score": 0.03898531571030617}, {"image_id": 4, "category_id": 3, "bbox": [802.7232666015625, 166.0010528564453, 477.2767333984375, 553.9989471435547], "score": 0.03709874674677849}, {"image_id": 4, "category_id": 3, "bbox": [233.95718383789062, 0.0, 683.8572692871094, 570.8981323242188], "score": 0.03680828958749771}, {"image_id": 4, "category_id": 3, "bbox": [585.8990478515625, 305.35723876953125, 694.1009521484375, 414.64276123046875], "score": 0.03561432659626007}, {"image_id": 4, "category_id": 3, "bbox": [322.98309326171875, 0.0, 313.616943359375, 405.09906005859375], "score": 0.03513503819704056}, {"image_id": 4, "category_id": 3, "bbox": [113.47465515136719, 45.97132873535156, 278.7163543701172, 427.8950958251953], "score": 0.03398764878511429}, {"image_id": 4, "category_id": 3, "bbox": [261.68212890625, 23.628725051879883, 281.277587890625, 354.17440605163574], "score": 0.03351886570453644}, {"image_id": 4, "category_id": 3, "bbox": [863.9315795898438, 344.4158020019531, 274.74139404296875, 375.5841979980469], "score": 0.03298121690750122}, {"image_id": 5, "category_id": 1, "bbox": [956.1685180664062, 260.16558837890625, 412.75115966796875, 536.9993286132812], "score": 0.30046936869621277}, {"image_id": 5, "category_id": 1, "bbox": [1621.6953125, 0.0, 296.770751953125, 626.8226318359375], "score": 0.14502152800559998}, {"image_id": 5, "category_id": 1, "bbox": [353.158447265625, 0.0, 987.421875, 1010.9869995117188], "score": 0.13432577252388}, {"image_id": 5, "category_id": 1, "bbox": [0.0, 3.7047958374023438, 966.9107666015625, 997.7532119750977], "score": 0.12530557811260223}, {"image_id": 5, "category_id": 1, "bbox": [945.591796875, 0.0, 484.7181396484375, 706.3046264648438], "score": 0.10867444425821304}, {"image_id": 5, "category_id": 1, "bbox": [583.670654296875, 108.8931884765625, 874.1475830078125, 971.1068115234375], "score": 0.08977926522493362}, {"image_id": 5, "category_id": 1, "bbox": [890.5581665039062, 305.8571472167969, 478.58905029296875, 708.2825622558594], "score": 0.07439488172531128}, {"image_id": 5, "category_id": 1, "bbox": [830.692626953125, 243.2830352783203, 472.4171142578125, 507.66258239746094], "score": 0.07203829288482666}, {"image_id": 5, "category_id": 1, "bbox": [128.74708557128906, 227.4475860595703, 1034.4153900146484, 852.5524139404297], "score": 0.06600533425807953}, {"image_id": 5, "category_id": 1, "bbox": [966.5064697265625, 3.0489635467529297, 917.572998046875, 988.5023059844971], "score": 0.06547219306230545}, {"image_id": 5, "category_id": 1, "bbox": [213.190673828125, 78.98770141601562, 378.62847900390625, 534.1972351074219], "score": 0.06259377300739288}, {"image_id": 5, "category_id": 1, "bbox": [795.945556640625, 225.63690185546875, 869.8419189453125, 854.3630981445312], "score": 0.05380347743630409}, {"image_id": 5, "category_id": 1, "bbox": [1171.639892578125, 241.8497772216797, 748.360107421875, 838.1502227783203], "score": 0.05033690482378006}, {"image_id": 5, "category_id": 1, "bbox": [0.0, 13.808887481689453, 554.8766479492188, 979.1710319519043], "score": 0.048862524330616}, {"image_id": 5, "category_id": 1, "bbox": [1697.041748046875, 0.0, 222.958251953125, 200.71592712402344], "score": 0.04558086395263672}, {"image_id": 5, "category_id": 1, "bbox": [1355.1815185546875, 18.338430404663086, 564.8184814453125, 964.8923435211182], "score": 0.045574888586997986}, {"image_id": 5, "category_id": 1, "bbox": [1057.018798828125, 259.3863220214844, 469.1553955078125, 667.5538024902344], "score": 0.043422408401966095}, {"image_id": 5, "category_id": 1, "bbox": [0.0, 474.4252624511719, 968.00537109375, 605.5747375488281], "score": 0.041644107550382614}, {"image_id": 5, "category_id": 1, "bbox": [820.8034057617188, 0.0, 793.5350952148438, 865.2833251953125], "score": 0.040661003440618515}, {"image_id": 5, "category_id": 1, "bbox": [323.7532653808594, 448.4374084472656, 1004.2499084472656, 631.5625915527344], "score": 0.0400838740170002}, {"image_id": 5, "category_id": 1, "bbox": [1636.8585205078125, 0.0, 267.6837158203125, 361.3399658203125], "score": 0.03650720790028572}, {"image_id": 5, "category_id": 1, "bbox": [1505.548583984375, 0.0, 387.53857421875, 493.5966796875], "score": 0.035585515201091766}, {"image_id": 5, "category_id": 1, "bbox": [533.3654174804688, 0.0, 364.6549072265625, 610.3468627929688], "score": 0.03535037115216255}, {"image_id": 5, "category_id": 1, "bbox": [937.1987915039062, 459.3435974121094, 972.5089721679688, 620.6564025878906], "score": 0.034255675971508026}, {"image_id": 5, "category_id": 1, "bbox": [1401.9259033203125, 0.0, 409.0919189453125, 541.6380615234375], "score": 0.033335719257593155}, {"image_id": 5, "category_id": 1, "bbox": [981.0538940429688, 0.0, 924.3976440429688, 547.3361206054688], "score": 0.03330210596323013}, {"image_id": 5, "category_id": 1, "bbox": [143.71266174316406, 581.6504516601562, 1034.221664428711, 498.34954833984375], "score": 0.032556820660829544}, {"image_id": 5, "category_id": 1, "bbox": [582.886474609375, 0.0, 910.189208984375, 540.8436889648438], "score": 0.03207579255104065}, {"image_id": 5, "category_id": 1, "bbox": [827.2205810546875, 33.344398498535156, 426.9951171875, 644.2256698608398], "score": 0.03157854080200195}, {"image_id": 5, "category_id": 1, "bbox": [1661.2877197265625, 59.752281188964844, 258.7122802734375, 233.23603057861328], "score": 0.03097069263458252}, {"image_id": 5, "category_id": 2, "bbox": [950.8788452148438, 186.25462341308594, 418.97528076171875, 525.0670928955078], "score": 0.22238703072071075}, {"image_id": 5, "category_id": 2, "bbox": [830.692626953125, 243.2830352783203, 472.4171142578125, 507.66258239746094], "score": 0.1210339292883873}, {"image_id": 5, "category_id": 2, "bbox": [909.3261108398438, 475.22796630859375, 376.38116455078125, 602.5188598632812], "score": 0.05901431292295456}, {"image_id": 5, "category_id": 2, "bbox": [981.8705444335938, 551.8323974609375, 432.29254150390625, 528.1676025390625], "score": 0.05487018823623657}, {"image_id": 5, "category_id": 2, "bbox": [1078.989990234375, 474.0835876464844, 439.7535400390625, 603.7419738769531], "score": 0.0526604950428009}, {"image_id": 5, "category_id": 2, "bbox": [1408.180419921875, 0.0, 409.4453125, 488.05767822265625], "score": 0.04824093356728554}, {"image_id": 5, "category_id": 2, "bbox": [1185.76708984375, 108.07569885253906, 404.654541015625, 608.6047210693359], "score": 0.0480555035173893}, {"image_id": 5, "category_id": 2, "bbox": [685.03564453125, 497.6167907714844, 502.0745849609375, 565.8171691894531], "score": 0.04603036493062973}, {"image_id": 5, "category_id": 2, "bbox": [990.8941650390625, 0.0, 379.1138916015625, 621.62353515625], "score": 0.04505329579114914}, {"image_id": 5, "category_id": 2, "bbox": [827.2205810546875, 33.344398498535156, 426.9951171875, 644.2256698608398], "score": 0.043478380888700485}, {"image_id": 5, "category_id": 2, "bbox": [1220.586181640625, 6.700046539306641, 335.5550537109375, 591.6674461364746], "score": 0.042202357202768326}, {"image_id": 5, "category_id": 2, "bbox": [1323.04248046875, 37.253074645996094, 338.216796875, 616.5127334594727], "score": 0.042107369750738144}, {"image_id": 5, "category_id": 2, "bbox": [834.2210693359375, 324.0699157714844, 424.1512451171875, 669.1179504394531], "score": 0.04172458499670029}, {"image_id": 5, "category_id": 2, "bbox": [780.5537719726562, 0.0, 478.80206298828125, 422.1895446777344], "score": 0.040771082043647766}, {"image_id": 5, "category_id": 2, "bbox": [1074.253173828125, 41.06483459472656, 384.6904296875, 636.8113250732422], "score": 0.039597347378730774}, {"image_id": 5, "category_id": 2, "bbox": [1184.2655029296875, 530.8296508789062, 446.2265625, 549.1703491210938], "score": 0.038608577102422714}, {"image_id": 5, "category_id": 2, "bbox": [714.4671020507812, 291.5511474609375, 464.58538818359375, 626.0831909179688], "score": 0.0385429672896862}, {"image_id": 5, "category_id": 2, "bbox": [210.35899353027344, 30.988628387451172, 369.7432403564453, 547.4008369445801], "score": 0.03807708993554115}, {"image_id": 5, "category_id": 2, "bbox": [1283.0614013671875, 599.5542602539062, 466.635498046875, 480.44573974609375], "score": 0.03753926604986191}, {"image_id": 5, "category_id": 2, "bbox": [715.2542724609375, 65.34293365478516, 433.3011474609375, 574.1610336303711], "score": 0.036861054599285126}, {"image_id": 5, "category_id": 2, "bbox": [801.68994140625, 117.8184814453125, 839.186279296875, 962.1815185546875], "score": 0.03675336018204689}, {"image_id": 5, "category_id": 2, "bbox": [156.614501953125, 96.91244506835938, 970.72998046875, 983.0875549316406], "score": 0.03631417080760002}, {"image_id": 5, "category_id": 2, "bbox": [1329.5518798828125, 0.0, 386.1292724609375, 442.8591003417969], "score": 0.03628207743167877}, {"image_id": 5, "category_id": 2, "bbox": [1513.5531005859375, 0.0, 378.713134765625, 553.240234375], "score": 0.03356136381626129}, {"image_id": 5, "category_id": 2, "bbox": [564.4074096679688, 220.264892578125, 926.8534545898438, 859.735107421875], "score": 0.03313019871711731}, {"image_id": 5, "category_id": 2, "bbox": [1171.376953125, 299.4496765136719, 448.0792236328125, 607.1381530761719], "score": 0.03227568045258522}, {"image_id": 5, "category_id": 2, "bbox": [208.6549530029297, 0.0, 476.92311096191406, 355.59039306640625], "score": 0.03215698525309563}, {"image_id": 5, "category_id": 2, "bbox": [1399.17822265625, 104.72794342041016, 393.0999755859375, 597.3866806030273], "score": 0.03195074573159218}, {"image_id": 5, "category_id": 2, "bbox": [621.8156127929688, 116.43562316894531, 367.104248046875, 574.0768280029297], "score": 0.03149815648794174}, {"image_id": 5, "category_id": 2, "bbox": [709.8687744140625, 0.0, 415.95458984375, 468.53009033203125], "score": 0.03108149766921997}, {"image_id": 5, "category_id": 2, "bbox": [1006.236572265625, 0.0, 863.6441650390625, 875.8999633789062], "score": 0.030956100672483444}, {"image_id": 5, "category_id": 2, "bbox": [1295.3104248046875, 211.69558715820312, 378.8870849609375, 599.9358825683594], "score": 0.030679363757371902}, {"image_id": 5, "category_id": 2, "bbox": [353.158447265625, 0.0, 987.421875, 1010.9869995117188], "score": 0.03061136044561863}, {"image_id": 5, "category_id": 2, "bbox": [531.3404541015625, 152.58302307128906, 400.82464599609375, 600.1512298583984], "score": 0.030512532219290733}, {"image_id": 5, "category_id": 2, "bbox": [4.3487548828125, 231.90806579589844, 445.3187255859375, 567.4317779541016], "score": 0.030426377430558205}, {"image_id": 5, "category_id": 3, "bbox": [950.8788452148438, 186.25462341308594, 418.97528076171875, 525.0670928955078], "score": 0.7375662922859192}, {"image_id": 5, "category_id": 3, "bbox": [830.692626953125, 243.2830352783203, 472.4171142578125, 507.66258239746094], "score": 0.3983542025089264}, {"image_id": 5, "category_id": 3, "bbox": [890.5581665039062, 305.8571472167969, 478.58905029296875, 708.2825622558594], "score": 0.14430004358291626}, {"image_id": 5, "category_id": 3, "bbox": [990.8941650390625, 0.0, 379.1138916015625, 621.62353515625], "score": 0.12069300562143326}, {"image_id": 5, "category_id": 3, "bbox": [577.3232421875, 2.845853805541992, 878.870361328125, 999.3872394561768], "score": 0.10505835711956024}, {"image_id": 5, "category_id": 3, "bbox": [362.6519470214844, 99.6335678100586, 949.3628234863281, 980.3664321899414], "score": 0.10019754618406296}, {"image_id": 5, "category_id": 3, "bbox": [189.5326690673828, 78.22709655761719, 398.23191833496094, 540.3027496337891], "score": 0.09464608877897263}, {"image_id": 5, "category_id": 3, "bbox": [1074.253173828125, 41.06483459472656, 384.6904296875, 636.8113250732422], "score": 0.08195967972278595}, {"image_id": 5, "category_id": 3, "bbox": [1036.265625, 206.93515014648438, 528.30078125, 681.2730407714844], "score": 0.08146850019693375}, {"image_id": 5, "category_id": 3, "bbox": [1643.39794921875, 39.05764389038086, 276.60205078125, 247.12887954711914], "score": 0.0806666761636734}, {"image_id": 5, "category_id": 3, "bbox": [0.81097412109375, 104.36997985839844, 939.8970336914062, 975.6300201416016], "score": 0.08053837716579437}, {"image_id": 5, "category_id": 3, "bbox": [1697.041748046875, 0.0, 222.958251953125, 200.71592712402344], "score": 0.07660679519176483}, {"image_id": 5, "category_id": 3, "bbox": [655.692138671875, 98.2972640991211, 529.0037841796875, 695.6805801391602], "score": 0.07519432157278061}, {"image_id": 5, "category_id": 3, "bbox": [1571.6898193359375, 0.0, 348.3101806640625, 178.37571716308594], "score": 0.07481846213340759}, {"image_id": 5, "category_id": 3, "bbox": [820.8034057617188, 0.0, 793.5350952148438, 865.2833251953125], "score": 0.07472988963127136}, {"image_id": 5, "category_id": 3, "bbox": [827.2205810546875, 33.344398498535156, 426.9951171875, 644.2256698608398], "score": 0.06232774257659912}, {"image_id": 5, "category_id": 3, "bbox": [1159.2353515625, 180.61441040039062, 469.951416015625, 631.8509216308594], "score": 0.061221808195114136}, {"image_id": 5, "category_id": 3, "bbox": [1006.236572265625, 0.0, 863.6441650390625, 875.8999633789062], "score": 0.058593012392520905}, {"image_id": 5, "category_id": 3, "bbox": [795.945556640625, 225.63690185546875, 869.8419189453125, 854.3630981445312], "score": 0.053846538066864014}, {"image_id": 5, "category_id": 3, "bbox": [155.44253540039062, 0.0, 998.8584899902344, 1008.457763671875], "score": 0.04744111746549606}, {"image_id": 5, "category_id": 3, "bbox": [383.4782409667969, 0.0, 494.1878356933594, 615.5613403320312], "score": 0.04694577679038048}, {"image_id": 5, "category_id": 3, "bbox": [1167.425048828125, 0.0, 752.574951171875, 645.7105712890625], "score": 0.04673805460333824}, {"image_id": 5, "category_id": 3, "bbox": [1076.22216796875, 333.46148681640625, 428.93505859375, 640.6642456054688], "score": 0.0464433953166008}, {"image_id": 5, "category_id": 3, "bbox": [121.07894897460938, 338.3389587402344, 1057.2566223144531, 741.6610412597656], "score": 0.04643399640917778}, {"image_id": 5, "category_id": 3, "bbox": [1190.261474609375, 130.0609893798828, 729.738525390625, 949.9390106201172], "score": 0.04445460066199303}, {"image_id": 5, "category_id": 3, "bbox": [520.1185302734375, 335.1689758300781, 1017.3902587890625, 744.8310241699219], "score": 0.04367470741271973}, {"image_id": 5, "category_id": 3, "bbox": [881.6383666992188, 0.0, 482.27435302734375, 426.3958740234375], "score": 0.04026566445827484}, {"image_id": 5, "category_id": 3, "bbox": [1408.180419921875, 0.0, 409.4453125, 488.05767822265625], "score": 0.03996983543038368}, {"image_id": 5, "category_id": 3, "bbox": [1201.527099609375, 50.129634857177734, 366.19775390625, 610.3369178771973], "score": 0.03832313418388367}, {"image_id": 5, "category_id": 3, "bbox": [349.0066223144531, 0.0, 1029.4253845214844, 743.968994140625], "score": 0.03815001621842384}, {"image_id": 5, "category_id": 3, "bbox": [1621.6953125, 0.0, 296.770751953125, 626.8226318359375], "score": 0.03798140957951546}, {"image_id": 5, "category_id": 3, "bbox": [582.886474609375, 0.0, 910.189208984375, 540.8436889648438], "score": 0.03742585331201553}, {"image_id": 5, "category_id": 3, "bbox": [263.39715576171875, 47.50605010986328, 459.880126953125, 520.3822555541992], "score": 0.03553660959005356}, {"image_id": 5, "category_id": 3, "bbox": [909.3261108398438, 475.22796630859375, 376.38116455078125, 602.5188598632812], "score": 0.035287342965602875}, {"image_id": 5, "category_id": 3, "bbox": [1340.2591552734375, 1.5999269485473633, 315.2366943359375, 589.5212888717651], "score": 0.03498566523194313}, {"image_id": 6, "category_id": 1, "bbox": [433.6170349121094, 267.14471435546875, 353.9197082519531, 447.96905517578125], "score": 0.7232762575149536}, {"image_id": 6, "category_id": 1, "bbox": [730.9089965820312, 361.3147277832031, 268.27508544921875, 358.6852722167969], "score": 0.6687101125717163}, {"image_id": 6, "category_id": 1, "bbox": [592.9573974609375, 268.8675842285156, 242.83209228515625, 448.2185363769531], "score": 0.392996609210968}, {"image_id": 6, "category_id": 1, "bbox": [818.9022216796875, 213.0418243408203, 207.84423828125, 473.7554168701172], "score": 0.28210991621017456}, {"image_id": 6, "category_id": 1, "bbox": [370.306640625, 263.79229736328125, 288.6119384765625, 447.26470947265625], "score": 0.275052011013031}, {"image_id": 6, "category_id": 1, "bbox": [421.39422607421875, 209.20753479003906, 286.05877685546875, 448.75865173339844], "score": 0.1997687667608261}, {"image_id": 6, "category_id": 1, "bbox": [871.7845458984375, 190.12167358398438, 225.8309326171875, 454.2715759277344], "score": 0.16719281673431396}, {"image_id": 6, "category_id": 1, "bbox": [762.9659423828125, 238.96583557128906, 321.26953125, 481.03416442871094], "score": 0.15111011266708374}, {"image_id": 6, "category_id": 1, "bbox": [300.63360595703125, 0.30095672607421875, 539.8870239257812, 679.0578079223633], "score": 0.14505526423454285}, {"image_id": 6, "category_id": 1, "bbox": [732.2869873046875, 193.60488891601562, 263.54925537109375, 455.8025817871094], "score": 0.1414945274591446}, {"image_id": 6, "category_id": 1, "bbox": [3.283203125, 5.3186187744140625, 609.5269775390625, 665.8978729248047], "score": 0.13905780017375946}, {"image_id": 6, "category_id": 1, "bbox": [698.2744750976562, 242.80313110351562, 241.82049560546875, 441.1030578613281], "score": 0.13851772248744965}, {"image_id": 6, "category_id": 1, "bbox": [567.4459228515625, 169.3579864501953, 230.49005126953125, 436.86595153808594], "score": 0.1346355378627777}, {"image_id": 6, "category_id": 1, "bbox": [634.8463745117188, 372.4523620605469, 238.80645751953125, 347.5476379394531], "score": 0.12593361735343933}, {"image_id": 6, "category_id": 1, "bbox": [856.287841796875, 306.77264404296875, 258.304443359375, 413.22735595703125], "score": 0.11297232657670975}, {"image_id": 6, "category_id": 1, "bbox": [800.4462890625, 172.54244995117188, 479.5537109375, 547.4575500488281], "score": 0.09734354168176651}, {"image_id": 6, "category_id": 1, "bbox": [809.2649536132812, 0.0, 470.73504638671875, 585.9813232421875], "score": 0.09273109585046768}, {"image_id": 6, "category_id": 1, "bbox": [1059.9739990234375, 0.0, 220.0260009765625, 386.0355224609375], "score": 0.09119664132595062}, {"image_id": 6, "category_id": 1, "bbox": [395.5287170410156, 0.0, 601.3409729003906, 584.2000732421875], "score": 0.08661521971225739}, {"image_id": 6, "category_id": 1, "bbox": [568.7140502929688, 370.76666259765625, 221.784423828125, 349.23333740234375], "score": 0.0863901674747467}, {"image_id": 6, "category_id": 1, "bbox": [526.452880859375, 90.46128845214844, 570.870361328125, 629.5387115478516], "score": 0.08026156574487686}, {"image_id": 6, "category_id": 1, "bbox": [916.8757934570312, 228.50283813476562, 242.80523681640625, 445.9043273925781], "score": 0.07866501808166504}, {"image_id": 6, "category_id": 1, "bbox": [129.92172241210938, 0.0, 635.9315490722656, 572.5153198242188], "score": 0.06521974503993988}, {"image_id": 6, "category_id": 1, "bbox": [399.91168212890625, 172.89259338378906, 545.3095092773438, 547.1074066162109], "score": 0.06293633580207825}, {"image_id": 6, "category_id": 1, "bbox": [459.8975830078125, 134.07974243164062, 317.45465087890625, 427.6603698730469], "score": 0.06238844245672226}, {"image_id": 6, "category_id": 1, "bbox": [209.79718017578125, 40.33187484741211, 241.13436889648438, 364.63241958618164], "score": 0.06163594871759415}, {"image_id": 6, "category_id": 1, "bbox": [434.5926513671875, 364.21771240234375, 264.0972900390625, 355.78228759765625], "score": 0.05956871807575226}, {"image_id": 6, "category_id": 1, "bbox": [97.56027221679688, 180.53509521484375, 661.4093322753906, 539.4649047851562], "score": 0.0561634786427021}, {"image_id": 6, "category_id": 1, "bbox": [843.3858032226562, 147.52825927734375, 208.11468505859375, 404.76678466796875], "score": 0.05534004792571068}, {"image_id": 6, "category_id": 1, "bbox": [536.8089599609375, 72.97591400146484, 256.58648681640625, 407.6761245727539], "score": 0.05499788746237755}, {"image_id": 6, "category_id": 1, "bbox": [888.5184936523438, 118.5954360961914, 210.44354248046875, 388.9878463745117], "score": 0.049914274364709854}, {"image_id": 6, "category_id": 1, "bbox": [910.0924682617188, 0.0, 294.40802001953125, 416.4046325683594], "score": 0.046580828726291656}, {"image_id": 6, "category_id": 1, "bbox": [656.02783203125, 0.0, 603.648681640625, 368.0651550292969], "score": 0.044743482023477554}, {"image_id": 6, "category_id": 1, "bbox": [14.65924072265625, 0.0, 625.1106567382812, 363.7034912109375], "score": 0.04195987805724144}, {"image_id": 6, "category_id": 1, "bbox": [266.24774169921875, 0.0, 628.0889282226562, 369.7441101074219], "score": 0.04093943163752556}, {"image_id": 6, "category_id": 1, "bbox": [0.0, 184.1331787109375, 494.2915954589844, 535.8668212890625], "score": 0.03971913084387779}, {"image_id": 6, "category_id": 1, "bbox": [243.65887451171875, 303.9045715332031, 646.0878295898438, 416.0954284667969], "score": 0.03893280029296875}, {"image_id": 6, "category_id": 1, "bbox": [912.9427490234375, 338.263671875, 307.8837890625, 381.736328125], "score": 0.03870004042983055}, {"image_id": 6, "category_id": 2, "bbox": [429.57159423828125, 290.2535705566406, 285.2738037109375, 429.7464294433594], "score": 0.07850144803524017}, {"image_id": 6, "category_id": 2, "bbox": [717.9749145507812, 330.6396484375, 252.09417724609375, 385.2900390625], "score": 0.06817291676998138}, {"image_id": 6, "category_id": 2, "bbox": [193.1670379638672, 41.12086486816406, 254.28199768066406, 361.0845489501953], "score": 0.06385660916566849}, {"image_id": 6, "category_id": 2, "bbox": [891.957763671875, 85.87422180175781, 212.38427734375, 372.2902374267578], "score": 0.04998211935162544}, {"image_id": 6, "category_id": 2, "bbox": [260.38427734375, 49.71299743652344, 287.51983642578125, 349.5011444091797], "score": 0.04769511520862579}, {"image_id": 6, "category_id": 2, "bbox": [871.7845458984375, 190.12167358398438, 225.8309326171875, 454.2715759277344], "score": 0.04754742980003357}, {"image_id": 6, "category_id": 2, "bbox": [732.2869873046875, 193.60488891601562, 263.54925537109375, 455.8025817871094], "score": 0.046569179743528366}, {"image_id": 6, "category_id": 2, "bbox": [214.9967498779297, 310.3576965332031, 295.81724548339844, 409.6423034667969], "score": 0.044801224023103714}, {"image_id": 6, "category_id": 2, "bbox": [414.4060974121094, 74.32221984863281, 290.8960266113281, 399.18067932128906], "score": 0.04291963204741478}, {"image_id": 6, "category_id": 2, "bbox": [905.2745361328125, 38.91288757324219, 270.8660888671875, 397.9369354248047], "score": 0.04242687672376633}, {"image_id": 6, "category_id": 2, "bbox": [841.3833618164062, 118.849853515625, 212.81243896484375, 380.1376647949219], "score": 0.04173169657588005}, {"image_id": 6, "category_id": 2, "bbox": [472.490478515625, 409.0013122558594, 277.50738525390625, 310.9986877441406], "score": 0.041416238993406296}, {"image_id": 6, "category_id": 2, "bbox": [536.770263671875, 269.4045715332031, 262.9224853515625, 445.5287780761719], "score": 0.041279640048742294}, {"image_id": 6, "category_id": 2, "bbox": [110.10400390625, 64.88662719726562, 287.9081115722656, 392.0303649902344], "score": 0.040577106177806854}, {"image_id": 6, "category_id": 2, "bbox": [210.54856872558594, 86.44107818603516, 270.25523376464844, 425.54552459716797], "score": 0.04053948074579239}, {"image_id": 6, "category_id": 2, "bbox": [1038.1358642578125, 84.25052642822266, 241.8641357421875, 388.01854705810547], "score": 0.040156442672014236}, {"image_id": 6, "category_id": 2, "bbox": [148.4224853515625, 271.4440612792969, 305.23858642578125, 431.8578796386719], "score": 0.03962787613272667}, {"image_id": 6, "category_id": 2, "bbox": [334.31036376953125, 70.24475860595703, 306.74700927734375, 391.44921112060547], "score": 0.03928137943148613}, {"image_id": 6, "category_id": 2, "bbox": [818.9022216796875, 213.0418243408203, 207.84423828125, 473.7554168701172], "score": 0.03923914209008217}, {"image_id": 6, "category_id": 2, "bbox": [381.75213623046875, 378.2670593261719, 259.62603759765625, 341.7329406738281], "score": 0.038922470062971115}, {"image_id": 6, "category_id": 3, "bbox": [425.3095703125, 309.4581298828125, 288.5113525390625, 410.5418701171875], "score": 0.262311726808548}, {"image_id": 6, "category_id": 3, "bbox": [205.91375732421875, 13.700637817382812, 243.03091430664062, 356.3284454345703], "score": 0.22856993973255157}, {"image_id": 6, "category_id": 3, "bbox": [713.9786987304688, 251.47422790527344, 312.60845947265625, 468.52577209472656], "score": 0.20784343779087067}, {"image_id": 6, "category_id": 3, "bbox": [560.8348388671875, 313.7969665527344, 241.9625244140625, 406.2030334472656], "score": 0.16312475502490997}, {"image_id": 6, "category_id": 3, "bbox": [719.1908569335938, 378.21649169921875, 249.84521484375, 341.78350830078125], "score": 0.1290975958108902}, {"image_id": 6, "category_id": 3, "bbox": [818.9022216796875, 213.0418243408203, 207.84423828125, 473.7554168701172], "score": 0.11350775510072708}, {"image_id": 6, "category_id": 3, "bbox": [297.7001953125, 172.6641845703125, 534.0914306640625, 547.3358154296875], "score": 0.11280528455972672}, {"image_id": 6, "category_id": 3, "bbox": [381.9609069824219, 238.135498046875, 595.6600036621094, 481.864501953125], "score": 0.10487653315067291}, {"image_id": 6, "category_id": 3, "bbox": [622.3531494140625, 339.7597351074219, 229.49725341796875, 380.2402648925781], "score": 0.10260099172592163}, {"image_id": 6, "category_id": 3, "bbox": [780.064697265625, 353.671875, 239.752197265625, 366.328125], "score": 0.0970265194773674}, {"image_id": 6, "category_id": 3, "bbox": [871.7845458984375, 190.12167358398438, 225.8309326171875, 454.2715759277344], "score": 0.09454332292079926}, {"image_id": 6, "category_id": 3, "bbox": [81.80694580078125, 245.64306640625, 701.6143188476562, 474.35693359375], "score": 0.08700421452522278}, {"image_id": 6, "category_id": 3, "bbox": [521.3284912109375, 162.7459716796875, 574.43505859375, 557.2540283203125], "score": 0.0857013389468193}, {"image_id": 6, "category_id": 3, "bbox": [260.38427734375, 49.71299743652344, 287.51983642578125, 349.5011444091797], "score": 0.08287009596824646}, {"image_id": 6, "category_id": 3, "bbox": [891.957763671875, 85.87422180175781, 212.38427734375, 372.2902374267578], "score": 0.08067719638347626}, {"image_id": 6, "category_id": 3, "bbox": [210.54856872558594, 86.44107818603516, 270.25523376464844, 425.54552459716797], "score": 0.0792476162314415}, {"image_id": 6, "category_id": 3, "bbox": [381.75213623046875, 378.2670593261719, 259.62603759765625, 341.7329406738281], "score": 0.07711130380630493}, {"image_id": 6, "category_id": 3, "bbox": [413.1280822753906, 2.032745361328125, 541.4817810058594, 675.5248718261719], "score": 0.0729210302233696}, {"image_id": 6, "category_id": 3, "bbox": [592.0294799804688, 210.34437561035156, 252.17401123046875, 443.3075408935547], "score": 0.07218067348003387}, {"image_id": 6, "category_id": 3, "bbox": [674.793701171875, 168.93099975585938, 550.854736328125, 551.0690002441406], "score": 0.06971462070941925}, {"image_id": 6, "category_id": 3, "bbox": [3.283203125, 5.3186187744140625, 609.5269775390625, 665.8978729248047], "score": 0.06447387486696243}, {"image_id": 6, "category_id": 3, "bbox": [265.24359130859375, 0.0, 606.7368774414062, 578.953369140625], "score": 0.06261224299669266}, {"image_id": 6, "category_id": 3, "bbox": [916.8757934570312, 228.50283813476562, 242.80523681640625, 445.9043273925781], "score": 0.060289591550827026}, {"image_id": 6, "category_id": 3, "bbox": [905.2745361328125, 38.91288757324219, 270.8660888671875, 397.9369354248047], "score": 0.05707626789808273}, {"image_id": 6, "category_id": 3, "bbox": [809.2649536132812, 0.0, 470.73504638671875, 585.9813232421875], "score": 0.05707518383860588}, {"image_id": 6, "category_id": 3, "bbox": [329.31072998046875, 0.0, 302.28057861328125, 388.0342712402344], "score": 0.05462457239627838}, {"image_id": 6, "category_id": 3, "bbox": [1018.1092529296875, 133.72325134277344, 203.4078369140625, 358.2541961669922], "score": 0.0540834441781044}, {"image_id": 6, "category_id": 3, "bbox": [216.6400146484375, 351.0606994628906, 292.1260070800781, 368.9393005371094], "score": 0.05303405225276947}, {"image_id": 6, "category_id": 3, "bbox": [0.0, 184.1331787109375, 494.2915954589844, 535.8668212890625], "score": 0.052502550184726715}, {"image_id": 6, "category_id": 3, "bbox": [856.287841796875, 306.77264404296875, 258.304443359375, 413.22735595703125], "score": 0.05247155949473381}, {"image_id": 6, "category_id": 3, "bbox": [472.490478515625, 409.0013122558594, 277.50738525390625, 310.9986877441406], "score": 0.05199763551354408}, {"image_id": 6, "category_id": 3, "bbox": [843.3858032226562, 147.52825927734375, 208.11468505859375, 404.76678466796875], "score": 0.05191649869084358}, {"image_id": 6, "category_id": 3, "bbox": [110.11699676513672, 23.88450050354004, 293.6206741333008, 397.9227809906006], "score": 0.05121239647269249}, {"image_id": 6, "category_id": 3, "bbox": [926.7747192382812, 159.368408203125, 222.42718505859375, 372.406005859375], "score": 0.05020151287317276}, {"image_id": 6, "category_id": 3, "bbox": [542.9820556640625, 0.0, 588.498779296875, 582.3479614257812], "score": 0.05018999055027962}, {"image_id": 6, "category_id": 3, "bbox": [557.3597412109375, 136.1082763671875, 235.44110107421875, 427.4346923828125], "score": 0.048441290855407715}, {"image_id": 6, "category_id": 3, "bbox": [832.282958984375, 48.93588638305664, 218.0028076171875, 374.3034324645996], "score": 0.04531622678041458}, {"image_id": 6, "category_id": 3, "bbox": [874.2672729492188, 168.25421142578125, 405.73272705078125, 551.7457885742188], "score": 0.04493051767349243}, {"image_id": 6, "category_id": 3, "bbox": [1038.1358642578125, 84.25052642822266, 241.8641357421875, 388.01854705810547], "score": 0.04405749589204788}, {"image_id": 6, "category_id": 3, "bbox": [510.9053039550781, 380.81280517578125, 639.5650329589844, 339.18719482421875], "score": 0.04021564498543739}, {"image_id": 6, "category_id": 3, "bbox": [0.0, 0.0, 660.2454833984375, 417.9510498046875], "score": 0.04002097249031067}, {"image_id": 6, "category_id": 3, "bbox": [464.74639892578125, 71.07989501953125, 304.58740234375, 408.3144226074219], "score": 0.03925594687461853}, {"image_id": 7, "category_id": 1, "bbox": [522.843994140625, 177.84120178222656, 327.021728515625, 473.42930603027344], "score": 0.810905933380127}, {"image_id": 7, "category_id": 1, "bbox": [778.442626953125, 361.8348388671875, 242.3804931640625, 358.1651611328125], "score": 0.5694953203201294}, {"image_id": 7, "category_id": 1, "bbox": [639.870849609375, 184.46751403808594, 266.36273193359375, 462.2499542236328], "score": 0.334099680185318}, {"image_id": 7, "category_id": 1, "bbox": [445.5447692871094, 0.0, 512.0068054199219, 615.5057373046875], "score": 0.18087266385555267}, {"image_id": 7, "category_id": 1, "bbox": [730.8032836914062, 247.83953857421875, 309.58087158203125, 472.16046142578125], "score": 0.14590883255004883}, {"image_id": 7, "category_id": 1, "bbox": [512.9580078125, 254.42291259765625, 265.53057861328125, 444.053466796875], "score": 0.12160052359104156}, {"image_id": 7, "category_id": 1, "bbox": [427.2283630371094, 170.4464874267578, 524.8873596191406, 549.5535125732422], "score": 0.1178697869181633}, {"image_id": 7, "category_id": 1, "bbox": [309.0687255859375, 40.086551666259766, 215.92291259765625, 401.6600914001465], "score": 0.10972802340984344}, {"image_id": 7, "category_id": 1, "bbox": [262.1940002441406, 34.48618698120117, 195.04763793945312, 408.2174873352051], "score": 0.10231529176235199}, {"image_id": 7, "category_id": 1, "bbox": [3.715423583984375, 11.931461334228516, 624.9228820800781, 652.4679527282715], "score": 0.09800948947668076}, {"image_id": 7, "category_id": 1, "bbox": [884.4578857421875, 194.41122436523438, 220.8170166015625, 439.9580383300781], "score": 0.09724008291959763}, {"image_id": 7, "category_id": 1, "bbox": [824.3033447265625, 368.6164245605469, 315.0167236328125, 351.3835754394531], "score": 0.09695769101381302}, {"image_id": 7, "category_id": 1, "bbox": [584.2323608398438, 294.058349609375, 282.48809814453125, 423.47149658203125], "score": 0.09571079909801483}, {"image_id": 7, "category_id": 1, "bbox": [494.826416015625, 133.13856506347656, 289.37261962890625, 447.7872772216797], "score": 0.09136433154344559}, {"image_id": 7, "category_id": 1, "bbox": [806.3973999023438, 188.29254150390625, 237.89031982421875, 449.7236328125], "score": 0.09000396728515625}, {"image_id": 7, "category_id": 1, "bbox": [311.63323974609375, 19.810340881347656, 540.438720703125, 671.5709457397461], "score": 0.08505669981241226}, {"image_id": 7, "category_id": 1, "bbox": [674.633056640625, 328.5689697265625, 290.8658447265625, 391.4310302734375], "score": 0.08347909152507782}, {"image_id": 7, "category_id": 1, "bbox": [526.7337036132812, 51.6053581237793, 319.0120849609375, 450.95329666137695], "score": 0.07708267122507095}, {"image_id": 7, "category_id": 1, "bbox": [528.7597045898438, 380.737060546875, 297.373291015625, 339.262939453125], "score": 0.06964173167943954}, {"image_id": 7, "category_id": 1, "bbox": [427.8203125, 186.9624481201172, 289.630615234375, 452.59620666503906], "score": 0.06572794914245605}, {"image_id": 7, "category_id": 1, "bbox": [945.6814575195312, 172.93392944335938, 204.49945068359375, 385.1662292480469], "score": 0.06370960175991058}, {"image_id": 7, "category_id": 1, "bbox": [535.8873291015625, 100.18126678466797, 531.7508544921875, 619.818733215332], "score": 0.06339125335216522}, {"image_id": 7, "category_id": 1, "bbox": [301.47540283203125, 0.0, 238.073486328125, 314.41986083984375], "score": 0.05803629755973816}, {"image_id": 7, "category_id": 1, "bbox": [895.14404296875, 298.97119140625, 302.1351318359375, 421.02880859375], "score": 0.0564383864402771}, {"image_id": 7, "category_id": 1, "bbox": [638.8367309570312, 0.0, 616.5746459960938, 580.903564453125], "score": 0.05629732087254524}, {"image_id": 7, "category_id": 1, "bbox": [121.99945068359375, 0.0, 660.9404907226562, 576.0885620117188], "score": 0.05100619047880173}, {"image_id": 7, "category_id": 1, "bbox": [939.4639892578125, 37.78023147583008, 212.6220703125, 412.1843376159668], "score": 0.0508987195789814}, {"image_id": 7, "category_id": 1, "bbox": [899.6971435546875, 80.30847930908203, 210.2349853515625, 384.90258026123047], "score": 0.04788746312260628}, {"image_id": 7, "category_id": 1, "bbox": [1060.2177734375, 0.0, 219.7822265625, 373.0710144042969], "score": 0.04752803221344948}, {"image_id": 7, "category_id": 1, "bbox": [1019.6378173828125, 42.70183181762695, 196.55859375, 408.1906547546387], "score": 0.04583562910556793}, {"image_id": 7, "category_id": 1, "bbox": [506.33013916015625, 378.3439636230469, 629.0087280273438, 341.6560363769531], "score": 0.04308799281716347}, {"image_id": 7, "category_id": 1, "bbox": [1034.0091552734375, 163.4327850341797, 190.4261474609375, 345.18922424316406], "score": 0.04247095063328743}, {"image_id": 7, "category_id": 1, "bbox": [629.88818359375, 155.7995147705078, 617.2288818359375, 564.2004852294922], "score": 0.04179849103093147}, {"image_id": 7, "category_id": 1, "bbox": [338.8783264160156, 54.934539794921875, 266.9826965332031, 424.19866943359375], "score": 0.04151194170117378}, {"image_id": 7, "category_id": 1, "bbox": [103.2518310546875, 318.9478759765625, 661.4342651367188, 401.0521240234375], "score": 0.04075940325856209}, {"image_id": 7, "category_id": 2, "bbox": [521.2008666992188, 310.2171936035156, 249.6707763671875, 394.4314880371094], "score": 0.1602218598127365}, {"image_id": 7, "category_id": 2, "bbox": [774.5703125, 329.8519287109375, 239.2647705078125, 382.48699951171875], "score": 0.1102508082985878}, {"image_id": 7, "category_id": 2, "bbox": [430.0217590332031, 321.9610290527344, 298.5321960449219, 376.3891906738281], "score": 0.10036588460206985}, {"image_id": 7, "category_id": 2, "bbox": [584.2323608398438, 294.058349609375, 282.48809814453125, 423.47149658203125], "score": 0.09535666555166245}, {"image_id": 7, "category_id": 2, "bbox": [522.843994140625, 177.84120178222656, 327.021728515625, 473.42930603027344], "score": 0.08202218264341354}, {"image_id": 7, "category_id": 2, "bbox": [884.4578857421875, 194.41122436523438, 220.8170166015625, 439.9580383300781], "score": 0.06776975840330124}, {"image_id": 7, "category_id": 2, "bbox": [926.75146484375, 237.88084411621094, 240.8316650390625, 420.30470275878906], "score": 0.06470651179552078}, {"image_id": 7, "category_id": 2, "bbox": [806.3973999023438, 188.29254150390625, 237.89031982421875, 449.7236328125], "score": 0.06407086551189423}, {"image_id": 7, "category_id": 2, "bbox": [899.6971435546875, 80.30847930908203, 210.2349853515625, 384.90258026123047], "score": 0.06020020693540573}, {"image_id": 7, "category_id": 2, "bbox": [945.6814575195312, 172.93392944335938, 204.49945068359375, 385.1662292480469], "score": 0.05841958895325661}, {"image_id": 7, "category_id": 2, "bbox": [822.0528564453125, 341.0125732421875, 309.125732421875, 363.02703857421875], "score": 0.057584721595048904}, {"image_id": 7, "category_id": 2, "bbox": [357.74456787109375, 277.4457702636719, 293.99676513671875, 406.3332214355469], "score": 0.05505778267979622}, {"image_id": 7, "category_id": 2, "bbox": [699.5720825195312, 367.8640441894531, 274.9591064453125, 352.1359558105469], "score": 0.05330352857708931}, {"image_id": 7, "category_id": 2, "bbox": [1029.7774658203125, 185.09007263183594, 202.06689453125, 356.6315460205078], "score": 0.05291159823536873}, {"image_id": 7, "category_id": 2, "bbox": [427.8203125, 186.9624481201172, 289.630615234375, 452.59620666503906], "score": 0.051854208111763}, {"image_id": 7, "category_id": 2, "bbox": [1027.9022216796875, 88.84107971191406, 195.1666259765625, 370.08152770996094], "score": 0.049873653799295425}, {"image_id": 7, "category_id": 2, "bbox": [991.55615234375, 239.65716552734375, 249.484375, 405.328369140625], "score": 0.0497349314391613}, {"image_id": 7, "category_id": 2, "bbox": [815.7748413085938, 73.67143249511719, 244.98785400390625, 396.30702209472656], "score": 0.04903285205364227}, {"image_id": 7, "category_id": 2, "bbox": [939.4639892578125, 37.78023147583008, 212.6220703125, 412.1843376159668], "score": 0.04702083393931389}, {"image_id": 7, "category_id": 2, "bbox": [309.3072204589844, 0.0, 209.96475219726562, 346.0223083496094], "score": 0.04650630056858063}, {"image_id": 7, "category_id": 2, "bbox": [1060.653076171875, 124.26954650878906, 219.346923828125, 365.34242248535156], "score": 0.046192657202482224}, {"image_id": 7, "category_id": 2, "bbox": [330.479736328125, 30.65621566772461, 268.5765380859375, 415.23019790649414], "score": 0.045382071286439896}, {"image_id": 7, "category_id": 2, "bbox": [269.64532470703125, 0.0, 177.04632568359375, 341.7298583984375], "score": 0.04416671395301819}, {"image_id": 7, "category_id": 2, "bbox": [1053.7572021484375, 199.8439178466797, 226.2427978515625, 404.87586975097656], "score": 0.044016383588314056}, {"image_id": 7, "category_id": 2, "bbox": [535.8873291015625, 100.18126678466797, 531.7508544921875, 619.818733215332], "score": 0.04236799478530884}, {"image_id": 7, "category_id": 2, "bbox": [642.918701171875, 82.4556884765625, 598.7584228515625, 634.781494140625], "score": 0.042188771069049835}, {"image_id": 7, "category_id": 2, "bbox": [581.2374877929688, 447.9969482421875, 322.9368896484375, 272.0030517578125], "score": 0.04089132323861122}, {"image_id": 7, "category_id": 3, "bbox": [513.51611328125, 238.11785888671875, 337.8643798828125, 475.80499267578125], "score": 0.6091630458831787}, {"image_id": 7, "category_id": 3, "bbox": [778.442626953125, 361.8348388671875, 242.3804931640625, 358.1651611328125], "score": 0.3154968023300171}, {"image_id": 7, "category_id": 3, "bbox": [523.6842651367188, 349.2694396972656, 239.65252685546875, 370.7305603027344], "score": 0.29756608605384827}, {"image_id": 7, "category_id": 3, "bbox": [591.6284790039062, 345.38836669921875, 269.35211181640625, 374.61163330078125], "score": 0.22062228620052338}, {"image_id": 7, "category_id": 3, "bbox": [430.0217590332031, 321.9610290527344, 298.5321960449219, 376.3891906738281], "score": 0.19543521106243134}, {"image_id": 7, "category_id": 3, "bbox": [730.8032836914062, 247.83953857421875, 309.58087158203125, 472.16046142578125], "score": 0.19198371469974518}, {"image_id": 7, "category_id": 3, "bbox": [646.5430297851562, 203.3941192626953, 274.52642822265625, 480.9701385498047], "score": 0.14423885941505432}, {"image_id": 7, "category_id": 3, "bbox": [674.633056640625, 328.5689697265625, 290.8658447265625, 391.4310302734375], "score": 0.14082498848438263}, {"image_id": 7, "category_id": 3, "bbox": [945.3626708984375, 80.13238525390625, 201.0537109375, 388.1072082519531], "score": 0.1328650414943695}, {"image_id": 7, "category_id": 3, "bbox": [841.4793701171875, 213.14962768554688, 293.6090087890625, 468.9776306152344], "score": 0.12994062900543213}, {"image_id": 7, "category_id": 3, "bbox": [427.2283630371094, 170.4464874267578, 524.8873596191406, 549.5535125732422], "score": 0.12925219535827637}, {"image_id": 7, "category_id": 3, "bbox": [578.6124267578125, 158.95530700683594, 285.48291015625, 462.80946350097656], "score": 0.1266169399023056}, {"image_id": 7, "category_id": 3, "bbox": [514.337646484375, 233.6387939453125, 585.683349609375, 486.3612060546875], "score": 0.11151713877916336}, {"image_id": 7, "category_id": 3, "bbox": [934.5341796875, 201.0697479248047, 224.0159912109375, 411.6237335205078], "score": 0.11066092550754547}, {"image_id": 7, "category_id": 3, "bbox": [806.3973999023438, 188.29254150390625, 237.89031982421875, 449.7236328125], "score": 0.10713295638561249}, {"image_id": 7, "category_id": 3, "bbox": [905.5654296875, 119.60369110107422, 199.8487548828125, 372.18079376220703], "score": 0.1069631278514862}, {"image_id": 7, "category_id": 3, "bbox": [1031.998291015625, 130.22055053710938, 189.00048828125, 351.1125183105469], "score": 0.10076043754816055}, {"image_id": 7, "category_id": 3, "bbox": [909.5149536132812, 271.1204833984375, 270.51776123046875, 427.2608642578125], "score": 0.09542480111122131}, {"image_id": 7, "category_id": 3, "bbox": [629.88818359375, 155.7995147705078, 617.2288818359375, 564.2004852294922], "score": 0.0905974730849266}, {"image_id": 7, "category_id": 3, "bbox": [824.3033447265625, 368.6164245605469, 315.0167236328125, 351.3835754394531], "score": 0.07538426667451859}, {"image_id": 7, "category_id": 3, "bbox": [750.9002075195312, 315.47760009765625, 529.0997924804688, 404.52239990234375], "score": 0.06876727193593979}, {"image_id": 7, "category_id": 3, "bbox": [269.64532470703125, 0.0, 177.04632568359375, 341.7298583984375], "score": 0.06799320876598358}, {"image_id": 7, "category_id": 3, "bbox": [427.8203125, 186.9624481201172, 289.630615234375, 452.59620666503906], "score": 0.06712918728590012}, {"image_id": 7, "category_id": 3, "bbox": [445.5447692871094, 0.0, 512.0068054199219, 615.5057373046875], "score": 0.06365185230970383}, {"image_id": 7, "category_id": 3, "bbox": [310.49481201171875, 99.51484680175781, 542.7959594726562, 620.4851531982422], "score": 0.06337001919746399}, {"image_id": 7, "category_id": 3, "bbox": [309.3072204589844, 0.0, 209.96475219726562, 346.0223083496094], "score": 0.06058220937848091}, {"image_id": 7, "category_id": 3, "bbox": [767.5650634765625, 16.741378784179688, 512.4349365234375, 631.7833404541016], "score": 0.05997508019208908}, {"image_id": 7, "category_id": 3, "bbox": [1059.432861328125, 82.38970184326172, 220.567138671875, 384.15088653564453], "score": 0.05920619145035744}, {"image_id": 7, "category_id": 3, "bbox": [406.2156677246094, 379.9667663574219, 599.9730529785156, 340.0332336425781], "score": 0.05429977923631668}, {"image_id": 7, "category_id": 3, "bbox": [630.9276123046875, 0.0, 629.076416015625, 494.47613525390625], "score": 0.052147381007671356}, {"image_id": 7, "category_id": 3, "bbox": [824.9594116210938, 107.75485229492188, 232.84234619140625, 396.5146484375], "score": 0.05053601786494255}, {"image_id": 7, "category_id": 3, "bbox": [1012.4347534179688, 208.18260192871094, 221.88079833984375, 382.4145050048828], "score": 0.05051179230213165}, {"image_id": 7, "category_id": 3, "bbox": [3.715423583984375, 11.931461334228516, 624.9228820800781, 652.4679527282715], "score": 0.050232287496328354}, {"image_id": 7, "category_id": 3, "bbox": [879.2562255859375, 158.1203155517578, 400.7437744140625, 561.8796844482422], "score": 0.05010835826396942}, {"image_id": 7, "category_id": 3, "bbox": [253.97061157226562, 321.9878845214844, 620.2734680175781, 398.0121154785156], "score": 0.047511033713817596}, {"image_id": 7, "category_id": 3, "bbox": [538.6010131835938, 28.392070770263672, 285.04302978515625, 423.3212471008301], "score": 0.04563326761126518}, {"image_id": 7, "category_id": 3, "bbox": [303.3406677246094, 49.57923889160156, 226.17623901367188, 427.3821563720703], "score": 0.044603247195482254}, {"image_id": 7, "category_id": 3, "bbox": [508.7242431640625, 158.34371948242188, 272.55584716796875, 459.6634826660156], "score": 0.04119155555963516}, {"image_id": 8, "category_id": 1, "bbox": [665.663818359375, 152.39283752441406, 173.102783203125, 392.58274841308594], "score": 0.7935628890991211}, {"image_id": 8, "category_id": 1, "bbox": [603.444580078125, 163.5847930908203, 276.23797607421875, 482.0481414794922], "score": 0.5158417820930481}, {"image_id": 8, "category_id": 1, "bbox": [645.0481567382812, 39.726253509521484, 275.86700439453125, 472.701114654541], "score": 0.3275265395641327}, {"image_id": 8, "category_id": 1, "bbox": [580.6118774414062, 118.05834197998047, 267.7015380859375, 401.0866165161133], "score": 0.26732927560806274}, {"image_id": 8, "category_id": 1, "bbox": [228.1268310546875, 7.659873962402344, 633.63330078125, 665.5712051391602], "score": 0.14328041672706604}, {"image_id": 8, "category_id": 1, "bbox": [0.0, 5.745334625244141, 645.560546875, 655.0135154724121], "score": 0.11333098262548447}, {"image_id": 8, "category_id": 1, "bbox": [587.3186645507812, 327.3491516113281, 257.1326904296875, 351.4235534667969], "score": 0.109589122235775}, {"image_id": 8, "category_id": 1, "bbox": [703.1038818359375, 131.08360290527344, 292.00518798828125, 431.6009063720703], "score": 0.09975525736808777}, {"image_id": 8, "category_id": 1, "bbox": [790.5262451171875, 181.62864685058594, 489.4737548828125, 538.3713531494141], "score": 0.08405900001525879}, {"image_id": 8, "category_id": 1, "bbox": [931.1613159179688, 171.046142578125, 217.45037841796875, 432.08660888671875], "score": 0.07558973878622055}, {"image_id": 8, "category_id": 1, "bbox": [485.79168701171875, 123.56275177001953, 288.66741943359375, 434.2578659057617], "score": 0.07269816100597382}, {"image_id": 8, "category_id": 1, "bbox": [384.7590026855469, 0.0, 588.5539855957031, 580.2883911132812], "score": 0.05890166759490967}, {"image_id": 8, "category_id": 1, "bbox": [1061.8720703125, 0.0, 218.1279296875, 374.499267578125], "score": 0.0538179874420166}, {"image_id": 8, "category_id": 1, "bbox": [217.96102905273438, 237.99899291992188, 674.2417297363281, 482.0010070800781], "score": 0.04796424135565758}, {"image_id": 8, "category_id": 1, "bbox": [0.0, 7.563434600830078, 363.154296875, 652.3073539733887], "score": 0.04559754207730293}, {"image_id": 8, "category_id": 1, "bbox": [543.5068969726562, 0.0, 565.7260131835938, 578.5079956054688], "score": 0.04303905740380287}, {"image_id": 8, "category_id": 1, "bbox": [510.69451904296875, 173.54978942871094, 579.1489868164062, 546.4502105712891], "score": 0.04269106686115265}, {"image_id": 8, "category_id": 1, "bbox": [0.0, 238.03318786621094, 660.3739624023438, 481.96681213378906], "score": 0.042365964502096176}, {"image_id": 8, "category_id": 1, "bbox": [168.9118194580078, 4.844078063964844, 285.22630310058594, 371.64212799072266], "score": 0.0406411811709404}, {"image_id": 8, "category_id": 1, "bbox": [768.4909057617188, 0.0, 511.50909423828125, 562.9136962890625], "score": 0.04060777649283409}, {"image_id": 8, "category_id": 1, "bbox": [377.0450439453125, 309.9543151855469, 603.254638671875, 410.0456848144531], "score": 0.04006066918373108}, {"image_id": 8, "category_id": 1, "bbox": [651.2030029296875, 0.0, 615.93359375, 361.8298034667969], "score": 0.03969964012503624}, {"image_id": 8, "category_id": 1, "bbox": [793.8764038085938, 160.30484008789062, 284.31317138671875, 438.4846496582031], "score": 0.03776444494724274}, {"image_id": 8, "category_id": 1, "bbox": [135.31094360351562, 0.0, 630.4281921386719, 361.2301330566406], "score": 0.03698703274130821}, {"image_id": 8, "category_id": 1, "bbox": [883.1241455078125, 114.16470336914062, 249.0965576171875, 396.08038330078125], "score": 0.03666641563177109}, {"image_id": 8, "category_id": 1, "bbox": [629.1734619140625, 0.0, 241.83477783203125, 422.0652770996094], "score": 0.03619374334812164}, {"image_id": 8, "category_id": 1, "bbox": [375.7569274902344, 0.0, 671.5497131347656, 361.474365234375], "score": 0.03534536436200142}, {"image_id": 8, "category_id": 2, "bbox": [587.3186645507812, 327.3491516113281, 257.1326904296875, 351.4235534667969], "score": 0.26853591203689575}, {"image_id": 8, "category_id": 2, "bbox": [479.67388916015625, 304.97381591796875, 327.68048095703125, 343.397216796875], "score": 0.19457639753818512}, {"image_id": 8, "category_id": 2, "bbox": [582.005126953125, 159.2288360595703, 266.313232421875, 402.9182586669922], "score": 0.0762784481048584}, {"image_id": 8, "category_id": 2, "bbox": [663.8770751953125, 120.42019653320312, 171.67694091796875, 408.6712951660156], "score": 0.06561380624771118}, {"image_id": 8, "category_id": 2, "bbox": [168.9118194580078, 4.844078063964844, 285.22630310058594, 371.64212799072266], "score": 0.06305363774299622}, {"image_id": 8, "category_id": 2, "bbox": [491.8997802734375, 155.90182495117188, 284.37548828125, 425.8738098144531], "score": 0.05854247882962227}, {"image_id": 8, "category_id": 2, "bbox": [394.40594482421875, 326.39251708984375, 324.7935791015625, 393.60748291015625], "score": 0.05410929024219513}, {"image_id": 8, "category_id": 2, "bbox": [645.8668212890625, 433.219970703125, 321.93115234375, 286.780029296875], "score": 0.051170408725738525}, {"image_id": 8, "category_id": 2, "bbox": [648.3655395507812, 215.671142578125, 261.50701904296875, 449.82757568359375], "score": 0.04937528073787689}, {"image_id": 8, "category_id": 2, "bbox": [253.94940185546875, 87.8854751586914, 604.6327514648438, 632.1145248413086], "score": 0.04764552414417267}, {"image_id": 8, "category_id": 2, "bbox": [855.9937744140625, 441.4383544921875, 303.3812255859375, 278.5616455078125], "score": 0.04565391317009926}, {"image_id": 8, "category_id": 2, "bbox": [418.8916015625, 211.04640197753906, 293.88104248046875, 400.89988708496094], "score": 0.04247104749083519}, {"image_id": 8, "category_id": 2, "bbox": [733.243408203125, 390.4681701660156, 287.28082275390625, 329.5318298339844], "score": 0.040410708636045456}, {"image_id": 8, "category_id": 2, "bbox": [334.345703125, 395.3023986816406, 314.244384765625, 324.6976013183594], "score": 0.039632998406887054}, {"image_id": 8, "category_id": 2, "bbox": [127.45001983642578, 73.11466217041016, 256.39400482177734, 369.5824508666992], "score": 0.039048776030540466}, {"image_id": 8, "category_id": 2, "bbox": [877.4991455078125, 77.9921646118164, 251.142333984375, 400.37877655029297], "score": 0.03759603202342987}, {"image_id": 8, "category_id": 2, "bbox": [876.625, 166.30929565429688, 271.54296875, 432.4727478027344], "score": 0.037062641233205795}, {"image_id": 8, "category_id": 2, "bbox": [95.22817993164062, 5.291770935058594, 670.6766662597656, 655.9396133422852], "score": 0.03648136183619499}, {"image_id": 8, "category_id": 2, "bbox": [256.916015625, 0.0, 346.2666015625, 324.1949768066406], "score": 0.03644473850727081}, {"image_id": 8, "category_id": 2, "bbox": [89.74346923828125, 236.83570861816406, 699.8688354492188, 483.16429138183594], "score": 0.03563397005200386}, {"image_id": 8, "category_id": 2, "bbox": [759.99169921875, 289.85394287109375, 236.22369384765625, 376.6177978515625], "score": 0.035421065986156464}, {"image_id": 8, "category_id": 2, "bbox": [645.0481567382812, 39.726253509521484, 275.86700439453125, 472.701114654541], "score": 0.03524329140782356}, {"image_id": 8, "category_id": 2, "bbox": [267.6597900390625, 361.9915466308594, 309.06182861328125, 358.0084533691406], "score": 0.034812647849321365}, {"image_id": 8, "category_id": 2, "bbox": [817.6239624023438, 6.816158294677734, 226.43743896484375, 385.2817726135254], "score": 0.03457380086183548}, {"image_id": 8, "category_id": 2, "bbox": [1061.735107421875, 79.91029357910156, 218.264892578125, 395.97120666503906], "score": 0.03414293751120567}, {"image_id": 8, "category_id": 2, "bbox": [16.963455200195312, 0.0, 324.23182678222656, 391.5501708984375], "score": 0.03395761922001839}, {"image_id": 8, "category_id": 2, "bbox": [793.5121459960938, 112.63729095458984, 267.67828369140625, 395.0856399536133], "score": 0.03381122276186943}, {"image_id": 8, "category_id": 2, "bbox": [936.3776245117188, 119.53823852539062, 215.60125732421875, 390.83636474609375], "score": 0.033398959785699844}, {"image_id": 8, "category_id": 2, "bbox": [220.51339721679688, 0.0, 668.2879943847656, 493.3916931152344], "score": 0.03337846323847771}, {"image_id": 8, "category_id": 2, "bbox": [135.31094360351562, 0.0, 630.4281921386719, 361.2301330566406], "score": 0.03331791236996651}, {"image_id": 8, "category_id": 2, "bbox": [807.5404663085938, 351.7864074707031, 256.06243896484375, 368.2135925292969], "score": 0.03286505863070488}, {"image_id": 8, "category_id": 2, "bbox": [1003.0045166015625, 84.89070129394531, 226.5458984375, 388.3848419189453], "score": 0.03283941000699997}, {"image_id": 8, "category_id": 2, "bbox": [536.8890991210938, 103.92884063720703, 535.1102905273438, 616.071159362793], "score": 0.03232428804039955}, {"image_id": 8, "category_id": 2, "bbox": [0.0, 0.0, 530.5169067382812, 573.360595703125], "score": 0.03219160437583923}, {"image_id": 8, "category_id": 3, "bbox": [573.6188354492188, 240.572021484375, 273.65655517578125, 402.4510498046875], "score": 0.5195037126541138}, {"image_id": 8, "category_id": 3, "bbox": [661.2098388671875, 162.1226043701172, 183.19122314453125, 400.48304748535156], "score": 0.3119368255138397}, {"image_id": 8, "category_id": 3, "bbox": [479.67388916015625, 304.97381591796875, 327.68048095703125, 343.397216796875], "score": 0.24355672299861908}, {"image_id": 8, "category_id": 3, "bbox": [645.0481567382812, 39.726253509521484, 275.86700439453125, 472.701114654541], "score": 0.18974509835243225}, {"image_id": 8, "category_id": 3, "bbox": [570.8430786132812, 345.7403564453125, 316.73919677734375, 374.2596435546875], "score": 0.15907122194766998}, {"image_id": 8, "category_id": 3, "bbox": [645.8896484375, 167.62088012695312, 277.21478271484375, 457.8659362792969], "score": 0.1382974535226822}, {"image_id": 8, "category_id": 3, "bbox": [230.13650512695312, 166.2631072998047, 645.1888122558594, 553.7368927001953], "score": 0.12171941250562668}, {"image_id": 8, "category_id": 3, "bbox": [168.30703735351562, 25.824737548828125, 277.3354797363281, 378.2430725097656], "score": 0.1165408194065094}, {"image_id": 8, "category_id": 3, "bbox": [491.8997802734375, 155.90182495117188, 284.37548828125, 425.8738098144531], "score": 0.11138635873794556}, {"image_id": 8, "category_id": 3, "bbox": [580.6118774414062, 118.05834197998047, 267.7015380859375, 401.0866165161133], "score": 0.0995236188173294}, {"image_id": 8, "category_id": 3, "bbox": [0.0, 5.745334625244141, 645.560546875, 655.0135154724121], "score": 0.08752159774303436}, {"image_id": 8, "category_id": 3, "bbox": [410.42987060546875, 93.65508270263672, 535.155029296875, 626.3449172973633], "score": 0.08489865809679031}, {"image_id": 8, "category_id": 3, "bbox": [622.3273315429688, 246.0663299560547, 354.77581787109375, 459.95948791503906], "score": 0.08211281150579453}, {"image_id": 8, "category_id": 3, "bbox": [699.86328125, 138.64825439453125, 306.8485107421875, 461.92254638671875], "score": 0.0789993479847908}, {"image_id": 8, "category_id": 3, "bbox": [242.5174560546875, 0.0, 639.1490478515625, 575.97900390625], "score": 0.07676305621862411}, {"image_id": 8, "category_id": 3, "bbox": [931.1613159179688, 171.046142578125, 217.45037841796875, 432.08660888671875], "score": 0.07516534626483917}, {"image_id": 8, "category_id": 3, "bbox": [375.2472839355469, 246.68218994140625, 351.2550964355469, 454.5001220703125], "score": 0.06846577674150467}, {"image_id": 8, "category_id": 3, "bbox": [790.5262451171875, 181.62864685058594, 489.4737548828125, 538.3713531494141], "score": 0.06669364869594574}, {"image_id": 8, "category_id": 3, "bbox": [0.0, 238.03318786621094, 660.3739624023438, 481.96681213378906], "score": 0.06660887598991394}, {"image_id": 8, "category_id": 3, "bbox": [510.69451904296875, 173.54978942871094, 579.1489868164062, 546.4502105712891], "score": 0.06590159237384796}, {"image_id": 8, "category_id": 3, "bbox": [931.024658203125, 79.00460815429688, 230.872314453125, 399.5826110839844], "score": 0.05484747886657715}, {"image_id": 8, "category_id": 3, "bbox": [794.1094360351562, 193.51702880859375, 296.49420166015625, 427.95947265625], "score": 0.04826638847589493}, {"image_id": 8, "category_id": 3, "bbox": [543.5068969726562, 0.0, 565.7260131835938, 578.5079956054688], "score": 0.046023812144994736}, {"image_id": 8, "category_id": 3, "bbox": [113.288818359375, 0.0, 645.3707275390625, 486.4755554199219], "score": 0.04525231942534447}, {"image_id": 8, "category_id": 3, "bbox": [649.7077026367188, 30.70452117919922, 625.2918090820312, 610.3676223754883], "score": 0.04008934274315834}, {"image_id": 8, "category_id": 3, "bbox": [0.0, 84.04225158691406, 347.22515869140625, 635.9577484130859], "score": 0.039597418159246445}, {"image_id": 8, "category_id": 3, "bbox": [767.8719482421875, 0.0, 512.1280517578125, 491.07366943359375], "score": 0.03899211436510086}, {"image_id": 8, "category_id": 3, "bbox": [865.5809936523438, 270.484375, 289.21539306640625, 400.56097412109375], "score": 0.03811891749501228}, {"image_id": 8, "category_id": 3, "bbox": [127.45001983642578, 73.11466217041016, 256.39400482177734, 369.5824508666992], "score": 0.037692222744226456}, {"image_id": 8, "category_id": 3, "bbox": [759.1818237304688, 316.54241943359375, 236.08203125, 389.36773681640625], "score": 0.03744470700621605}, {"image_id": 8, "category_id": 3, "bbox": [792.5711669921875, 75.39281463623047, 265.3968505859375, 402.7140579223633], "score": 0.03560689464211464}, {"image_id": 8, "category_id": 3, "bbox": [1003.0045166015625, 84.89070129394531, 226.5458984375, 388.3848419189453], "score": 0.03509070351719856}, {"image_id": 8, "category_id": 3, "bbox": [252.36434936523438, 10.85342788696289, 359.9450378417969, 365.3527488708496], "score": 0.03406136482954025}, {"image_id": 8, "category_id": 3, "bbox": [1061.735107421875, 79.91029357910156, 218.264892578125, 395.97120666503906], "score": 0.03344514220952988}, {"image_id": 8, "category_id": 3, "bbox": [810.4077758789062, 321.1385803222656, 249.58477783203125, 374.1233215332031], "score": 0.03322984278202057}, {"image_id": 8, "category_id": 3, "bbox": [9.353057861328125, 0.0, 635.5494079589844, 360.3963928222656], "score": 0.03306083381175995}, {"image_id": 8, "category_id": 3, "bbox": [333.5661926269531, 0.0, 726.9255065917969, 422.0791931152344], "score": 0.032958127558231354}, {"image_id": 8, "category_id": 3, "bbox": [714.070068359375, 32.730003356933594, 271.0740966796875, 419.9611587524414], "score": 0.0329093374311924}, {"image_id": 8, "category_id": 3, "bbox": [568.7703247070312, 505.9291687011719, 239.439208984375, 159.01528930664062], "score": 0.032404933124780655}, {"image_id": 9, "category_id": 1, "bbox": [889.1114501953125, 187.39031982421875, 382.9091796875, 676.3524169921875], "score": 0.7223817706108093}, {"image_id": 9, "category_id": 1, "bbox": [865.7237548828125, 370.8044738769531, 391.9989013671875, 668.3415222167969], "score": 0.25203457474708557}, {"image_id": 9, "category_id": 1, "bbox": [752.97802734375, 193.40643310546875, 421.24267578125, 657.9037475585938], "score": 0.18453121185302734}, {"image_id": 9, "category_id": 1, "bbox": [967.6717529296875, 251.79933166503906, 393.7650146484375, 692.3752288818359], "score": 0.17169229686260223}, {"image_id": 9, "category_id": 1, "bbox": [1637.3138427734375, 0.0, 281.7838134765625, 615.146728515625], "score": 0.12930940091609955}, {"image_id": 9, "category_id": 1, "bbox": [690.2908325195312, 30.261703491210938, 689.2783813476562, 983.0678253173828], "score": 0.11736828088760376}, {"image_id": 9, "category_id": 1, "bbox": [875.5936889648438, 524.7146606445312, 315.47930908203125, 555.2853393554688], "score": 0.09018765389919281}, {"image_id": 9, "category_id": 1, "bbox": [196.3014678955078, 44.26654052734375, 963.7056121826172, 947.0564575195312], "score": 0.0869138240814209}, {"image_id": 9, "category_id": 1, "bbox": [814.8283081054688, 146.5294647216797, 776.2117309570312, 933.4705352783203], "score": 0.0732828825712204}, {"image_id": 9, "category_id": 1, "bbox": [382.62506103515625, 161.43165588378906, 923.2289428710938, 918.5683441162109], "score": 0.06880498677492142}, {"image_id": 9, "category_id": 1, "bbox": [1697.9613037109375, 0.0, 222.0386962890625, 209.8900146484375], "score": 0.06348712742328644}, {"image_id": 9, "category_id": 1, "bbox": [748.1891479492188, 397.57373046875, 423.50604248046875, 625.0096435546875], "score": 0.06334269046783447}, {"image_id": 9, "category_id": 1, "bbox": [913.7741088867188, 33.03159713745117, 384.30438232421875, 663.2516670227051], "score": 0.06316301971673965}, {"image_id": 9, "category_id": 1, "bbox": [973.6824340820312, 0.0, 886.1763305664062, 878.662841796875], "score": 0.05772004276514053}, {"image_id": 9, "category_id": 1, "bbox": [192.5206298828125, 65.4250259399414, 419.83282470703125, 540.7440414428711], "score": 0.0568239800632}, {"image_id": 9, "category_id": 1, "bbox": [9.146759033203125, 231.2586669921875, 938.3722229003906, 848.7413330078125], "score": 0.05652674660086632}, {"image_id": 9, "category_id": 1, "bbox": [999.0114135742188, 465.76947021484375, 345.35601806640625, 596.8626098632812], "score": 0.053130317479372025}, {"image_id": 9, "category_id": 1, "bbox": [806.4303588867188, 0.0, 845.3193969726562, 756.9859619140625], "score": 0.0494290329515934}, {"image_id": 9, "category_id": 1, "bbox": [926.6956787109375, 217.00596618652344, 978.7991943359375, 862.9940338134766], "score": 0.04761534929275513}, {"image_id": 9, "category_id": 1, "bbox": [0.0, 52.33103942871094, 728.7782592773438, 934.3983306884766], "score": 0.04725166782736778}, {"image_id": 9, "category_id": 1, "bbox": [855.007568359375, 0.0, 354.7464599609375, 618.0314331054688], "score": 0.04514048248529434}, {"image_id": 9, "category_id": 1, "bbox": [1003.9676513671875, 2.959691047668457, 342.69873046875, 610.9883680343628], "score": 0.043780822306871414}, {"image_id": 9, "category_id": 1, "bbox": [929.9562377929688, 600.7599487304688, 342.21356201171875, 479.24005126953125], "score": 0.03989557549357414}, {"image_id": 9, "category_id": 1, "bbox": [1519.3963623046875, 0.0, 377.24755859375, 541.1200561523438], "score": 0.03928453102707863}, {"image_id": 9, "category_id": 1, "bbox": [1628.7227783203125, 0.0, 287.5455322265625, 341.0650634765625], "score": 0.038693640381097794}, {"image_id": 9, "category_id": 1, "bbox": [606.4978637695312, 355.6337585449219, 837.4956665039062, 724.3662414550781], "score": 0.038476474583148956}, {"image_id": 9, "category_id": 1, "bbox": [370.0670471191406, 596.2114868164062, 929.7160339355469, 483.78851318359375], "score": 0.03617327660322189}, {"image_id": 9, "category_id": 1, "bbox": [1345.5986328125, 12.008554458618164, 574.4013671875, 963.9489650726318], "score": 0.03453870490193367}, {"image_id": 9, "category_id": 1, "bbox": [985.373291015625, 0.0, 928.54541015625, 540.2647705078125], "score": 0.03436574712395668}, {"image_id": 9, "category_id": 2, "bbox": [869.1336059570312, 593.8689575195312, 312.81732177734375, 486.13104248046875], "score": 0.1305200159549713}, {"image_id": 9, "category_id": 2, "bbox": [918.9109497070312, 528.9691162109375, 352.72760009765625, 551.0308837890625], "score": 0.12104882299900055}, {"image_id": 9, "category_id": 2, "bbox": [874.972412109375, 219.84915161132812, 398.60888671875, 675.1905822753906], "score": 0.10471069812774658}, {"image_id": 9, "category_id": 2, "bbox": [1004.7874145507812, 599.82421875, 352.66571044921875, 480.17578125], "score": 0.10019751638174057}, {"image_id": 9, "category_id": 2, "bbox": [756.2210083007812, 468.63616943359375, 400.74957275390625, 585.0780639648438], "score": 0.09097898006439209}, {"image_id": 9, "category_id": 2, "bbox": [745.3209228515625, 232.7140655517578, 437.5478515625, 643.2607269287109], "score": 0.08049364387989044}, {"image_id": 9, "category_id": 2, "bbox": [1087.399658203125, 593.08837890625, 390.3126220703125, 486.91162109375], "score": 0.07093648612499237}, {"image_id": 9, "category_id": 2, "bbox": [1303.8785400390625, 0.0, 422.1475830078125, 468.3970031738281], "score": 0.06301257759332657}, {"image_id": 9, "category_id": 2, "bbox": [1417.6380615234375, 0.0, 406.1064453125, 474.9261779785156], "score": 0.0579981729388237}, {"image_id": 9, "category_id": 2, "bbox": [625.705078125, 484.6343688964844, 436.9879150390625, 585.0500793457031], "score": 0.04986549913883209}, {"image_id": 9, "category_id": 2, "bbox": [996.6914672851562, 398.78289794921875, 347.25286865234375, 609.4979248046875], "score": 0.04856163635849953}, {"image_id": 9, "category_id": 2, "bbox": [1206.7264404296875, 11.123605728149414, 365.9471435546875, 573.7444972991943], "score": 0.047296661883592606}, {"image_id": 9, "category_id": 2, "bbox": [1311.547119140625, 65.01437377929688, 349.19091796875, 570.0803527832031], "score": 0.0460880808532238}, {"image_id": 9, "category_id": 2, "bbox": [643.153076171875, 253.77182006835938, 380.432861328125, 621.2456359863281], "score": 0.04481982812285423}, {"image_id": 9, "category_id": 2, "bbox": [987.1060791015625, 767.1405029296875, 419.4876708984375, 312.8594970703125], "score": 0.0425034835934639}, {"image_id": 9, "category_id": 2, "bbox": [1102.4873046875, 424.0892028808594, 361.5018310546875, 578.9554748535156], "score": 0.04213244840502739}, {"image_id": 9, "category_id": 2, "bbox": [827.9692993164062, 22.665790557861328, 738.8683471679688, 977.8414726257324], "score": 0.04084686562418938}, {"image_id": 9, "category_id": 2, "bbox": [386.4203186035156, 246.82350158691406, 909.8437194824219, 833.1764984130859], "score": 0.04046650230884552}, {"image_id": 9, "category_id": 2, "bbox": [1519.3963623046875, 0.0, 377.24755859375, 541.1200561523438], "score": 0.04035676643252373}, {"image_id": 9, "category_id": 2, "bbox": [141.18504333496094, 213.68504333496094, 442.54115295410156, 581.0876007080078], "score": 0.03932327777147293}, {"image_id": 9, "category_id": 2, "bbox": [1199.660888671875, 526.3402099609375, 401.3760986328125, 553.6597900390625], "score": 0.03891701251268387}, {"image_id": 9, "category_id": 2, "bbox": [1096.899658203125, 7.249045372009277, 373.27392578125, 589.977578163147], "score": 0.03846901282668114}, {"image_id": 9, "category_id": 2, "bbox": [8.34375, 238.2019500732422, 468.5223083496094, 555.0114898681641], "score": 0.03835919126868248}, {"image_id": 9, "category_id": 2, "bbox": [954.0681762695312, 0.9164657592773438, 894.5785522460938, 977.8441543579102], "score": 0.03787016496062279}, {"image_id": 9, "category_id": 2, "bbox": [1304.5330810546875, 222.01351928710938, 351.49560546875, 590.0872497558594], "score": 0.037600286304950714}, {"image_id": 9, "category_id": 2, "bbox": [222.160400390625, 277.077880859375, 439.6475830078125, 571.2656860351562], "score": 0.0375177338719368}, {"image_id": 9, "category_id": 2, "bbox": [1413.204345703125, 116.49420166015625, 370.9608154296875, 575.35205078125], "score": 0.03735986724495888}, {"image_id": 9, "category_id": 2, "bbox": [907.872802734375, 0.0, 354.2291259765625, 433.2292175292969], "score": 0.03727457672357559}, {"image_id": 9, "category_id": 2, "bbox": [418.7904968261719, 220.21754455566406, 410.7906799316406, 589.8796234130859], "score": 0.035559963434934616}, {"image_id": 9, "category_id": 2, "bbox": [515.1683349609375, 437.50653076171875, 434.42218017578125, 582.7530517578125], "score": 0.03517492488026619}, {"image_id": 9, "category_id": 2, "bbox": [1071.5477294921875, 0.0, 449.6427001953125, 358.6690979003906], "score": 0.03480236232280731}, {"image_id": 9, "category_id": 2, "bbox": [1006.7305297851562, 0.0, 354.43902587890625, 499.3772277832031], "score": 0.03451896086335182}, {"image_id": 9, "category_id": 2, "bbox": [1210.3297119140625, 0.0, 398.09228515625, 351.8616027832031], "score": 0.03443637490272522}, {"image_id": 9, "category_id": 2, "bbox": [315.5869445800781, 216.51611328125, 407.5579528808594, 587.8486938476562], "score": 0.03434937074780464}, {"image_id": 9, "category_id": 2, "bbox": [522.8697509765625, 260.4696044921875, 414.389892578125, 613.7008056640625], "score": 0.03433384373784065}, {"image_id": 9, "category_id": 2, "bbox": [1203.7042236328125, 162.83468627929688, 370.1141357421875, 608.9102478027344], "score": 0.03426138684153557}, {"image_id": 9, "category_id": 3, "bbox": [883.216552734375, 301.9679870605469, 421.8533935546875, 684.6893005371094], "score": 0.5312616229057312}, {"image_id": 9, "category_id": 3, "bbox": [872.7015380859375, 443.1317443847656, 349.3468017578125, 626.3217468261719], "score": 0.4374043941497803}, {"image_id": 9, "category_id": 3, "bbox": [929.9562377929688, 600.7599487304688, 342.21356201171875, 479.24005126953125], "score": 0.23758870363235474}, {"image_id": 9, "category_id": 3, "bbox": [756.2210083007812, 468.63616943359375, 400.74957275390625, 585.0780639648438], "score": 0.19779707491397858}, {"image_id": 9, "category_id": 3, "bbox": [1003.01806640625, 530.2113037109375, 351.581298828125, 549.7886962890625], "score": 0.1728420853614807}, {"image_id": 9, "category_id": 3, "bbox": [961.9026489257812, 192.82447814941406, 365.77435302734375, 669.9158782958984], "score": 0.15172970294952393}, {"image_id": 9, "category_id": 3, "bbox": [742.6372680664062, 259.85797119140625, 437.62713623046875, 675.4840087890625], "score": 0.1360728144645691}, {"image_id": 9, "category_id": 3, "bbox": [1697.9613037109375, 0.0, 222.0386962890625, 209.8900146484375], "score": 0.12628720700740814}, {"image_id": 9, "category_id": 3, "bbox": [1642.754638671875, 35.96492004394531, 277.245361328125, 254.3386993408203], "score": 0.10174356400966644}, {"image_id": 9, "category_id": 3, "bbox": [690.2908325195312, 30.261703491210938, 689.2783813476562, 983.0678253173828], "score": 0.10051657259464264}, {"image_id": 9, "category_id": 3, "bbox": [1568.9835205078125, 0.0, 351.0164794921875, 182.57945251464844], "score": 0.09867986291646957}, {"image_id": 9, "category_id": 3, "bbox": [814.8283081054688, 146.5294647216797, 776.2117309570312, 933.4705352783203], "score": 0.09319436550140381}, {"image_id": 9, "category_id": 3, "bbox": [845.4409790039062, 647.54296875, 357.75726318359375, 432.45703125], "score": 0.09200572222471237}, {"image_id": 9, "category_id": 3, "bbox": [954.0681762695312, 0.9164657592773438, 894.5785522460938, 977.8441543579102], "score": 0.0784541666507721}, {"image_id": 9, "category_id": 3, "bbox": [1087.399658203125, 593.08837890625, 390.3126220703125, 486.91162109375], "score": 0.07257501035928726}, {"image_id": 9, "category_id": 3, "bbox": [386.4203186035156, 246.82350158691406, 909.8437194824219, 833.1764984130859], "score": 0.07088606059551239}, {"image_id": 9, "category_id": 3, "bbox": [192.5206298828125, 65.4250259399414, 419.83282470703125, 540.7440414428711], "score": 0.0679088905453682}, {"image_id": 9, "category_id": 3, "bbox": [1163.99951171875, 126.90815734863281, 756.00048828125, 937.6510467529297], "score": 0.06595122069120407}, {"image_id": 9, "category_id": 3, "bbox": [863.382080078125, 148.6351776123047, 398.533203125, 677.5047149658203], "score": 0.0651460736989975}, {"image_id": 9, "category_id": 3, "bbox": [934.9749755859375, 0.0, 327.4337158203125, 618.9953002929688], "score": 0.055259350687265396}, {"image_id": 9, "category_id": 3, "bbox": [1074.646728515625, 282.2994689941406, 414.116455078125, 651.0233459472656], "score": 0.052545834332704544}, {"image_id": 9, "category_id": 3, "bbox": [9.146759033203125, 231.2586669921875, 938.3722229003906, 848.7413330078125], "score": 0.05249667167663574}, {"image_id": 9, "category_id": 3, "bbox": [584.2129516601562, 457.4913024902344, 858.2162475585938, 622.5086975097656], "score": 0.04852967709302902}, {"image_id": 9, "category_id": 3, "bbox": [1155.611572265625, 0.0, 764.388427734375, 641.387939453125], "score": 0.044523682445287704}, {"image_id": 9, "category_id": 3, "bbox": [1754.9678955078125, 0.0, 165.0321044921875, 150.7854766845703], "score": 0.042292144149541855}, {"image_id": 9, "category_id": 3, "bbox": [990.1677856445312, 49.7282829284668, 367.74835205078125, 631.6631355285645], "score": 0.04215564951300621}, {"image_id": 9, "category_id": 3, "bbox": [387.43487548828125, 24.543113708496094, 467.367919921875, 634.1206436157227], "score": 0.0409880056977272}, {"image_id": 9, "category_id": 3, "bbox": [806.4303588867188, 0.0, 845.3193969726562, 756.9859619140625], "score": 0.0390448197722435}, {"image_id": 9, "category_id": 3, "bbox": [370.0670471191406, 596.2114868164062, 929.7160339355469, 483.78851318359375], "score": 0.03802560269832611}, {"image_id": 9, "category_id": 3, "bbox": [1524.9852294921875, 0.0, 377.5037841796875, 475.1366271972656], "score": 0.03740619868040085}, {"image_id": 9, "category_id": 3, "bbox": [1637.3138427734375, 0.0, 281.7838134765625, 615.146728515625], "score": 0.037343766540288925}, {"image_id": 9, "category_id": 3, "bbox": [933.290283203125, 440.8107604980469, 969.9603271484375, 639.1892395019531], "score": 0.03587207570672035}, {"image_id": 9, "category_id": 3, "bbox": [287.03021240234375, 0.0, 471.83868408203125, 528.10400390625], "score": 0.035080332309007645}, {"image_id": 9, "category_id": 3, "bbox": [1363.7510986328125, 0.0, 556.2489013671875, 860.6771850585938], "score": 0.03499571606516838}, {"image_id": 9, "category_id": 3, "bbox": [196.3014678955078, 44.26654052734375, 963.7056121826172, 947.0564575195312], "score": 0.03427572548389435}, {"image_id": 10, "category_id": 1, "bbox": [389.0790710449219, 324.2745056152344, 271.5130920410156, 393.3929748535156], "score": 0.7442577481269836}, {"image_id": 10, "category_id": 1, "bbox": [297.68072509765625, 330.6492614746094, 314.12945556640625, 380.9777526855469], "score": 0.3360668122768402}, {"image_id": 10, "category_id": 1, "bbox": [370.8835754394531, 203.0166015625, 339.3212585449219, 492.65496826171875], "score": 0.2846806049346924}, {"image_id": 10, "category_id": 1, "bbox": [579.8006591796875, 337.4202575683594, 238.719970703125, 362.9407653808594], "score": 0.23496483266353607}, {"image_id": 10, "category_id": 1, "bbox": [300.2844543457031, 2.8632431030273438, 521.7403869628906, 681.3232192993164], "score": 0.14386814832687378}, {"image_id": 10, "category_id": 1, "bbox": [531.772705078125, 3.0875701904296875, 594.0716552734375, 671.8444976806641], "score": 0.13764312863349915}, {"image_id": 10, "category_id": 1, "bbox": [467.5552673339844, 297.2355651855469, 325.1661071777344, 422.7644348144531], "score": 0.13432545959949493}, {"image_id": 10, "category_id": 1, "bbox": [18.1900634765625, 13.796905517578125, 596.4985961914062, 657.0960998535156], "score": 0.11631930619478226}, {"image_id": 10, "category_id": 1, "bbox": [785.2698974609375, 168.3120574951172, 494.7301025390625, 551.6879425048828], "score": 0.10609588027000427}, {"image_id": 10, "category_id": 1, "bbox": [393.3828430175781, 85.31024169921875, 547.6279602050781, 634.6897583007812], "score": 0.0894695594906807}, {"image_id": 10, "category_id": 1, "bbox": [169.45785522460938, 0.0, 267.36767578125, 365.1966552734375], "score": 0.08634939044713974}, {"image_id": 10, "category_id": 1, "bbox": [289.78155517578125, 312.5377502441406, 535.6991577148438, 407.4622497558594], "score": 0.08382374793291092}, {"image_id": 10, "category_id": 1, "bbox": [201.1585693359375, 288.68701171875, 316.8004150390625, 431.31298828125], "score": 0.07856669276952744}, {"image_id": 10, "category_id": 1, "bbox": [575.1112060546875, 215.41513061523438, 221.79736328125, 409.7818908691406], "score": 0.071946881711483}, {"image_id": 10, "category_id": 1, "bbox": [1056.5418701171875, 0.0, 223.4581298828125, 417.36822509765625], "score": 0.06853409856557846}, {"image_id": 10, "category_id": 1, "bbox": [639.8323974609375, 0.0, 617.7271728515625, 577.2957763671875], "score": 0.06736540049314499}, {"image_id": 10, "category_id": 1, "bbox": [568.4288330078125, 209.0952606201172, 338.16375732421875, 482.7400665283203], "score": 0.061245813965797424}, {"image_id": 10, "category_id": 1, "bbox": [898.6610717773438, 153.36146545410156, 200.79644775390625, 370.7881317138672], "score": 0.06032082810997963}, {"image_id": 10, "category_id": 1, "bbox": [492.207275390625, 297.7644958496094, 639.2918701171875, 422.2355041503906], "score": 0.0595410093665123}, {"image_id": 10, "category_id": 1, "bbox": [160.17434692382812, 88.34381103515625, 545.2200622558594, 631.6561889648438], "score": 0.05829990282654762}, {"image_id": 10, "category_id": 1, "bbox": [868.7470703125, 224.55393981933594, 266.2698974609375, 436.4813995361328], "score": 0.05756758153438568}, {"image_id": 10, "category_id": 1, "bbox": [6.046333312988281, 0.0, 195.6231460571289, 358.6008605957031], "score": 0.049494966864585876}, {"image_id": 10, "category_id": 1, "bbox": [811.5802001953125, 189.00022888183594, 272.9951171875, 430.0568389892578], "score": 0.048183657228946686}, {"image_id": 10, "category_id": 1, "bbox": [0.0, 319.09356689453125, 649.17822265625, 400.90643310546875], "score": 0.04791088029742241}, {"image_id": 10, "category_id": 1, "bbox": [917.3824462890625, 169.743896484375, 249.3902587890625, 403.61883544921875], "score": 0.0450928658246994}, {"image_id": 10, "category_id": 1, "bbox": [605.0147094726562, 0.0, 293.39044189453125, 411.45452880859375], "score": 0.04176070913672447}, {"image_id": 10, "category_id": 1, "bbox": [0.0, 8.131771087646484, 362.3641357421875, 651.5839881896973], "score": 0.04174632579088211}, {"image_id": 10, "category_id": 1, "bbox": [810.03515625, 303.3697204589844, 257.748779296875, 416.6302795410156], "score": 0.04055804759263992}, {"image_id": 10, "category_id": 1, "bbox": [461.21405029296875, 450.72076416015625, 291.03973388671875, 269.27923583984375], "score": 0.038975659757852554}, {"image_id": 10, "category_id": 1, "bbox": [4.3099365234375, 0.0, 630.94384765625, 355.42138671875], "score": 0.03738445043563843}, {"image_id": 10, "category_id": 1, "bbox": [1047.0552978515625, 153.89389038085938, 231.66552734375, 392.3066101074219], "score": 0.03562911972403526}, {"image_id": 10, "category_id": 1, "bbox": [883.27587890625, 21.453380584716797, 396.72412109375, 638.0078010559082], "score": 0.035418808460235596}, {"image_id": 10, "category_id": 1, "bbox": [391.38861083984375, 368.7728271484375, 600.3511962890625, 351.2271728515625], "score": 0.03505528345704079}, {"image_id": 10, "category_id": 1, "bbox": [921.7478637695312, 24.219051361083984, 260.03045654296875, 438.5713233947754], "score": 0.03447166830301285}, {"image_id": 10, "category_id": 1, "bbox": [829.838623046875, 113.33003234863281, 249.63818359375, 384.6968231201172], "score": 0.034223318099975586}, {"image_id": 10, "category_id": 1, "bbox": [539.3633422851562, 0.0, 621.9898071289062, 350.66632080078125], "score": 0.0341445729136467}, {"image_id": 10, "category_id": 1, "bbox": [721.650634765625, 0.0, 298.64801025390625, 409.31451416015625], "score": 0.03344011679291725}, {"image_id": 10, "category_id": 2, "bbox": [579.8006591796875, 337.4202575683594, 238.719970703125, 362.9407653808594], "score": 0.1350891888141632}, {"image_id": 10, "category_id": 2, "bbox": [373.5992431640625, 324.5570983886719, 304.6561279296875, 390.6744079589844], "score": 0.06817129254341125}, {"image_id": 10, "category_id": 2, "bbox": [575.1112060546875, 215.41513061523438, 221.79736328125, 409.7818908691406], "score": 0.06624499708414078}, {"image_id": 10, "category_id": 2, "bbox": [173.09429931640625, 28.674659729003906, 271.95831298828125, 384.85286712646484], "score": 0.049130119383335114}, {"image_id": 10, "category_id": 2, "bbox": [662.710693359375, 185.15591430664062, 273.07354736328125, 387.5263366699219], "score": 0.042818404734134674}, {"image_id": 10, "category_id": 2, "bbox": [467.5552673339844, 297.2355651855469, 325.1661071777344, 422.7644348144531], "score": 0.042173612862825394}, {"image_id": 10, "category_id": 2, "bbox": [826.733642578125, 171.38670349121094, 260.137939453125, 386.18067932128906], "score": 0.04186572879552841}, {"image_id": 10, "category_id": 2, "bbox": [852.854736328125, 428.51239013671875, 317.9669189453125, 291.48760986328125], "score": 0.041632235050201416}, {"image_id": 10, "category_id": 2, "bbox": [746.851806640625, 208.27972412109375, 274.92144775390625, 401.13555908203125], "score": 0.04153254255652428}, {"image_id": 10, "category_id": 2, "bbox": [198.08523559570312, 85.95380401611328, 287.1762390136719, 417.7829818725586], "score": 0.03966912627220154}, {"image_id": 10, "category_id": 2, "bbox": [471.72674560546875, 178.18275451660156, 296.00372314453125, 402.6244354248047], "score": 0.03835233300924301}, {"image_id": 10, "category_id": 2, "bbox": [400.72711181640625, 443.1658020019531, 298.59002685546875, 276.8341979980469], "score": 0.0380789078772068}, {"image_id": 10, "category_id": 2, "bbox": [543.4129638671875, 155.33338928222656, 265.2623291015625, 381.1384735107422], "score": 0.037433043122291565}, {"image_id": 10, "category_id": 2, "bbox": [732.1581420898438, 415.4356384277344, 274.5880126953125, 304.5643615722656], "score": 0.03662361949682236}, {"image_id": 10, "category_id": 2, "bbox": [269.05145263671875, 100.9578628540039, 306.67987060546875, 398.42818450927734], "score": 0.036257557570934296}, {"image_id": 10, "category_id": 2, "bbox": [165.4124755859375, 4.162925720214844, 534.1226196289062, 678.2037734985352], "score": 0.03536653518676758}, {"image_id": 10, "category_id": 2, "bbox": [393.3828430175781, 85.31024169921875, 547.6279602050781, 634.6897583007812], "score": 0.0348159484565258}, {"image_id": 10, "category_id": 2, "bbox": [293.46014404296875, 308.5218200683594, 309.1217041015625, 392.1636657714844], "score": 0.034752726554870605}, {"image_id": 10, "category_id": 2, "bbox": [896.1513061523438, 117.64118957519531, 198.44207763671875, 382.98268127441406], "score": 0.0346565879881382}, {"image_id": 10, "category_id": 2, "bbox": [502.25567626953125, 153.09417724609375, 604.8479614257812, 566.9058227539062], "score": 0.03422350808978081}, {"image_id": 10, "category_id": 2, "bbox": [115.18372344970703, 100.02859497070312, 268.5941696166992, 395.8730163574219], "score": 0.03333759307861328}, {"image_id": 10, "category_id": 2, "bbox": [736.26513671875, 92.42265319824219, 285.78564453125, 415.95558166503906], "score": 0.03307328745722771}, {"image_id": 10, "category_id": 2, "bbox": [818.6984252929688, 63.394630432128906, 251.44183349609375, 421.63008880615234], "score": 0.032618552446365356}, {"image_id": 10, "category_id": 2, "bbox": [226.78399658203125, 428.9842224121094, 270.03009033203125, 291.0157775878906], "score": 0.03260747715830803}, {"image_id": 10, "category_id": 2, "bbox": [812.3998413085938, 265.87847900390625, 265.65093994140625, 425.08294677734375], "score": 0.032001152634620667}, {"image_id": 10, "category_id": 2, "bbox": [880.6431274414062, 196.0501251220703, 244.78240966796875, 415.7500457763672], "score": 0.0319473035633564}, {"image_id": 10, "category_id": 3, "bbox": [578.3311767578125, 349.2962646484375, 237.2122802734375, 370.7037353515625], "score": 0.494576096534729}, {"image_id": 10, "category_id": 3, "bbox": [548.9738159179688, 218.44863891601562, 286.26715087890625, 470.0356750488281], "score": 0.28212037682533264}, {"image_id": 10, "category_id": 3, "bbox": [373.5992431640625, 324.5570983886719, 304.6561279296875, 390.6744079589844], "score": 0.27067214250564575}, {"image_id": 10, "category_id": 3, "bbox": [467.5552673339844, 297.2355651855469, 325.1661071777344, 422.7644348144531], "score": 0.13249295949935913}, {"image_id": 10, "category_id": 3, "bbox": [896.1513061523438, 117.64118957519531, 198.44207763671875, 382.98268127441406], "score": 0.10880740731954575}, {"image_id": 10, "category_id": 3, "bbox": [393.3828430175781, 85.31024169921875, 547.6279602050781, 634.6897583007812], "score": 0.10310947895050049}, {"image_id": 10, "category_id": 3, "bbox": [637.6883544921875, 168.18508911132812, 612.2178955078125, 551.8149108886719], "score": 0.10308360308408737}, {"image_id": 10, "category_id": 3, "bbox": [174.1712188720703, 6.7405242919921875, 266.13523864746094, 378.88536071777344], "score": 0.09947553277015686}, {"image_id": 10, "category_id": 3, "bbox": [296.7677001953125, 170.230712890625, 519.7216796875, 549.769287109375], "score": 0.09129945188760757}, {"image_id": 10, "category_id": 3, "bbox": [826.733642578125, 171.38670349121094, 260.137939453125, 386.18067932128906], "score": 0.08395978063344955}, {"image_id": 10, "category_id": 3, "bbox": [485.71160888671875, 224.36862182617188, 645.4370727539062, 495.6313781738281], "score": 0.07705561816692352}, {"image_id": 10, "category_id": 3, "bbox": [916.2285766601562, 64.34891510009766, 252.43377685546875, 430.0176315307617], "score": 0.07579183578491211}, {"image_id": 10, "category_id": 3, "bbox": [643.5166625976562, 229.2483367919922, 292.51153564453125, 445.35829162597656], "score": 0.07248637080192566}, {"image_id": 10, "category_id": 3, "bbox": [594.7740478515625, 179.73355102539062, 256.73358154296875, 402.4815368652344], "score": 0.07112925499677658}, {"image_id": 10, "category_id": 3, "bbox": [868.7470703125, 224.55393981933594, 266.2698974609375, 436.4813995361328], "score": 0.06753429025411606}, {"image_id": 10, "category_id": 3, "bbox": [781.571533203125, 21.847721099853516, 498.428466796875, 641.6221885681152], "score": 0.06662929803133011}, {"image_id": 10, "category_id": 3, "bbox": [165.4124755859375, 4.162925720214844, 534.1226196289062, 678.2037734985352], "score": 0.06633540987968445}, {"image_id": 10, "category_id": 3, "bbox": [145.40203857421875, 246.52053833007812, 599.8041381835938, 473.4794616699219], "score": 0.0638943538069725}, {"image_id": 10, "category_id": 3, "bbox": [531.772705078125, 3.0875701904296875, 594.0716552734375, 671.8444976806641], "score": 0.06067802011966705}, {"image_id": 10, "category_id": 3, "bbox": [812.3998413085938, 265.87847900390625, 265.65093994140625, 425.08294677734375], "score": 0.05827464535832405}, {"image_id": 10, "category_id": 3, "bbox": [757.3848266601562, 308.89361572265625, 522.6151733398438, 411.10638427734375], "score": 0.05640203505754471}, {"image_id": 10, "category_id": 3, "bbox": [746.851806640625, 208.27972412109375, 274.92144775390625, 401.13555908203125], "score": 0.05361001566052437}, {"image_id": 10, "category_id": 3, "bbox": [278.788330078125, 0.0, 576.9332275390625, 585.3538818359375], "score": 0.053244464099407196}, {"image_id": 10, "category_id": 3, "bbox": [7.01348876953125, 101.964111328125, 616.5953979492188, 618.035888671875], "score": 0.04968293383717537}, {"image_id": 10, "category_id": 3, "bbox": [818.6984252929688, 63.394630432128906, 251.44183349609375, 421.63008880615234], "score": 0.04850213974714279}, {"image_id": 10, "category_id": 3, "bbox": [400.72711181640625, 443.1658020019531, 298.59002685546875, 276.8341979980469], "score": 0.04821787402033806}, {"image_id": 10, "category_id": 3, "bbox": [391.38861083984375, 368.7728271484375, 600.3511962890625, 351.2271728515625], "score": 0.04708481580018997}, {"image_id": 10, "category_id": 3, "bbox": [725.1651611328125, 306.8961486816406, 280.68939208984375, 413.1038513183594], "score": 0.04690851271152496}, {"image_id": 10, "category_id": 3, "bbox": [479.06951904296875, 213.5166015625, 292.7242431640625, 411.220947265625], "score": 0.04583293944597244}, {"image_id": 10, "category_id": 3, "bbox": [1052.1214599609375, 76.08323669433594, 223.7564697265625, 415.9194793701172], "score": 0.04404227435588837}, {"image_id": 10, "category_id": 3, "bbox": [161.62945556640625, 87.08604431152344, 280.5509033203125, 412.05906677246094], "score": 0.04380013793706894}, {"image_id": 10, "category_id": 3, "bbox": [620.9952392578125, 0.0, 659.0047607421875, 415.83154296875], "score": 0.04279925674200058}, {"image_id": 10, "category_id": 3, "bbox": [201.1585693359375, 288.68701171875, 316.8004150390625, 431.31298828125], "score": 0.042384203523397446}, {"image_id": 10, "category_id": 3, "bbox": [0.0, 0.0, 638.43798828125, 413.6981201171875], "score": 0.03762006014585495}, {"image_id": 10, "category_id": 3, "bbox": [736.26513671875, 92.42265319824219, 285.78564453125, 415.95558166503906], "score": 0.03704670071601868}, {"image_id": 10, "category_id": 3, "bbox": [0.0, 89.72232055664062, 354.8092956542969, 630.2776794433594], "score": 0.033437248319387436}, {"image_id": 10, "category_id": 3, "bbox": [298.434326171875, 342.61175537109375, 307.6517333984375, 377.38824462890625], "score": 0.03261891379952431}, {"image_id": 11, "category_id": 1, "bbox": [793.6387329101562, 44.77378463745117, 143.69805908203125, 387.8983039855957], "score": 0.5578745007514954}, {"image_id": 11, "category_id": 1, "bbox": [711.1932983398438, 44.032447814941406, 224.10675048828125, 442.65377044677734], "score": 0.4551839232444763}, {"image_id": 11, "category_id": 1, "bbox": [786.4583129882812, 47.87508773803711, 241.32440185546875, 380.9295082092285], "score": 0.28654342889785767}, {"image_id": 11, "category_id": 1, "bbox": [769.1420288085938, 133.29940795898438, 192.7838134765625, 390.6347961425781], "score": 0.1889616698026657}, {"image_id": 11, "category_id": 1, "bbox": [712.239501953125, 174.77545166015625, 186.2564697265625, 370.8365478515625], "score": 0.13099342584609985}, {"image_id": 11, "category_id": 1, "bbox": [0.0, 1.7886085510253906, 620.8497314453125, 664.2385520935059], "score": 0.12405853718519211}, {"image_id": 11, "category_id": 1, "bbox": [250.65359497070312, 5.982845306396484, 636.0046691894531, 666.0437660217285], "score": 0.11166057735681534}, {"image_id": 11, "category_id": 1, "bbox": [763.9708251953125, 0.0, 217.5845947265625, 357.166015625], "score": 0.09830010682344437}, {"image_id": 11, "category_id": 1, "bbox": [381.2557678222656, 91.54817962646484, 599.4175720214844, 628.4518203735352], "score": 0.08873201906681061}, {"image_id": 11, "category_id": 1, "bbox": [619.0498046875, 0.0, 286.61004638671875, 409.74615478515625], "score": 0.07125087827444077}, {"image_id": 11, "category_id": 1, "bbox": [210.31129455566406, 0.0, 236.59617614746094, 353.86468505859375], "score": 0.06919298321008682}, {"image_id": 11, "category_id": 1, "bbox": [812.4999389648438, 159.4417266845703, 467.50006103515625, 560.5582733154297], "score": 0.068832628428936}, {"image_id": 11, "category_id": 1, "bbox": [91.84622192382812, 78.440185546875, 673.4509582519531, 641.559814453125], "score": 0.06656650453805923}, {"image_id": 11, "category_id": 1, "bbox": [786.0416259765625, 196.91603088378906, 269.0340576171875, 415.3472137451172], "score": 0.05511593818664551}, {"image_id": 11, "category_id": 1, "bbox": [749.3510131835938, 236.79835510253906, 228.67022705078125, 402.2859344482422], "score": 0.048651888966560364}, {"image_id": 11, "category_id": 1, "bbox": [0.0, 10.799766540527344, 359.23394775390625, 647.4769058227539], "score": 0.047336239367723465}, {"image_id": 11, "category_id": 1, "bbox": [689.7314453125, 233.7025604248047, 213.201904296875, 407.6909942626953], "score": 0.047131944447755814}, {"image_id": 11, "category_id": 1, "bbox": [841.9393310546875, 25.88986587524414, 271.529296875, 417.8709373474121], "score": 0.04279177263379097}, {"image_id": 11, "category_id": 1, "bbox": [529.36669921875, 162.9161834716797, 547.17919921875, 557.0838165283203], "score": 0.04241880029439926}, {"image_id": 11, "category_id": 1, "bbox": [376.3558044433594, 0.0, 605.0807189941406, 499.2565002441406], "score": 0.04227181151509285}, {"image_id": 11, "category_id": 1, "bbox": [919.2169189453125, 191.57882690429688, 247.582763671875, 435.2640686035156], "score": 0.04164009913802147}, {"image_id": 11, "category_id": 1, "bbox": [186.5535125732422, 63.54481887817383, 256.63575744628906, 386.8127861022949], "score": 0.04006248340010643}, {"image_id": 11, "category_id": 2, "bbox": [706.548095703125, 202.36170959472656, 192.71453857421875, 386.77684020996094], "score": 0.13562677800655365}, {"image_id": 11, "category_id": 2, "bbox": [757.0086669921875, 204.5770721435547, 220.2513427734375, 389.55201721191406], "score": 0.12697230279445648}, {"image_id": 11, "category_id": 2, "bbox": [787.6527099609375, 144.8037567138672, 247.432861328125, 379.32423400878906], "score": 0.10269901156425476}, {"image_id": 11, "category_id": 2, "bbox": [614.524169921875, 239.48757934570312, 276.04376220703125, 400.5533142089844], "score": 0.09959699958562851}, {"image_id": 11, "category_id": 2, "bbox": [548.5499267578125, 207.22146606445312, 288.7354736328125, 391.9248962402344], "score": 0.08847960084676743}, {"image_id": 11, "category_id": 2, "bbox": [639.822265625, 128.90811157226562, 258.61077880859375, 397.7748107910156], "score": 0.08713990449905396}, {"image_id": 11, "category_id": 2, "bbox": [711.1932983398438, 44.032447814941406, 224.10675048828125, 442.65377044677734], "score": 0.08265963941812515}, {"image_id": 11, "category_id": 2, "bbox": [673.3705444335938, 328.29998779296875, 242.902099609375, 379.56036376953125], "score": 0.07812044769525528}, {"image_id": 11, "category_id": 2, "bbox": [745.1964721679688, 331.04437255859375, 239.24407958984375, 376.7371826171875], "score": 0.07513707876205444}, {"image_id": 11, "category_id": 2, "bbox": [789.6646728515625, 234.23495483398438, 275.8387451171875, 413.3244934082031], "score": 0.07496144622564316}, {"image_id": 11, "category_id": 2, "bbox": [793.6387329101562, 44.77378463745117, 143.69805908203125, 387.8983039855957], "score": 0.07436255365610123}, {"image_id": 11, "category_id": 2, "bbox": [191.84173583984375, 42.33481979370117, 246.46170043945312, 364.1914863586426], "score": 0.0716821476817131}, {"image_id": 11, "category_id": 2, "bbox": [786.4583129882812, 47.87508773803711, 241.32440185546875, 380.9295082092285], "score": 0.06926408410072327}, {"image_id": 11, "category_id": 2, "bbox": [875.1549682617188, 226.61598205566406, 274.02301025390625, 428.79856872558594], "score": 0.06744860857725143}, {"image_id": 11, "category_id": 2, "bbox": [816.7042236328125, 355.6716003417969, 265.404541015625, 364.3283996582031], "score": 0.0643807202577591}, {"image_id": 11, "category_id": 2, "bbox": [862.2566528320312, 112.5934829711914, 255.68756103515625, 386.2497787475586], "score": 0.059149909764528275}, {"image_id": 11, "category_id": 2, "bbox": [859.0827026367188, 435.7423400878906, 292.17779541015625, 284.2576599121094], "score": 0.05637622997164726}, {"image_id": 11, "category_id": 2, "bbox": [620.6749267578125, 16.327417373657227, 290.6788330078125, 425.2966060638428], "score": 0.05152783915400505}, {"image_id": 11, "category_id": 2, "bbox": [934.269775390625, 171.0397491455078, 232.5654296875, 394.7007293701172], "score": 0.05053185299038887}, {"image_id": 11, "category_id": 2, "bbox": [938.1800537109375, 75.1349105834961, 231.962646484375, 392.09989166259766], "score": 0.05013241991400719}, {"image_id": 11, "category_id": 2, "bbox": [250.65359497070312, 5.982845306396484, 636.0046691894531, 666.0437660217285], "score": 0.04979142174124718}, {"image_id": 11, "category_id": 2, "bbox": [549.3214111328125, 107.98202514648438, 287.062255859375, 388.8477478027344], "score": 0.04845895990729332}, {"image_id": 11, "category_id": 2, "bbox": [109.46361541748047, 111.83271789550781, 275.68958282470703, 377.4551544189453], "score": 0.04803529381752014}, {"image_id": 11, "category_id": 2, "bbox": [266.8457336425781, 54.528079986572266, 265.4692687988281, 347.3446617126465], "score": 0.04687260091304779}, {"image_id": 11, "category_id": 2, "bbox": [381.2557678222656, 91.54817962646484, 599.4175720214844, 628.4518203735352], "score": 0.04559338092803955}, {"image_id": 11, "category_id": 2, "bbox": [259.9463806152344, 0.0, 326.3260803222656, 263.7514953613281], "score": 0.0448700487613678}, {"image_id": 11, "category_id": 2, "bbox": [531.012939453125, 0.0, 291.1553955078125, 269.7636413574219], "score": 0.04416465014219284}, {"image_id": 11, "category_id": 2, "bbox": [212.7058868408203, 90.98492431640625, 258.49937438964844, 407.0175476074219], "score": 0.0441020205616951}, {"image_id": 11, "category_id": 2, "bbox": [521.9603271484375, 322.5997009277344, 326.62841796875, 387.4262390136719], "score": 0.043646764010190964}, {"image_id": 11, "category_id": 2, "bbox": [1021.3606567382812, 86.2380599975586, 205.48004150390625, 373.6719436645508], "score": 0.04340348765254021}, {"image_id": 11, "category_id": 2, "bbox": [994.951904296875, 179.69537353515625, 238.6934814453125, 376.9566650390625], "score": 0.043371375650167465}, {"image_id": 11, "category_id": 2, "bbox": [843.751708984375, 0.0, 266.088134765625, 407.647216796875], "score": 0.04326639696955681}, {"image_id": 11, "category_id": 2, "bbox": [203.57489013671875, 0.0, 284.6365966796875, 275.9992980957031], "score": 0.0428997166454792}, {"image_id": 11, "category_id": 2, "bbox": [24.003799438476562, 79.28978729248047, 283.2124481201172, 378.4701919555664], "score": 0.040737055242061615}, {"image_id": 11, "category_id": 2, "bbox": [535.6162109375, 16.47736358642578, 540.6273193359375, 633.7865524291992], "score": 0.04055088758468628}, {"image_id": 11, "category_id": 2, "bbox": [1059.6229248046875, 119.06136322021484, 220.3770751953125, 380.90991973876953], "score": 0.039530105888843536}, {"image_id": 11, "category_id": 3, "bbox": [769.1420288085938, 133.29940795898438, 192.7838134765625, 390.6347961425781], "score": 0.4628440737724304}, {"image_id": 11, "category_id": 3, "bbox": [709.7943725585938, 137.6330108642578, 197.05853271484375, 382.3435516357422], "score": 0.4555150270462036}, {"image_id": 11, "category_id": 3, "bbox": [787.3982543945312, 53.27720260620117, 155.37750244140625, 416.9135932922363], "score": 0.3086288571357727}, {"image_id": 11, "category_id": 3, "bbox": [689.7314453125, 233.7025604248047, 213.201904296875, 407.6909942626953], "score": 0.21731334924697876}, {"image_id": 11, "category_id": 3, "bbox": [214.0520782470703, 11.216835021972656, 227.8765411376953, 363.00984954833984], "score": 0.2143668681383133}, {"image_id": 11, "category_id": 3, "bbox": [640.3848266601562, 160.651123046875, 251.390869140625, 397.68792724609375], "score": 0.2022867053747177}, {"image_id": 11, "category_id": 3, "bbox": [749.3510131835938, 236.79835510253906, 228.67022705078125, 402.2859344482422], "score": 0.17962859570980072}, {"image_id": 11, "category_id": 3, "bbox": [720.796630859375, 12.614604949951172, 210.92010498046875, 396.1550178527832], "score": 0.17814677953720093}, {"image_id": 11, "category_id": 3, "bbox": [786.4583129882812, 47.87508773803711, 241.32440185546875, 380.9295082092285], "score": 0.17768555879592896}, {"image_id": 11, "category_id": 3, "bbox": [789.5359497070312, 176.71551513671875, 249.43682861328125, 381.107421875], "score": 0.16221560537815094}, {"image_id": 11, "category_id": 3, "bbox": [380.6868591308594, 10.405426025390625, 602.2237243652344, 656.6643371582031], "score": 0.12607181072235107}, {"image_id": 11, "category_id": 3, "bbox": [624.8347778320312, 46.60826110839844, 287.212646484375, 430.30625915527344], "score": 0.11593650281429291}, {"image_id": 11, "category_id": 3, "bbox": [232.94973754882812, 80.417724609375, 660.5360412597656, 639.582275390625], "score": 0.10764239728450775}, {"image_id": 11, "category_id": 3, "bbox": [265.3182678222656, 19.485694885253906, 270.7479553222656, 351.0743637084961], "score": 0.08211584389209747}, {"image_id": 11, "category_id": 3, "bbox": [540.24609375, 99.18075561523438, 529.450927734375, 620.8192443847656], "score": 0.08209103345870972}, {"image_id": 11, "category_id": 3, "bbox": [812.4999389648438, 159.4417266845703, 467.50006103515625, 560.5582733154297], "score": 0.08085548877716064}, {"image_id": 11, "category_id": 3, "bbox": [0.0, 1.7886085510253906, 620.8497314453125, 664.2385520935059], "score": 0.07544174790382385}, {"image_id": 11, "category_id": 3, "bbox": [875.1549682617188, 226.61598205566406, 274.02301025390625, 428.79856872558594], "score": 0.07184703648090363}, {"image_id": 11, "category_id": 3, "bbox": [607.3319091796875, 284.2518310546875, 282.6614990234375, 388.13885498046875], "score": 0.07103989273309708}, {"image_id": 11, "category_id": 3, "bbox": [934.269775390625, 171.0397491455078, 232.5654296875, 394.7007293701172], "score": 0.07020433992147446}, {"image_id": 11, "category_id": 3, "bbox": [212.7058868408203, 90.98492431640625, 258.49937438964844, 407.0175476074219], "score": 0.07000125199556351}, {"image_id": 11, "category_id": 3, "bbox": [555.46484375, 169.51124572753906, 280.4832763671875, 392.46983337402344], "score": 0.0695919618010521}, {"image_id": 11, "category_id": 3, "bbox": [673.3705444335938, 328.29998779296875, 242.902099609375, 379.56036376953125], "score": 0.06693938374519348}, {"image_id": 11, "category_id": 3, "bbox": [873.6351928710938, 145.146728515625, 259.80023193359375, 385.9412841796875], "score": 0.06670709699392319}, {"image_id": 11, "category_id": 3, "bbox": [798.6546020507812, 284.19696044921875, 272.60223388671875, 397.07501220703125], "score": 0.06562507152557373}, {"image_id": 11, "category_id": 3, "bbox": [863.506103515625, 351.28948974609375, 277.0791015625, 368.71051025390625], "score": 0.061407022178173065}, {"image_id": 11, "category_id": 3, "bbox": [938.1800537109375, 75.1349105834961, 231.962646484375, 392.09989166259766], "score": 0.061358742415905}, {"image_id": 11, "category_id": 3, "bbox": [841.9393310546875, 25.88986587524414, 271.529296875, 417.8709373474121], "score": 0.061117153614759445}, {"image_id": 11, "category_id": 3, "bbox": [245.55477905273438, 0.0, 645.3086242675781, 496.8207092285156], "score": 0.0560259148478508}, {"image_id": 11, "category_id": 3, "bbox": [114.1761474609375, 0.0, 649.2318115234375, 572.5965576171875], "score": 0.0545351542532444}, {"image_id": 11, "category_id": 3, "bbox": [356.19488525390625, 235.07875061035156, 669.8319702148438, 484.92124938964844], "score": 0.05093385651707649}, {"image_id": 11, "category_id": 3, "bbox": [665.0260009765625, 18.582687377929688, 567.8739013671875, 618.4192657470703], "score": 0.05024787783622742}, {"image_id": 11, "category_id": 3, "bbox": [104.71992492675781, 16.680953979492188, 296.1352081298828, 414.3992462158203], "score": 0.048947885632514954}, {"image_id": 11, "category_id": 3, "bbox": [333.08575439453125, 0.0, 286.94757080078125, 401.19097900390625], "score": 0.04768714681267738}, {"image_id": 11, "category_id": 3, "bbox": [763.9708251953125, 0.0, 217.5845947265625, 357.166015625], "score": 0.04744655638933182}, {"image_id": 11, "category_id": 3, "bbox": [1021.3606567382812, 86.2380599975586, 205.48004150390625, 373.6719436645508], "score": 0.043542586266994476}, {"image_id": 11, "category_id": 3, "bbox": [788.1318359375, 0.0, 491.8681640625, 489.9587097167969], "score": 0.042598579078912735}, {"image_id": 11, "category_id": 3, "bbox": [535.5707397460938, 0.0, 550.4198608398438, 493.98736572265625], "score": 0.04168575629591942}, {"image_id": 11, "category_id": 3, "bbox": [70.59246826171875, 233.10952758789062, 733.9910278320312, 486.8904724121094], "score": 0.040473390370607376}, {"image_id": 11, "category_id": 3, "bbox": [365.4703369140625, 0.0, 651.92919921875, 353.3249206542969], "score": 0.040036506950855255}, {"image_id": 11, "category_id": 3, "bbox": [203.57489013671875, 0.0, 284.6365966796875, 275.9992980957031], "score": 0.03872827813029289}, {"image_id": 11, "category_id": 3, "bbox": [641.4378662109375, 0.0, 627.26513671875, 355.7915954589844], "score": 0.03799927234649658}, {"image_id": 12, "category_id": 1, "bbox": [584.5387573242188, 448.1968078613281, 497.77740478515625, 587.0398864746094], "score": 0.756415843963623}, {"image_id": 12, "category_id": 1, "bbox": [575.6952514648438, 319.00439453125, 400.10455322265625, 653.0479125976562], "score": 0.24910491704940796}, {"image_id": 12, "category_id": 1, "bbox": [749.107177734375, 369.7765197753906, 507.744873046875, 699.7013854980469], "score": 0.14256791770458221}, {"image_id": 12, "category_id": 1, "bbox": [423.25244140625, 12.339380264282227, 850.8375244140625, 998.6625118255615], "score": 0.12180855125188828}, {"image_id": 12, "category_id": 1, "bbox": [938.3679809570312, 445.4434509277344, 317.28155517578125, 562.7604675292969], "score": 0.10487187653779984}, {"image_id": 12, "category_id": 1, "bbox": [1312.2138671875, 71.54427337646484, 317.2955322265625, 665.6665420532227], "score": 0.10107286274433136}, {"image_id": 12, "category_id": 1, "bbox": [210.58319091796875, 69.52527618408203, 382.6287841796875, 546.4225387573242], "score": 0.08729560673236847}, {"image_id": 12, "category_id": 1, "bbox": [1.3367156982421875, 25.294269561767578, 941.5722808837891, 965.3807182312012], "score": 0.08666644245386124}, {"image_id": 12, "category_id": 1, "bbox": [615.8834838867188, 133.17747497558594, 812.3528442382812, 946.8225250244141], "score": 0.07625043392181396}, {"image_id": 12, "category_id": 1, "bbox": [1629.343994140625, 0.0, 290.656005859375, 622.9232788085938], "score": 0.07619617879390717}, {"image_id": 12, "category_id": 1, "bbox": [212.35372924804688, 143.08282470703125, 911.4363098144531, 936.9171752929688], "score": 0.06915077567100525}, {"image_id": 12, "category_id": 1, "bbox": [511.025390625, 529.1082153320312, 468.38751220703125, 550.8917846679688], "score": 0.06796782463788986}, {"image_id": 12, "category_id": 1, "bbox": [879.9591064453125, 625.5614013671875, 427.17333984375, 454.4385986328125], "score": 0.06433165073394775}, {"image_id": 12, "category_id": 1, "bbox": [400.519775390625, 470.6707763671875, 879.107421875, 609.3292236328125], "score": 0.05502168461680412}, {"image_id": 12, "category_id": 1, "bbox": [656.6890869140625, 215.79364013671875, 498.1884765625, 707.1752319335938], "score": 0.054581817239522934}, {"image_id": 12, "category_id": 1, "bbox": [0.0, 17.50044822692871, 553.5313720703125, 970.9961338043213], "score": 0.05268816649913788}, {"image_id": 12, "category_id": 1, "bbox": [826.6155395507812, 0.0, 842.4539184570312, 865.6671752929688], "score": 0.046238094568252563}, {"image_id": 12, "category_id": 1, "bbox": [1191.22216796875, 0.0, 728.77783203125, 759.3123168945312], "score": 0.045333489775657654}, {"image_id": 12, "category_id": 1, "bbox": [782.0577392578125, 219.59231567382812, 855.30224609375, 860.4076843261719], "score": 0.04510819539427757}, {"image_id": 12, "category_id": 1, "bbox": [987.031494140625, 232.63682556152344, 869.517333984375, 847.3631744384766], "score": 0.041379477828741074}, {"image_id": 12, "category_id": 1, "bbox": [1696.134521484375, 0.0, 223.865478515625, 197.80824279785156], "score": 0.04018330201506615}, {"image_id": 12, "category_id": 1, "bbox": [0.0, 374.3476867675781, 991.8929443359375, 705.6523132324219], "score": 0.04013173282146454}, {"image_id": 12, "category_id": 1, "bbox": [1330.46826171875, 244.0127716064453, 589.53173828125, 835.9872283935547], "score": 0.03922770917415619}, {"image_id": 12, "category_id": 1, "bbox": [613.7601928710938, 0.0, 899.5416870117188, 755.4652099609375], "score": 0.03913819417357445}, {"image_id": 12, "category_id": 1, "bbox": [365.0588684082031, 0.0, 983.5356140136719, 644.54443359375], "score": 0.03741612285375595}, {"image_id": 12, "category_id": 1, "bbox": [1314.6383056640625, 282.92706298828125, 333.597412109375, 632.1914672851562], "score": 0.0366983488202095}, {"image_id": 12, "category_id": 1, "bbox": [1499.5968017578125, 0.0, 389.786865234375, 561.987060546875], "score": 0.036690179258584976}, {"image_id": 12, "category_id": 1, "bbox": [128.3730926513672, 88.61483764648438, 472.3463897705078, 703.7524719238281], "score": 0.03647608309984207}, {"image_id": 12, "category_id": 1, "bbox": [20.4525146484375, 595.5023193359375, 323.2842102050781, 484.4976806640625], "score": 0.03625425323843956}, {"image_id": 12, "category_id": 1, "bbox": [979.1226196289062, 0.0, 902.2715454101562, 540.760009765625], "score": 0.03460397198796272}, {"image_id": 12, "category_id": 1, "bbox": [389.6744384765625, 418.8507385253906, 517.5331420898438, 661.1492614746094], "score": 0.03444761782884598}, {"image_id": 12, "category_id": 1, "bbox": [0.0, 0.0, 948.7913818359375, 542.5052490234375], "score": 0.033912550657987595}, {"image_id": 12, "category_id": 1, "bbox": [617.5381469726562, 816.321044921875, 351.2027587890625, 237.97998046875], "score": 0.03331560268998146}, {"image_id": 12, "category_id": 2, "bbox": [541.2962036132812, 661.6316528320312, 573.4873657226562, 418.36834716796875], "score": 0.2944899797439575}, {"image_id": 12, "category_id": 2, "bbox": [758.9971923828125, 650.1434326171875, 549.228759765625, 429.8565673828125], "score": 0.20387965440750122}, {"image_id": 12, "category_id": 2, "bbox": [949.1124267578125, 507.5289001464844, 313.06591796875, 530.2667541503906], "score": 0.17492473125457764}, {"image_id": 12, "category_id": 2, "bbox": [963.5787963867188, 568.38525390625, 420.40521240234375, 511.61474609375], "score": 0.12331625819206238}, {"image_id": 12, "category_id": 2, "bbox": [1076.1163330078125, 558.4424438476562, 412.7891845703125, 521.5575561523438], "score": 0.09299363940954208}, {"image_id": 12, "category_id": 2, "bbox": [970.0588989257812, 377.2192687988281, 407.92681884765625, 593.8536682128906], "score": 0.07303420454263687}, {"image_id": 12, "category_id": 2, "bbox": [445.9786682128906, 656.8173217773438, 438.2475891113281, 423.18267822265625], "score": 0.06459607183933258}, {"image_id": 12, "category_id": 2, "bbox": [657.1699829101562, 456.80523681640625, 499.50494384765625, 608.5025024414062], "score": 0.0626990795135498}, {"image_id": 12, "category_id": 2, "bbox": [1321.7589111328125, 61.96654510498047, 297.1748046875, 621.1035842895508], "score": 0.060950856655836105}, {"image_id": 12, "category_id": 2, "bbox": [1211.5887451171875, 546.5008544921875, 367.0682373046875, 533.4991455078125], "score": 0.05962763726711273}, {"image_id": 12, "category_id": 2, "bbox": [1361.6451416015625, 55.82767868041992, 392.0062255859375, 623.3733711242676], "score": 0.048995036631822586}, {"image_id": 12, "category_id": 2, "bbox": [915.3291015625, 306.80596923828125, 361.397216796875, 638.5580444335938], "score": 0.04844856634736061}, {"image_id": 12, "category_id": 2, "bbox": [576.2388916015625, 409.4638671875, 412.06591796875, 583.6846923828125], "score": 0.04771701619029045}, {"image_id": 12, "category_id": 2, "bbox": [1089.3013916015625, 365.08074951171875, 408.4337158203125, 613.6117553710938], "score": 0.04565882682800293}, {"image_id": 12, "category_id": 2, "bbox": [297.6064453125, 653.34375, 499.4632568359375, 426.65625], "score": 0.04482521489262581}, {"image_id": 12, "category_id": 2, "bbox": [100.96220397949219, 663.0093383789062, 487.3482208251953, 416.99066162109375], "score": 0.04296933859586716}, {"image_id": 12, "category_id": 2, "bbox": [1263.8453369140625, 192.77976989746094, 344.5308837890625, 610.1031036376953], "score": 0.04245179519057274}, {"image_id": 12, "category_id": 2, "bbox": [203.8378143310547, 0.0, 369.76808166503906, 550.9328002929688], "score": 0.04202678054571152}, {"image_id": 12, "category_id": 2, "bbox": [1118.8660888671875, 138.14096069335938, 409.3714599609375, 628.0019836425781], "score": 0.04164997860789299}, {"image_id": 12, "category_id": 2, "bbox": [1231.9417724609375, 3.377652168273926, 348.696044921875, 625.5098600387573], "score": 0.04057168588042259}, {"image_id": 12, "category_id": 2, "bbox": [1429.2681884765625, 0.0, 391.950439453125, 492.71368408203125], "score": 0.040231019258499146}, {"image_id": 12, "category_id": 2, "bbox": [709.5217895507812, 437.5854187011719, 960.2532348632812, 642.4145812988281], "score": 0.038646381348371506}, {"image_id": 12, "category_id": 2, "bbox": [1314.6383056640625, 282.92706298828125, 333.597412109375, 632.1914672851562], "score": 0.038616426289081573}, {"image_id": 12, "category_id": 2, "bbox": [1297.1588134765625, 537.7833251953125, 407.5615234375, 542.2166748046875], "score": 0.03687765449285507}, {"image_id": 12, "category_id": 2, "bbox": [21.873390197753906, 605.2442626953125, 437.2166976928711, 474.7557373046875], "score": 0.036657724529504776}, {"image_id": 12, "category_id": 2, "bbox": [208.83056640625, 599.34228515625, 470.3197021484375, 480.65771484375], "score": 0.03537286818027496}, {"image_id": 12, "category_id": 2, "bbox": [579.3394165039062, 239.1968994140625, 847.3419799804688, 840.8031005859375], "score": 0.03512510657310486}, {"image_id": 12, "category_id": 2, "bbox": [825.8699340820312, 43.52079391479492, 425.59295654296875, 611.4515571594238], "score": 0.03500821441411972}, {"image_id": 12, "category_id": 2, "bbox": [907.3282470703125, 85.83909606933594, 427.7047119140625, 631.6411895751953], "score": 0.03373294696211815}, {"image_id": 12, "category_id": 2, "bbox": [310.00872802734375, 224.31784057617188, 438.25579833984375, 574.5038757324219], "score": 0.03315527364611626}, {"image_id": 12, "category_id": 3, "bbox": [949.1124267578125, 507.5289001464844, 313.06591796875, 530.2667541503906], "score": 0.3932914733886719}, {"image_id": 12, "category_id": 3, "bbox": [767.5947875976562, 619.55712890625, 529.7793579101562, 460.44287109375], "score": 0.25304874777793884}, {"image_id": 12, "category_id": 3, "bbox": [552.9085083007812, 585.7274169921875, 549.4083862304688, 494.2725830078125], "score": 0.24541833996772766}, {"image_id": 12, "category_id": 3, "bbox": [657.1699829101562, 456.80523681640625, 499.50494384765625, 608.5025024414062], "score": 0.19563481211662292}, {"image_id": 12, "category_id": 3, "bbox": [1312.2138671875, 71.54427337646484, 317.2955322265625, 665.6665420532227], "score": 0.14432771503925323}, {"image_id": 12, "category_id": 3, "bbox": [192.53404235839844, 0.0, 407.4032745361328, 558.9723510742188], "score": 0.13048040866851807}, {"image_id": 12, "category_id": 3, "bbox": [659.2424926757812, 692.0776977539062, 537.0935668945312, 387.92230224609375], "score": 0.11470801383256912}, {"image_id": 12, "category_id": 3, "bbox": [963.5787963867188, 568.38525390625, 420.40521240234375, 511.61474609375], "score": 0.1111811175942421}, {"image_id": 12, "category_id": 3, "bbox": [485.5148620605469, 460.069091796875, 508.6385192871094, 619.930908203125], "score": 0.10485245287418365}, {"image_id": 12, "category_id": 3, "bbox": [753.72021484375, 317.97552490234375, 488.5916748046875, 710.6469116210938], "score": 0.10411663353443146}, {"image_id": 12, "category_id": 3, "bbox": [970.0588989257812, 377.2192687988281, 407.92681884765625, 593.8536682128906], "score": 0.09336720407009125}, {"image_id": 12, "category_id": 3, "bbox": [517.0243530273438, 699.9879150390625, 466.81011962890625, 380.0120849609375], "score": 0.09277234226465225}, {"image_id": 12, "category_id": 3, "bbox": [404.27471923828125, 362.3662109375, 893.3655395507812, 717.6337890625], "score": 0.09228695183992386}, {"image_id": 12, "category_id": 3, "bbox": [572.546875, 257.7568359375, 527.5782470703125, 729.3822631835938], "score": 0.0918555036187172}, {"image_id": 12, "category_id": 3, "bbox": [915.3291015625, 306.80596923828125, 361.397216796875, 638.5580444335938], "score": 0.07815804332494736}, {"image_id": 12, "category_id": 3, "bbox": [1647.8848876953125, 39.52253341674805, 272.1151123046875, 245.59517288208008], "score": 0.07187265902757645}, {"image_id": 12, "category_id": 3, "bbox": [1515.110595703125, 0.0, 404.889404296875, 231.63916015625], "score": 0.06878863275051117}, {"image_id": 12, "category_id": 3, "bbox": [1696.134521484375, 0.0, 223.865478515625, 197.80824279785156], "score": 0.06449871510267258}, {"image_id": 12, "category_id": 3, "bbox": [128.3730926513672, 88.61483764648438, 472.3463897705078, 703.7524719238281], "score": 0.062326688319444656}, {"image_id": 12, "category_id": 3, "bbox": [1074.5633544921875, 504.58721923828125, 415.5159912109375, 545.8851928710938], "score": 0.057570166885852814}, {"image_id": 12, "category_id": 3, "bbox": [371.5150451660156, 2.842609405517578, 508.8008117675781, 674.5514945983887], "score": 0.05682412162423134}, {"image_id": 12, "category_id": 3, "bbox": [197.62326049804688, 260.92144775390625, 939.3100891113281, 819.0785522460938], "score": 0.055999334901571274}, {"image_id": 12, "category_id": 3, "bbox": [269.88787841796875, 85.14139556884766, 461.7232666015625, 522.3687973022461], "score": 0.055146798491477966}, {"image_id": 12, "category_id": 3, "bbox": [876.803466796875, 725.79638671875, 477.71533203125, 354.20361328125], "score": 0.05181723088026047}, {"image_id": 12, "category_id": 3, "bbox": [1260.519775390625, 239.19972229003906, 335.839599609375, 610.2167205810547], "score": 0.04980340227484703}, {"image_id": 12, "category_id": 3, "bbox": [1231.9417724609375, 3.377652168273926, 348.696044921875, 625.5098600387573], "score": 0.04958599805831909}, {"image_id": 12, "category_id": 3, "bbox": [615.8834838867188, 133.17747497558594, 812.3528442382812, 946.8225250244141], "score": 0.047887831926345825}, {"image_id": 12, "category_id": 3, "bbox": [423.25244140625, 12.339380264282227, 850.8375244140625, 998.6625118255615], "score": 0.046084120869636536}, {"image_id": 12, "category_id": 3, "bbox": [1345.7681884765625, 158.47833251953125, 408.243896484375, 600.09521484375], "score": 0.0427207350730896}, {"image_id": 12, "category_id": 3, "bbox": [1040.0928955078125, 0.0, 775.478515625, 860.419921875], "score": 0.04202926903963089}, {"image_id": 12, "category_id": 3, "bbox": [906.7560424804688, 440.9178161621094, 1013.2439575195312, 639.0821838378906], "score": 0.04044117033481598}, {"image_id": 12, "category_id": 3, "bbox": [738.7144775390625, 323.0152587890625, 925.2144775390625, 756.9847412109375], "score": 0.039636388421058655}, {"image_id": 12, "category_id": 3, "bbox": [0.0, 485.9674987792969, 977.0608520507812, 594.0325012207031], "score": 0.03837479650974274}, {"image_id": 12, "category_id": 3, "bbox": [523.482421875, 582.2359008789062, 958.474853515625, 497.76409912109375], "score": 0.03790397197008133}, {"image_id": 12, "category_id": 3, "bbox": [620.9669189453125, 842.1637573242188, 351.44891357421875, 237.83624267578125], "score": 0.0359388031065464}, {"image_id": 12, "category_id": 3, "bbox": [1495.7440185546875, 0.0, 393.548828125, 497.8324890136719], "score": 0.03562435135245323}, {"image_id": 12, "category_id": 3, "bbox": [0.0, 253.5995635986328, 760.8349609375, 826.4004364013672], "score": 0.03541457653045654}, {"image_id": 13, "category_id": 1, "bbox": [1074.7713623046875, 136.86016845703125, 259.803466796875, 667.9913330078125], "score": 0.6553751230239868}, {"image_id": 13, "category_id": 1, "bbox": [1130.195068359375, 141.2550506591797, 314.416748046875, 666.0061798095703], "score": 0.5566073060035706}, {"image_id": 13, "category_id": 1, "bbox": [1033.12158203125, 253.00706481933594, 359.0496826171875, 678.7765045166016], "score": 0.2951386868953705}, {"image_id": 13, "category_id": 1, "bbox": [831.1115112304688, 391.1956481933594, 360.27960205078125, 517.3589172363281], "score": 0.17238332331180573}, {"image_id": 13, "category_id": 1, "bbox": [889.08251953125, 143.29176330566406, 446.940185546875, 724.8990936279297], "score": 0.17166927456855774}, {"image_id": 13, "category_id": 1, "bbox": [963.5154418945312, 104.7214584350586, 337.03155517578125, 637.3277969360352], "score": 0.09610675275325775}, {"image_id": 13, "category_id": 1, "bbox": [1627.4656982421875, 0.0, 290.734375, 628.128173828125], "score": 0.09153296053409576}, {"image_id": 13, "category_id": 1, "bbox": [1062.542724609375, 0.0, 297.9461669921875, 624.887939453125], "score": 0.07524855434894562}, {"image_id": 13, "category_id": 1, "bbox": [1124.8050537109375, 332.6353454589844, 340.0445556640625, 621.3121643066406], "score": 0.0751338079571724}, {"image_id": 13, "category_id": 1, "bbox": [0.0, 39.15789794921875, 918.4506225585938, 941.2945556640625], "score": 0.06607746332883835}, {"image_id": 13, "category_id": 1, "bbox": [347.1631164550781, 343.3568115234375, 992.1274108886719, 736.6431884765625], "score": 0.06480108201503754}, {"image_id": 13, "category_id": 1, "bbox": [1217.11865234375, 250.1517333984375, 702.88134765625, 829.8482666015625], "score": 0.061579711735248566}, {"image_id": 13, "category_id": 1, "bbox": [1212.784423828125, 0.0, 707.215576171875, 868.1076049804688], "score": 0.06056167557835579}, {"image_id": 13, "category_id": 1, "bbox": [840.6110229492188, 139.87380981445312, 357.87628173828125, 638.3418884277344], "score": 0.05867668241262436}, {"image_id": 13, "category_id": 1, "bbox": [397.9236145019531, 20.726308822631836, 904.2044372558594, 968.5628147125244], "score": 0.056537602096796036}, {"image_id": 13, "category_id": 1, "bbox": [1124.483154296875, 0.0, 311.80810546875, 565.9744262695312], "score": 0.054349448531866074}, {"image_id": 13, "category_id": 1, "bbox": [763.2921752929688, 289.1500549316406, 426.43572998046875, 534.4155578613281], "score": 0.050726328045129776}, {"image_id": 13, "category_id": 1, "bbox": [178.27944946289062, 139.77359008789062, 969.9604187011719, 939.8384704589844], "score": 0.050499215722084045}, {"image_id": 13, "category_id": 1, "bbox": [11.747817993164062, 0.0, 889.1437225341797, 643.8350219726562], "score": 0.050244640558958054}, {"image_id": 13, "category_id": 1, "bbox": [197.41732788085938, 91.66526794433594, 400.3829650878906, 514.3546295166016], "score": 0.04928546026349068}, {"image_id": 13, "category_id": 1, "bbox": [594.5315551757812, 141.6844482421875, 834.1611938476562, 938.3155517578125], "score": 0.04543047770857811}, {"image_id": 13, "category_id": 1, "bbox": [1697.281982421875, 0.0, 222.718017578125, 198.22459411621094], "score": 0.04076142609119415}, {"image_id": 13, "category_id": 1, "bbox": [0.0, 359.6885986328125, 997.9845581054688, 720.3114013671875], "score": 0.03825155273079872}, {"image_id": 13, "category_id": 1, "bbox": [808.2947387695312, 345.9832763671875, 815.8150024414062, 734.0167236328125], "score": 0.036845456808805466}, {"image_id": 13, "category_id": 1, "bbox": [0.0, 135.920166015625, 545.6119995117188, 944.079833984375], "score": 0.03497258201241493}, {"image_id": 13, "category_id": 2, "bbox": [821.4561767578125, 419.1445617675781, 375.6944580078125, 542.8995666503906], "score": 0.22885750234127045}, {"image_id": 13, "category_id": 2, "bbox": [811.1502685546875, 288.8132019042969, 376.37841796875, 547.8455505371094], "score": 0.18652735650539398}, {"image_id": 13, "category_id": 2, "bbox": [758.3272094726562, 149.2642364501953, 381.75836181640625, 615.4462127685547], "score": 0.12020286917686462}, {"image_id": 13, "category_id": 2, "bbox": [872.0560913085938, 23.271669387817383, 327.60540771484375, 544.5407695770264], "score": 0.10528126358985901}, {"image_id": 13, "category_id": 2, "bbox": [755.7078247070312, 14.671099662780762, 354.11431884765625, 561.5770082473755], "score": 0.09780748188495636}, {"image_id": 13, "category_id": 2, "bbox": [962.566650390625, 12.803046226501465, 297.1458740234375, 578.411431312561], "score": 0.08802969753742218}, {"image_id": 13, "category_id": 2, "bbox": [966.9734497070312, 124.4070816040039, 351.34356689453125, 676.1025009155273], "score": 0.08723946660757065}, {"image_id": 13, "category_id": 2, "bbox": [656.0617065429688, 62.502052307128906, 365.6512451171875, 574.4585800170898], "score": 0.07274441421031952}, {"image_id": 13, "category_id": 2, "bbox": [1075.3515625, 216.67401123046875, 261.7979736328125, 669.9424438476562], "score": 0.06055903062224388}, {"image_id": 13, "category_id": 2, "bbox": [1127.139892578125, 220.52235412597656, 326.8631591796875, 665.3699188232422], "score": 0.05660703405737877}, {"image_id": 13, "category_id": 2, "bbox": [1062.542724609375, 0.0, 297.9461669921875, 624.887939453125], "score": 0.05490793660283089}, {"image_id": 13, "category_id": 2, "bbox": [613.8175048828125, 380.0867614746094, 473.4202880859375, 635.3552551269531], "score": 0.05338072404265404}, {"image_id": 13, "category_id": 2, "bbox": [812.044189453125, 0.0, 438.284423828125, 414.5367431640625], "score": 0.052644871175289154}, {"image_id": 13, "category_id": 2, "bbox": [1017.9915771484375, 392.8103332519531, 361.8612060546875, 612.7197570800781], "score": 0.05171917751431465}, {"image_id": 13, "category_id": 2, "bbox": [626.480712890625, 188.63287353515625, 439.7523193359375, 634.4279174804688], "score": 0.05163908749818802}, {"image_id": 13, "category_id": 2, "bbox": [1345.4298095703125, 0.0, 366.5203857421875, 474.37896728515625], "score": 0.05092242732644081}, {"image_id": 13, "category_id": 2, "bbox": [1118.9794921875, 394.92108154296875, 348.972900390625, 610.056640625], "score": 0.04867662489414215}, {"image_id": 13, "category_id": 2, "bbox": [519.1578369140625, 262.014404296875, 447.7056884765625, 591.2722778320312], "score": 0.04586339369416237}, {"image_id": 13, "category_id": 2, "bbox": [397.9236145019531, 20.726308822631836, 904.2044372558594, 968.5628147125244], "score": 0.04516908898949623}, {"image_id": 13, "category_id": 2, "bbox": [178.27944946289062, 139.77359008789062, 969.9604187011719, 939.8384704589844], "score": 0.04314858466386795}, {"image_id": 13, "category_id": 2, "bbox": [1124.483154296875, 0.0, 311.80810546875, 565.9744262695312], "score": 0.043004997074604034}, {"image_id": 13, "category_id": 2, "bbox": [1158.1627197265625, 121.0126724243164, 392.1632080078125, 693.0181503295898], "score": 0.04253359138965607}, {"image_id": 13, "category_id": 2, "bbox": [523.4107055664062, 415.86627197265625, 447.1729736328125, 591.714599609375], "score": 0.04248499125242233}, {"image_id": 13, "category_id": 2, "bbox": [1201.449951171875, 13.144429206848145, 349.013671875, 587.8254194259644], "score": 0.04234619438648224}, {"image_id": 13, "category_id": 2, "bbox": [207.17198181152344, 47.13739776611328, 400.96253967285156, 501.11339569091797], "score": 0.042296670377254486}, {"image_id": 13, "category_id": 2, "bbox": [1005.9633178710938, 0.0, 392.75653076171875, 417.71063232421875], "score": 0.04184853658080101}, {"image_id": 13, "category_id": 2, "bbox": [1298.947265625, 521.94482421875, 395.3671875, 558.05517578125], "score": 0.040786366909742355}, {"image_id": 13, "category_id": 2, "bbox": [508.56866455078125, 61.86715316772461, 425.3851318359375, 572.4971046447754], "score": 0.039179496467113495}, {"image_id": 13, "category_id": 2, "bbox": [0.0, 34.23614501953125, 736.5712890625, 956.2137451171875], "score": 0.03873138874769211}, {"image_id": 13, "category_id": 2, "bbox": [315.70654296875, 328.93927001953125, 439.64544677734375, 584.994384765625], "score": 0.03793958202004433}, {"image_id": 13, "category_id": 2, "bbox": [118.01749420166016, 283.4339904785156, 456.74361419677734, 571.8712463378906], "score": 0.03707306832075119}, {"image_id": 13, "category_id": 2, "bbox": [8.775077819824219, 234.9263916015625, 433.6321792602539, 571.1407470703125], "score": 0.03699303790926933}, {"image_id": 13, "category_id": 2, "bbox": [594.5315551757812, 141.6844482421875, 834.1611938476562, 938.3155517578125], "score": 0.0365842804312706}, {"image_id": 13, "category_id": 2, "bbox": [1377.388916015625, 0.0, 434.25146484375, 595.5211791992188], "score": 0.03592078387737274}, {"image_id": 13, "category_id": 2, "bbox": [1344.1549072265625, 47.030616760253906, 312.815185546875, 617.4065170288086], "score": 0.03589899465441704}, {"image_id": 13, "category_id": 2, "bbox": [1224.0465087890625, 387.4518737792969, 380.960205078125, 649.6054992675781], "score": 0.03580160066485405}, {"image_id": 13, "category_id": 2, "bbox": [409.7948913574219, 216.1328582763672, 444.9522399902344, 583.7818145751953], "score": 0.03573210909962654}, {"image_id": 13, "category_id": 2, "bbox": [664.7590942382812, 0.0, 324.62591552734375, 469.1226806640625], "score": 0.03503477945923805}, {"image_id": 13, "category_id": 2, "bbox": [414.2143249511719, 376.79656982421875, 448.4488220214844, 592.0216674804688], "score": 0.03478070721030235}, {"image_id": 13, "category_id": 3, "bbox": [818.6689453125, 348.40997314453125, 370.3272705078125, 519.1138305664062], "score": 0.6983469724655151}, {"image_id": 13, "category_id": 3, "bbox": [804.944091796875, 421.5149841308594, 416.357666015625, 657.5244445800781], "score": 0.36962100863456726}, {"image_id": 13, "category_id": 3, "bbox": [840.6110229492188, 139.87380981445312, 357.87628173828125, 638.3418884277344], "score": 0.313664972782135}, {"image_id": 13, "category_id": 3, "bbox": [1033.12158203125, 253.00706481933594, 359.0496826171875, 678.7765045166016], "score": 0.3099805414676666}, {"image_id": 13, "category_id": 3, "bbox": [1074.7713623046875, 136.86016845703125, 259.803466796875, 667.9913330078125], "score": 0.27094778418540955}, {"image_id": 13, "category_id": 3, "bbox": [889.08251953125, 143.29176330566406, 446.940185546875, 724.8990936279297], "score": 0.2198207974433899}, {"image_id": 13, "category_id": 3, "bbox": [1130.195068359375, 141.2550506591797, 314.416748046875, 666.0061798095703], "score": 0.20341090857982635}, {"image_id": 13, "category_id": 3, "bbox": [755.4949951171875, 112.84942626953125, 368.728759765625, 595.9637451171875], "score": 0.13557077944278717}, {"image_id": 13, "category_id": 3, "bbox": [1124.8050537109375, 332.6353454589844, 340.0445556640625, 621.3121643066406], "score": 0.12772610783576965}, {"image_id": 13, "category_id": 3, "bbox": [598.0113525390625, 218.7879180908203, 482.9976806640625, 662.4211883544922], "score": 0.11590449512004852}, {"image_id": 13, "category_id": 3, "bbox": [872.0560913085938, 23.271669387817383, 327.60540771484375, 544.5407695770264], "score": 0.11419356614351273}, {"image_id": 13, "category_id": 3, "bbox": [962.566650390625, 12.803046226501465, 297.1458740234375, 578.411431312561], "score": 0.10777939110994339}, {"image_id": 13, "category_id": 3, "bbox": [191.04415893554688, 44.96326446533203, 411.1267395019531, 510.0557174682617], "score": 0.1074787899851799}, {"image_id": 13, "category_id": 3, "bbox": [1062.542724609375, 0.0, 297.9461669921875, 624.887939453125], "score": 0.09838511794805527}, {"image_id": 13, "category_id": 3, "bbox": [963.5154418945312, 104.7214584350586, 337.03155517578125, 637.3277969360352], "score": 0.09707893431186676}, {"image_id": 13, "category_id": 3, "bbox": [594.5315551757812, 141.6844482421875, 834.1611938476562, 938.3155517578125], "score": 0.0874122902750969}, {"image_id": 13, "category_id": 3, "bbox": [1568.3453369140625, 0.0, 351.6546630859375, 179.18341064453125], "score": 0.07589671015739441}, {"image_id": 13, "category_id": 3, "bbox": [1646.181884765625, 40.00691604614258, 273.818115234375, 244.18827438354492], "score": 0.07006432116031647}, {"image_id": 13, "category_id": 3, "bbox": [1124.483154296875, 0.0, 311.80810546875, 565.9744262695312], "score": 0.06869136542081833}, {"image_id": 13, "category_id": 3, "bbox": [748.2572631835938, 0.0, 350.39422607421875, 517.6879272460938], "score": 0.0681183859705925}, {"image_id": 13, "category_id": 3, "bbox": [1697.281982421875, 0.0, 222.718017578125, 198.22459411621094], "score": 0.0667494386434555}, {"image_id": 13, "category_id": 3, "bbox": [1236.0400390625, 11.620513916015625, 683.9599609375, 978.7690734863281], "score": 0.0651257187128067}, {"image_id": 13, "category_id": 3, "bbox": [397.9236145019531, 20.726308822631836, 904.2044372558594, 968.5628147125244], "score": 0.059572793543338776}, {"image_id": 13, "category_id": 3, "bbox": [836.6048583984375, 139.68841552734375, 754.8648681640625, 933.9765014648438], "score": 0.059398502111434937}, {"image_id": 13, "category_id": 3, "bbox": [986.154052734375, 427.1053161621094, 431.180419921875, 652.8946838378906], "score": 0.05196673795580864}, {"image_id": 13, "category_id": 3, "bbox": [347.1631164550781, 343.3568115234375, 992.1274108886719, 736.6431884765625], "score": 0.0518779382109642}, {"image_id": 13, "category_id": 3, "bbox": [1175.58349609375, 0.0, 744.41650390625, 624.3744506835938], "score": 0.04704233631491661}, {"image_id": 13, "category_id": 3, "bbox": [1377.5113525390625, 249.35748291015625, 542.4886474609375, 830.6425170898438], "score": 0.04640166461467743}, {"image_id": 13, "category_id": 3, "bbox": [651.2105102539062, 109.55200958251953, 381.91107177734375, 591.4548263549805], "score": 0.04631352424621582}, {"image_id": 13, "category_id": 3, "bbox": [1001.5380249023438, 127.78234100341797, 854.4579467773438, 938.9619216918945], "score": 0.04360387101769447}, {"image_id": 13, "category_id": 3, "bbox": [178.27944946289062, 139.77359008789062, 969.9604187011719, 939.8384704589844], "score": 0.042786259204149246}, {"image_id": 13, "category_id": 3, "bbox": [121.89191436767578, 91.74436950683594, 485.27220916748047, 702.5570220947266], "score": 0.04192251339554787}, {"image_id": 13, "category_id": 3, "bbox": [1370.984619140625, 0.0, 444.767578125, 483.56195068359375], "score": 0.040454600006341934}, {"image_id": 13, "category_id": 3, "bbox": [1005.9633178710938, 0.0, 392.75653076171875, 417.71063232421875], "score": 0.03943023830652237}, {"image_id": 13, "category_id": 3, "bbox": [560.9658203125, 0.0, 929.433349609375, 739.1602172851562], "score": 0.03881663456559181}, {"image_id": 13, "category_id": 3, "bbox": [371.1028137207031, 0.0, 505.8035583496094, 616.2576904296875], "score": 0.036350179463624954}, {"image_id": 14, "category_id": 1, "bbox": [567.2349853515625, 507.35601806640625, 430.724609375, 555.8296508789062], "score": 0.7526895999908447}, {"image_id": 14, "category_id": 1, "bbox": [590.30126953125, 267.6692810058594, 280.43048095703125, 692.3636169433594], "score": 0.6221360564231873}, {"image_id": 14, "category_id": 1, "bbox": [605.8358154296875, 105.60836029052734, 246.06884765625, 620.2003555297852], "score": 0.43953657150268555}, {"image_id": 14, "category_id": 1, "bbox": [492.1051025390625, 227.9801025390625, 366.5050048828125, 608.7412109375], "score": 0.3361971974372864}, {"image_id": 14, "category_id": 1, "bbox": [445.098388671875, 541.566650390625, 474.5845947265625, 538.433349609375], "score": 0.2989497780799866}, {"image_id": 14, "category_id": 1, "bbox": [601.396728515625, 235.50003051757812, 366.38763427734375, 600.2229309082031], "score": 0.27120867371559143}, {"image_id": 14, "category_id": 1, "bbox": [1640.0386962890625, 0.0, 279.9613037109375, 616.9367065429688], "score": 0.26389074325561523}, {"image_id": 14, "category_id": 1, "bbox": [662.9815673828125, 522.5117797851562, 464.5909423828125, 557.4882202148438], "score": 0.15966013073921204}, {"image_id": 14, "category_id": 1, "bbox": [964.0552368164062, 508.35107421875, 296.98394775390625, 526.0711669921875], "score": 0.15910416841506958}, {"image_id": 14, "category_id": 1, "bbox": [434.9674072265625, 314.8119812011719, 450.33843994140625, 720.6828918457031], "score": 0.1045728549361229}, {"image_id": 14, "category_id": 1, "bbox": [601.0831909179688, 45.43880081176758, 375.8907470703125, 644.2913017272949], "score": 0.09438355267047882}, {"image_id": 14, "category_id": 1, "bbox": [208.09274291992188, 79.05366516113281, 402.2446594238281, 526.2733001708984], "score": 0.08329986035823822}, {"image_id": 14, "category_id": 1, "bbox": [543.2828979492188, 298.719482421875, 557.5109252929688, 744.8590087890625], "score": 0.07453954219818115}, {"image_id": 14, "category_id": 1, "bbox": [461.41510009765625, 17.290180206298828, 778.1782836914062, 997.6200370788574], "score": 0.07449167966842651}, {"image_id": 14, "category_id": 1, "bbox": [253.7256317138672, 19.2048397064209, 796.9589385986328, 987.3511905670166], "score": 0.0730411484837532}, {"image_id": 14, "category_id": 1, "bbox": [822.0243530273438, 21.193862915039062, 847.4710083007812, 990.4091033935547], "score": 0.0709768682718277}, {"image_id": 14, "category_id": 1, "bbox": [1202.78076171875, 16.97161102294922, 717.21923828125, 977.2130813598633], "score": 0.06690913438796997}, {"image_id": 14, "category_id": 1, "bbox": [614.737548828125, 127.54102325439453, 816.534912109375, 952.4589767456055], "score": 0.06335742771625519}, {"image_id": 14, "category_id": 1, "bbox": [0.0, 37.49079132080078, 692.2682495117188, 937.9087448120117], "score": 0.061122409999370575}, {"image_id": 14, "category_id": 1, "bbox": [915.9555053710938, 350.829345703125, 405.46893310546875, 642.7755737304688], "score": 0.0605589896440506}, {"image_id": 14, "category_id": 1, "bbox": [973.8382568359375, 608.9243774414062, 406.0906982421875, 471.07562255859375], "score": 0.05796436965465546}, {"image_id": 14, "category_id": 1, "bbox": [1525.612060546875, 0.0, 376.1258544921875, 492.5939025878906], "score": 0.05238942429423332}, {"image_id": 14, "category_id": 1, "bbox": [413.649169921875, 12.840768814086914, 453.1514892578125, 683.0646877288818], "score": 0.05153421312570572}, {"image_id": 14, "category_id": 1, "bbox": [839.6356811523438, 600.3726196289062, 421.07501220703125, 479.62738037109375], "score": 0.050987571477890015}, {"image_id": 14, "category_id": 1, "bbox": [2.492523193359375, 258.643310546875, 936.2057800292969, 821.356689453125], "score": 0.04724672809243202}, {"image_id": 14, "category_id": 1, "bbox": [291.099609375, 668.8362426757812, 493.7979736328125, 411.16375732421875], "score": 0.04526422917842865}, {"image_id": 14, "category_id": 1, "bbox": [20.304656982421875, 0.0, 895.2794494628906, 551.8087158203125], "score": 0.04254207760095596}, {"image_id": 14, "category_id": 1, "bbox": [349.6882629394531, 215.2140350341797, 427.0466613769531, 601.7118682861328], "score": 0.04094712808728218}, {"image_id": 14, "category_id": 1, "bbox": [1012.6121215820312, 132.66036987304688, 841.9822387695312, 947.3396301269531], "score": 0.04003438353538513}, {"image_id": 14, "category_id": 1, "bbox": [0.0, 249.84130859375, 561.037353515625, 830.15869140625], "score": 0.03949246555566788}, {"image_id": 14, "category_id": 1, "bbox": [384.57177734375, 353.7644348144531, 939.123291015625, 726.2355651855469], "score": 0.03920765221118927}, {"image_id": 14, "category_id": 1, "bbox": [1632.779296875, 0.0, 274.2659912109375, 373.1532897949219], "score": 0.038811247795820236}, {"image_id": 14, "category_id": 1, "bbox": [137.48428344726562, 464.8445129394531, 1047.3683776855469, 615.1554870605469], "score": 0.0384720116853714}, {"image_id": 14, "category_id": 1, "bbox": [1076.3553466796875, 523.0106811523438, 416.0303955078125, 556.9893188476562], "score": 0.038211073726415634}, {"image_id": 14, "category_id": 2, "bbox": [960.9970703125, 459.5618896484375, 296.2281494140625, 528.3929443359375], "score": 0.21660299599170685}, {"image_id": 14, "category_id": 2, "bbox": [960.7950439453125, 542.0023803710938, 395.251220703125, 537.9976196289062], "score": 0.15212364494800568}, {"image_id": 14, "category_id": 2, "bbox": [567.2349853515625, 507.35601806640625, 430.724609375, 555.8296508789062], "score": 0.10515573620796204}, {"image_id": 14, "category_id": 2, "bbox": [845.2739868164062, 384.4676208496094, 402.37957763671875, 574.6394348144531], "score": 0.10337568819522858}, {"image_id": 14, "category_id": 2, "bbox": [839.6356811523438, 600.3726196289062, 421.07501220703125, 479.62738037109375], "score": 0.08198925852775574}, {"image_id": 14, "category_id": 2, "bbox": [1080.6722412109375, 586.7930908203125, 408.9112548828125, 493.2069091796875], "score": 0.07446958869695663}, {"image_id": 14, "category_id": 2, "bbox": [287.3277282714844, 734.7307739257812, 500.1032409667969, 345.26922607421875], "score": 0.07171515375375748}, {"image_id": 14, "category_id": 2, "bbox": [445.098388671875, 541.566650390625, 474.5845947265625, 538.433349609375], "score": 0.05800225958228111}, {"image_id": 14, "category_id": 2, "bbox": [955.2617797851562, 344.5020751953125, 438.27447509765625, 654.572998046875], "score": 0.05690337345004082}, {"image_id": 14, "category_id": 2, "bbox": [900.77294921875, 749.476806640625, 424.463134765625, 330.523193359375], "score": 0.0548742301762104}, {"image_id": 14, "category_id": 2, "bbox": [100.0835952758789, 673.2844848632812, 517.0613632202148, 406.71551513671875], "score": 0.054640933871269226}, {"image_id": 14, "category_id": 2, "bbox": [686.8622436523438, 692.5205688476562, 416.43243408203125, 387.47943115234375], "score": 0.0499107800424099}, {"image_id": 14, "category_id": 2, "bbox": [590.30126953125, 267.6692810058594, 280.43048095703125, 692.3636169433594], "score": 0.04850592836737633}, {"image_id": 14, "category_id": 2, "bbox": [1249.2939453125, 594.5696411132812, 331.4652099609375, 485.43035888671875], "score": 0.04813047870993614}, {"image_id": 14, "category_id": 2, "bbox": [687.1847534179688, 356.4091796875, 434.76361083984375, 644.8225708007812], "score": 0.04804122820496559}, {"image_id": 14, "category_id": 2, "bbox": [936.4876098632812, 274.38409423828125, 346.63616943359375, 597.5467529296875], "score": 0.046264853328466415}, {"image_id": 14, "category_id": 2, "bbox": [351.6494445800781, 252.49285888671875, 429.8062438964844, 620.0924072265625], "score": 0.045265890657901764}, {"image_id": 14, "category_id": 2, "bbox": [1126.8775634765625, 376.07647705078125, 351.744873046875, 597.73779296875], "score": 0.043686844408512115}, {"image_id": 14, "category_id": 2, "bbox": [1269.508544921875, 655.9939575195312, 485.4710693359375, 424.00604248046875], "score": 0.042864374816417694}, {"image_id": 14, "category_id": 2, "bbox": [608.7178955078125, 177.033935546875, 230.68719482421875, 589.6218872070312], "score": 0.04267600551247597}, {"image_id": 14, "category_id": 2, "bbox": [237.9971160888672, 599.7903442382812, 437.81922912597656, 480.20965576171875], "score": 0.042275045067071915}, {"image_id": 14, "category_id": 2, "bbox": [5.378929138183594, 716.0562744140625, 489.92969512939453, 363.9437255859375], "score": 0.04160819575190544}, {"image_id": 14, "category_id": 2, "bbox": [1314.7264404296875, 83.51842498779297, 326.9503173828125, 635.6460037231445], "score": 0.04131321609020233}, {"image_id": 14, "category_id": 2, "bbox": [1426.92333984375, 0.0, 402.13818359375, 483.20111083984375], "score": 0.041259489953517914}, {"image_id": 14, "category_id": 2, "bbox": [465.905029296875, 778.0617065429688, 549.37255859375, 301.93829345703125], "score": 0.04022002965211868}, {"image_id": 14, "category_id": 2, "bbox": [203.6924285888672, 41.09975051879883, 399.9777374267578, 527.4033622741699], "score": 0.03917103260755539}, {"image_id": 14, "category_id": 2, "bbox": [1237.5703125, 133.84193420410156, 354.728759765625, 647.5585174560547], "score": 0.03888211399316788}, {"image_id": 14, "category_id": 2, "bbox": [1398.0546875, 97.52527618408203, 357.835205078125, 606.1565475463867], "score": 0.0385441929101944}, {"image_id": 14, "category_id": 2, "bbox": [434.9674072265625, 314.8119812011719, 450.33843994140625, 720.6828918457031], "score": 0.03813561797142029}, {"image_id": 14, "category_id": 3, "bbox": [960.9970703125, 459.5618896484375, 296.2281494140625, 528.3929443359375], "score": 0.5388631224632263}, {"image_id": 14, "category_id": 3, "bbox": [561.0100708007812, 533.8311767578125, 490.40179443359375, 546.1688232421875], "score": 0.34010806679725647}, {"image_id": 14, "category_id": 3, "bbox": [960.7950439453125, 542.0023803710938, 395.251220703125, 537.9976196289062], "score": 0.3287065923213959}, {"image_id": 14, "category_id": 3, "bbox": [845.2739868164062, 384.4676208496094, 402.37957763671875, 574.6394348144531], "score": 0.17846789956092834}, {"image_id": 14, "category_id": 3, "bbox": [834.0624389648438, 538.7200317382812, 429.97491455078125, 541.2799682617188], "score": 0.17276792228221893}, {"image_id": 14, "category_id": 3, "bbox": [955.2617797851562, 344.5020751953125, 438.27447509765625, 654.572998046875], "score": 0.16832970082759857}, {"image_id": 14, "category_id": 3, "bbox": [602.9393920898438, 255.1500701904297, 255.1710205078125, 635.6842803955078], "score": 0.1525108367204666}, {"image_id": 14, "category_id": 3, "bbox": [605.8358154296875, 105.60836029052734, 246.06884765625, 620.2003555297852], "score": 0.13716329634189606}, {"image_id": 14, "category_id": 3, "bbox": [181.08714294433594, 80.27454376220703, 407.9887237548828, 528.368278503418], "score": 0.10675925761461258}, {"image_id": 14, "category_id": 3, "bbox": [543.2828979492188, 298.719482421875, 557.5109252929688, 744.8590087890625], "score": 0.09941019862890244}, {"image_id": 14, "category_id": 3, "bbox": [594.1728515625, 270.671875, 383.41650390625, 692.5737915039062], "score": 0.09923187643289566}, {"image_id": 14, "category_id": 3, "bbox": [1076.3553466796875, 523.0106811523438, 416.0303955078125, 556.9893188476562], "score": 0.09775498509407043}, {"image_id": 14, "category_id": 3, "bbox": [473.5721740722656, 269.05010986328125, 387.4275207519531, 689.786376953125], "score": 0.08130435645580292}, {"image_id": 14, "category_id": 3, "bbox": [445.098388671875, 541.566650390625, 474.5845947265625, 538.433349609375], "score": 0.07966332882642746}, {"image_id": 14, "category_id": 3, "bbox": [564.1710815429688, 350.7880859375, 904.2788696289062, 729.2119140625], "score": 0.07961498945951462}, {"image_id": 14, "category_id": 3, "bbox": [232.97412109375, 731.0089721679688, 467.35980224609375, 348.99102783203125], "score": 0.07857915759086609}, {"image_id": 14, "category_id": 3, "bbox": [361.6155090332031, 735.734619140625, 519.7944030761719, 344.265380859375], "score": 0.0658208504319191}, {"image_id": 14, "category_id": 3, "bbox": [0.0, 350.018798828125, 978.42919921875, 729.981201171875], "score": 0.06571479886770248}, {"image_id": 14, "category_id": 3, "bbox": [465.905029296875, 778.0617065429688, 549.37255859375, 301.93829345703125], "score": 0.06359872221946716}, {"image_id": 14, "category_id": 3, "bbox": [453.34039306640625, 127.3935317993164, 773.4208374023438, 952.6064682006836], "score": 0.06020958349108696}, {"image_id": 14, "category_id": 3, "bbox": [900.77294921875, 749.476806640625, 424.463134765625, 330.523193359375], "score": 0.05750727280974388}, {"image_id": 14, "category_id": 3, "bbox": [1516.5997314453125, 0.0, 403.4002685546875, 228.4379425048828], "score": 0.056951526552438736}, {"image_id": 14, "category_id": 3, "bbox": [686.8622436523438, 692.5205688476562, 416.43243408203125, 387.47943115234375], "score": 0.056113481521606445}, {"image_id": 14, "category_id": 3, "bbox": [297.166259765625, 553.3778076171875, 481.62823486328125, 526.6221923828125], "score": 0.05128338560461998}, {"image_id": 14, "category_id": 3, "bbox": [1640.0386962890625, 0.0, 279.9613037109375, 616.9367065429688], "score": 0.04986603930592537}, {"image_id": 14, "category_id": 3, "bbox": [1745.5712890625, 0.0, 174.4287109375, 192.38876342773438], "score": 0.04901042580604553}, {"image_id": 14, "category_id": 3, "bbox": [1599.0345458984375, 0.0, 320.9654541015625, 171.02035522460938], "score": 0.04875486344099045}, {"image_id": 14, "category_id": 3, "bbox": [1673.6326904296875, 40.22676086425781, 246.3673095703125, 236.8329315185547], "score": 0.04815680906176567}, {"image_id": 14, "category_id": 3, "bbox": [413.649169921875, 12.840768814086914, 453.1514892578125, 683.0646877288818], "score": 0.04797089472413063}, {"image_id": 14, "category_id": 3, "bbox": [754.4046020507812, 456.9328918457031, 908.8635864257812, 623.0671081542969], "score": 0.044650591909885406}, {"image_id": 14, "category_id": 3, "bbox": [307.0609436035156, 122.27685546875, 453.7564392089844, 654.30517578125], "score": 0.04232749715447426}, {"image_id": 14, "category_id": 3, "bbox": [1246.54345703125, 477.42132568359375, 335.5970458984375, 602.5786743164062], "score": 0.04227519780397415}, {"image_id": 14, "category_id": 3, "bbox": [936.4876098632812, 274.38409423828125, 346.63616943359375, 597.5467529296875], "score": 0.041940104216337204}, {"image_id": 14, "category_id": 3, "bbox": [638.6244506835938, 469.55633544921875, 559.0007934570312, 610.4436645507812], "score": 0.04112297669053078}, {"image_id": 14, "category_id": 3, "bbox": [601.0831909179688, 45.43880081176758, 375.8907470703125, 644.2913017272949], "score": 0.04040771350264549}, {"image_id": 14, "category_id": 3, "bbox": [635.4819946289062, 20.757568359375, 817.6876831054688, 985.9949951171875], "score": 0.040165361016988754}, {"image_id": 14, "category_id": 3, "bbox": [1202.78076171875, 16.97161102294922, 717.21923828125, 977.2130813598633], "score": 0.03838537633419037}, {"image_id": 15, "category_id": 1, "bbox": [284.91510009765625, 450.2974548339844, 301.21368408203125, 269.7025451660156], "score": 0.2997525632381439}, {"image_id": 15, "category_id": 1, "bbox": [555.9175415039062, 328.0311279296875, 252.61529541015625, 379.929931640625], "score": 0.20892192423343658}, {"image_id": 15, "category_id": 1, "bbox": [386.26690673828125, 447.0202331542969, 325.08245849609375, 272.9797668457031], "score": 0.13350306451320648}, {"image_id": 15, "category_id": 1, "bbox": [119.32919311523438, 8.192298889160156, 619.7023620605469, 659.0658798217773], "score": 0.11995235830545425}, {"image_id": 15, "category_id": 1, "bbox": [306.305908203125, 289.4626770019531, 340.86572265625, 430.5373229980469], "score": 0.10713429749011993}, {"image_id": 15, "category_id": 1, "bbox": [897.9622802734375, 216.91004943847656, 259.6658935546875, 431.10447692871094], "score": 0.09514526277780533}, {"image_id": 15, "category_id": 1, "bbox": [1067.0791015625, 0.0, 212.9208984375, 406.5905456542969], "score": 0.08614256978034973}, {"image_id": 15, "category_id": 1, "bbox": [460.75286865234375, 409.401611328125, 321.98492431640625, 310.598388671875], "score": 0.08206141740083694}, {"image_id": 15, "category_id": 1, "bbox": [265.2857666015625, 80.8100357055664, 580.7918701171875, 639.1899642944336], "score": 0.07750962674617767}, {"image_id": 15, "category_id": 1, "bbox": [570.964599609375, 222.68138122558594, 232.615478515625, 365.35633850097656], "score": 0.07376915216445923}, {"image_id": 15, "category_id": 1, "bbox": [381.3116455078125, 5.584384918212891, 612.2816162109375, 661.5071067810059], "score": 0.07220371812582016}, {"image_id": 15, "category_id": 1, "bbox": [916.8201904296875, 158.23199462890625, 232.10302734375, 362.18060302734375], "score": 0.0666867345571518}, {"image_id": 15, "category_id": 1, "bbox": [583.2916259765625, 400.37310791015625, 284.61285400390625, 319.62689208984375], "score": 0.06342270225286484}, {"image_id": 15, "category_id": 1, "bbox": [0.0, 15.127452850341797, 494.71258544921875, 638.1953010559082], "score": 0.061983272433280945}, {"image_id": 15, "category_id": 1, "bbox": [550.5108032226562, 0.0, 562.9132690429688, 584.5116577148438], "score": 0.05714055150747299}, {"image_id": 15, "category_id": 1, "bbox": [799.176025390625, 165.86282348632812, 480.823974609375, 554.1371765136719], "score": 0.05553276464343071}, {"image_id": 15, "category_id": 1, "bbox": [506.4452209472656, 160.68675231933594, 620.6016540527344, 559.3132476806641], "score": 0.04692748934030533}, {"image_id": 15, "category_id": 1, "bbox": [238.74917602539062, 317.44024658203125, 636.9797058105469, 402.55975341796875], "score": 0.046654924750328064}, {"image_id": 15, "category_id": 1, "bbox": [469.9813232421875, 233.91444396972656, 313.74835205078125, 429.4609832763672], "score": 0.04583228379487991}, {"image_id": 15, "category_id": 1, "bbox": [717.3346557617188, 344.4520568847656, 289.74493408203125, 375.5479431152344], "score": 0.04422125592827797}, {"image_id": 15, "category_id": 1, "bbox": [0.0, 243.67950439453125, 659.1595458984375, 476.32049560546875], "score": 0.04374482482671738}, {"image_id": 15, "category_id": 1, "bbox": [534.0018310546875, 0.0, 617.26904296875, 360.382080078125], "score": 0.04262200742959976}, {"image_id": 15, "category_id": 1, "bbox": [929.6754150390625, 21.531417846679688, 232.0682373046875, 436.56944274902344], "score": 0.04210285842418671}, {"image_id": 15, "category_id": 1, "bbox": [8.2193603515625, 0.0, 636.0980224609375, 357.1515197753906], "score": 0.0388530008494854}, {"image_id": 15, "category_id": 1, "bbox": [643.494384765625, 326.48583984375, 288.2838134765625, 393.51416015625], "score": 0.038426004350185394}, {"image_id": 15, "category_id": 1, "bbox": [246.31805419921875, 0.0, 672.9280395507812, 359.90020751953125], "score": 0.03661080822348595}, {"image_id": 15, "category_id": 1, "bbox": [807.4751586914062, 183.95301818847656, 297.55621337890625, 442.38731384277344], "score": 0.03621523082256317}, {"image_id": 15, "category_id": 2, "bbox": [558.860595703125, 296.6305236816406, 245.22216796875, 387.9569396972656], "score": 0.18008014559745789}, {"image_id": 15, "category_id": 2, "bbox": [592.989990234375, 231.11990356445312, 235.485595703125, 350.7550354003906], "score": 0.11358044296503067}, {"image_id": 15, "category_id": 2, "bbox": [486.3038635253906, 222.37753295898438, 292.3921203613281, 365.4058532714844], "score": 0.09818989783525467}, {"image_id": 15, "category_id": 2, "bbox": [382.8323059082031, 318.0440979003906, 344.9108581542969, 399.6003112792969], "score": 0.07726331055164337}, {"image_id": 15, "category_id": 2, "bbox": [648.5999755859375, 265.9183654785156, 272.69512939453125, 363.5362854003906], "score": 0.07118812948465347}, {"image_id": 15, "category_id": 2, "bbox": [309.4896240234375, 445.9449462890625, 335.2369384765625, 274.0550537109375], "score": 0.06269417703151703}, {"image_id": 15, "category_id": 2, "bbox": [911.2342529296875, 172.74163818359375, 238.197021484375, 380.03057861328125], "score": 0.05578232929110527}, {"image_id": 15, "category_id": 2, "bbox": [400.4683532714844, 219.5218048095703, 317.6741027832031, 382.5673065185547], "score": 0.05446565896272659}, {"image_id": 15, "category_id": 2, "bbox": [218.48794555664062, 0.0, 281.9642639160156, 272.89208984375], "score": 0.05397253856062889}, {"image_id": 15, "category_id": 2, "bbox": [318.94927978515625, 281.7162780761719, 311.876953125, 421.7878723144531], "score": 0.05256935581564903}, {"image_id": 15, "category_id": 2, "bbox": [261.9578857421875, 447.0304260253906, 265.19805908203125, 272.9695739746094], "score": 0.050740357488393784}, {"image_id": 15, "category_id": 2, "bbox": [236.7536163330078, 52.073150634765625, 219.83119201660156, 367.0986022949219], "score": 0.05021743103861809}, {"image_id": 15, "category_id": 2, "bbox": [853.0221557617188, 441.9833984375, 313.57476806640625, 278.0166015625], "score": 0.04955644905567169}, {"image_id": 15, "category_id": 2, "bbox": [653.7049560546875, 423.1524963378906, 312.6859130859375, 296.8475036621094], "score": 0.047936584800481796}, {"image_id": 15, "category_id": 2, "bbox": [192.6336669921875, 17.61650848388672, 218.5576171875, 359.18738555908203], "score": 0.04739083722233772}, {"image_id": 15, "category_id": 2, "bbox": [716.12353515625, 289.7076110839844, 290.47662353515625, 391.2383117675781], "score": 0.046481166034936905}, {"image_id": 15, "category_id": 2, "bbox": [583.2916259765625, 400.37310791015625, 284.61285400390625, 319.62689208984375], "score": 0.04603557661175728}, {"image_id": 15, "category_id": 2, "bbox": [476.6762390136719, 359.8182373046875, 306.7839660644531, 360.1817626953125], "score": 0.045040156692266464}, {"image_id": 15, "category_id": 2, "bbox": [95.72367858886719, 125.01004791259766, 275.9160614013672, 362.58629608154297], "score": 0.04476787894964218}, {"image_id": 15, "category_id": 2, "bbox": [263.78436279296875, 0.0, 308.42047119140625, 312.9623718261719], "score": 0.04180523380637169}, {"image_id": 15, "category_id": 2, "bbox": [826.1325073242188, 109.44786071777344, 258.76776123046875, 390.80470275878906], "score": 0.04159163311123848}, {"image_id": 15, "category_id": 2, "bbox": [737.7461547851562, 170.91270446777344, 257.1009521484375, 400.6968536376953], "score": 0.04107937961816788}, {"image_id": 15, "category_id": 2, "bbox": [537.5369873046875, 148.45718383789062, 281.99810791015625, 398.4436340332031], "score": 0.04044336453080177}, {"image_id": 15, "category_id": 2, "bbox": [894.1014404296875, 63.635677337646484, 228.24365234375, 427.52194595336914], "score": 0.0401245579123497}, {"image_id": 15, "category_id": 2, "bbox": [106.47913360595703, 8.222768783569336, 253.17900848388672, 379.88401222229004], "score": 0.039341866970062256}, {"image_id": 15, "category_id": 2, "bbox": [218.0297088623047, 315.354248046875, 305.1634063720703, 404.645751953125], "score": 0.03925473988056183}, {"image_id": 15, "category_id": 2, "bbox": [150.67440795898438, 161.9541015625, 289.331298828125, 365.28924560546875], "score": 0.03906649351119995}, {"image_id": 15, "category_id": 2, "bbox": [1015.0320434570312, 83.0521240234375, 208.35797119140625, 395.8305969238281], "score": 0.03903981298208237}, {"image_id": 15, "category_id": 2, "bbox": [888.8126220703125, 250.46983337402344, 268.2392578125, 424.55726623535156], "score": 0.0379801020026207}, {"image_id": 15, "category_id": 2, "bbox": [144.16204833984375, 0.0, 307.8683776855469, 235.1691436767578], "score": 0.03780936077237129}, {"image_id": 15, "category_id": 2, "bbox": [76.85116577148438, 445.4505615234375, 312.6731262207031, 274.5494384765625], "score": 0.036969225853681564}, {"image_id": 15, "category_id": 2, "bbox": [21.388458251953125, 124.09181213378906, 282.1029357910156, 370.08824157714844], "score": 0.03691590204834938}, {"image_id": 15, "category_id": 2, "bbox": [809.4547729492188, 222.21839904785156, 292.79351806640625, 429.74778747558594], "score": 0.0366983637213707}, {"image_id": 15, "category_id": 2, "bbox": [1062.2579345703125, 15.140653610229492, 217.7420654296875, 433.8019428253174], "score": 0.036641690880060196}, {"image_id": 15, "category_id": 2, "bbox": [797.1039428710938, 395.4020690917969, 272.64801025390625, 324.5979309082031], "score": 0.03639151155948639}, {"image_id": 15, "category_id": 2, "bbox": [272.7364807128906, 47.5428352355957, 263.5800476074219, 377.7568473815918], "score": 0.03638328239321709}, {"image_id": 15, "category_id": 2, "bbox": [328.54547119140625, 0.0, 325.69390869140625, 272.14227294921875], "score": 0.03636637702584267}, {"image_id": 15, "category_id": 2, "bbox": [71.16325378417969, 0.0, 321.3257598876953, 276.5601806640625], "score": 0.036110471934080124}, {"image_id": 15, "category_id": 2, "bbox": [665.8670043945312, 132.3923797607422, 273.0784912109375, 412.49916076660156], "score": 0.035840559750795364}, {"image_id": 15, "category_id": 3, "bbox": [555.9175415039062, 328.0311279296875, 252.61529541015625, 379.929931640625], "score": 0.5262241363525391}, {"image_id": 15, "category_id": 3, "bbox": [570.964599609375, 222.68138122558594, 232.615478515625, 365.35633850097656], "score": 0.33346280455589294}, {"image_id": 15, "category_id": 3, "bbox": [248.63467407226562, 239.8922119140625, 619.3224792480469, 480.1077880859375], "score": 0.1497812420129776}, {"image_id": 15, "category_id": 3, "bbox": [469.9813232421875, 233.91444396972656, 313.74835205078125, 429.4609832763672], "score": 0.14499850571155548}, {"image_id": 15, "category_id": 3, "bbox": [919.9066772460938, 118.68367004394531, 223.05987548828125, 382.4033660888672], "score": 0.1257125586271286}, {"image_id": 15, "category_id": 3, "bbox": [382.8323059082031, 318.0440979003906, 344.9108581542969, 399.6003112792969], "score": 0.12555484473705292}, {"image_id": 15, "category_id": 3, "bbox": [460.75286865234375, 409.401611328125, 321.98492431640625, 310.598388671875], "score": 0.10890904068946838}, {"image_id": 15, "category_id": 3, "bbox": [648.97509765625, 344.55706787109375, 289.83587646484375, 375.44293212890625], "score": 0.10702326148748398}, {"image_id": 15, "category_id": 3, "bbox": [506.4452209472656, 160.68675231933594, 620.6016540527344, 559.3132476806641], "score": 0.1064506247639656}, {"image_id": 15, "category_id": 3, "bbox": [583.2916259765625, 400.37310791015625, 284.61285400390625, 319.62689208984375], "score": 0.10637763887643814}, {"image_id": 15, "category_id": 3, "bbox": [284.86669921875, 366.37158203125, 298.44378662109375, 353.62841796875], "score": 0.10520832985639572}, {"image_id": 15, "category_id": 3, "bbox": [897.9622802734375, 216.91004943847656, 259.6658935546875, 431.10447692871094], "score": 0.1001012772321701}, {"image_id": 15, "category_id": 3, "bbox": [376.6915283203125, 84.57168579101562, 590.23974609375, 635.4283142089844], "score": 0.09627997875213623}, {"image_id": 15, "category_id": 3, "bbox": [648.5999755859375, 265.9183654785156, 272.69512939453125, 363.5362854003906], "score": 0.09408961236476898}, {"image_id": 15, "category_id": 3, "bbox": [799.176025390625, 165.86282348632812, 480.823974609375, 554.1371765136719], "score": 0.07922138273715973}, {"image_id": 15, "category_id": 3, "bbox": [140.79421997070312, 85.06013488769531, 599.9707336425781, 634.9398651123047], "score": 0.06963465362787247}, {"image_id": 15, "category_id": 3, "bbox": [715.276611328125, 317.3141784667969, 289.43798828125, 402.6858215332031], "score": 0.06945259869098663}, {"image_id": 15, "category_id": 3, "bbox": [242.28701782226562, 5.334720611572266, 626.4225769042969, 662.9099082946777], "score": 0.06461895257234573}, {"image_id": 15, "category_id": 3, "bbox": [603.8661499023438, 309.64794921875, 676.1338500976562, 410.35205078125], "score": 0.059298790991306305}, {"image_id": 15, "category_id": 3, "bbox": [550.5108032226562, 0.0, 562.9132690429688, 584.5116577148438], "score": 0.053432121872901917}, {"image_id": 15, "category_id": 3, "bbox": [0.0, 243.67950439453125, 659.1595458984375, 476.32049560546875], "score": 0.05238478630781174}, {"image_id": 15, "category_id": 3, "bbox": [832.3271484375, 150.47891235351562, 262.1251220703125, 370.2017517089844], "score": 0.050997294485569}, {"image_id": 15, "category_id": 3, "bbox": [809.4547729492188, 222.21839904785156, 292.79351806640625, 429.74778747558594], "score": 0.050015583634376526}, {"image_id": 15, "category_id": 3, "bbox": [1015.0320434570312, 83.0521240234375, 208.35797119140625, 395.8305969238281], "score": 0.04959661886096001}, {"image_id": 15, "category_id": 3, "bbox": [883.2669677734375, 15.153596878051758, 237.442626953125, 448.83813285827637], "score": 0.04927991330623627}, {"image_id": 15, "category_id": 3, "bbox": [305.38671875, 482.4426574707031, 343.713623046875, 237.55734252929688], "score": 0.04926776885986328}, {"image_id": 15, "category_id": 3, "bbox": [1067.0791015625, 0.0, 212.9208984375, 406.5905456542969], "score": 0.04636219143867493}, {"image_id": 15, "category_id": 3, "bbox": [358.5914001464844, 385.6519470214844, 670.0138244628906, 334.3480529785156], "score": 0.04156940057873726}, {"image_id": 15, "category_id": 3, "bbox": [0.0, 9.65753173828125, 608.2761840820312, 651.2119140625], "score": 0.03883548080921173}, {"image_id": 15, "category_id": 3, "bbox": [875.565673828125, 342.5406494140625, 257.29736328125, 377.4593505859375], "score": 0.03813798725605011}, {"image_id": 15, "category_id": 3, "bbox": [625.810546875, 0.0, 618.1572265625, 424.5748596191406], "score": 0.0377938374876976}, {"image_id": 15, "category_id": 3, "bbox": [892.9049682617188, 24.841358184814453, 387.09503173828125, 624.0291862487793], "score": 0.03615352511405945}, {"image_id": 15, "category_id": 3, "bbox": [236.7536163330078, 52.073150634765625, 219.83119201660156, 367.0986022949219], "score": 0.0358269102871418}, {"image_id": 15, "category_id": 3, "bbox": [806.228271484375, 346.5971984863281, 257.928466796875, 373.4028015136719], "score": 0.03558814898133278}, {"image_id": 16, "category_id": 1, "bbox": [362.6244812011719, 299.6726989746094, 205.22348022460938, 371.9792175292969], "score": 0.588880717754364}, {"image_id": 16, "category_id": 1, "bbox": [342.2330322265625, 224.3411102294922, 196.91412353515625, 388.96644592285156], "score": 0.455890029668808}, {"image_id": 16, "category_id": 1, "bbox": [324.1534423828125, 280.1866149902344, 313.3809814453125, 439.8133850097656], "score": 0.21698173880577087}, {"image_id": 16, "category_id": 1, "bbox": [381.912353515625, 216.9392547607422, 271.4776611328125, 400.02284240722656], "score": 0.14732572436332703}, {"image_id": 16, "category_id": 1, "bbox": [121.05645751953125, 6.633819580078125, 613.6747436523438, 669.7938537597656], "score": 0.10174722969532013}, {"image_id": 16, "category_id": 1, "bbox": [533.361572265625, 19.518310546875, 584.92333984375, 651.4182739257812], "score": 0.09975647926330566}, {"image_id": 16, "category_id": 1, "bbox": [365.91461181640625, 333.45257568359375, 369.83966064453125, 358.20172119140625], "score": 0.08535637706518173}, {"image_id": 16, "category_id": 1, "bbox": [263.83026123046875, 289.4198303222656, 254.74212646484375, 385.9544982910156], "score": 0.077920101583004}, {"image_id": 16, "category_id": 1, "bbox": [908.3897705078125, 170.68675231933594, 210.3905029296875, 387.3239288330078], "score": 0.07092487812042236}, {"image_id": 16, "category_id": 1, "bbox": [0.0, 15.668083190917969, 487.0750732421875, 639.7146072387695], "score": 0.06925756484270096}, {"image_id": 16, "category_id": 1, "bbox": [334.9927978515625, 119.18450927734375, 283.7127685546875, 458.4326171875], "score": 0.06871666014194489}, {"image_id": 16, "category_id": 1, "bbox": [385.1438293457031, 0.0, 650.6490173339844, 586.452392578125], "score": 0.06418672204017639}, {"image_id": 16, "category_id": 1, "bbox": [164.21063232421875, 15.601238250732422, 288.81182861328125, 367.2226753234863], "score": 0.0548994205892086}, {"image_id": 16, "category_id": 1, "bbox": [258.0205078125, 80.78785705566406, 608.81005859375, 639.2121429443359], "score": 0.05021756887435913}, {"image_id": 16, "category_id": 1, "bbox": [926.6473388671875, 235.84046936035156, 232.6497802734375, 417.4474334716797], "score": 0.04941415414214134}, {"image_id": 16, "category_id": 1, "bbox": [1058.0938720703125, 0.0, 221.9061279296875, 375.61273193359375], "score": 0.0490218810737133}, {"image_id": 16, "category_id": 1, "bbox": [665.4188232421875, 105.33186340332031, 547.4813232421875, 614.6681365966797], "score": 0.04763634875416756}, {"image_id": 16, "category_id": 1, "bbox": [442.40130615234375, 415.6287536621094, 312.31805419921875, 304.3712463378906], "score": 0.04652315378189087}, {"image_id": 16, "category_id": 1, "bbox": [928.5352783203125, 23.192533493041992, 220.8953857421875, 437.0414447784424], "score": 0.045879822224378586}, {"image_id": 16, "category_id": 1, "bbox": [464.6447448730469, 219.59957885742188, 267.9160461425781, 391.7303771972656], "score": 0.04544048383831978}, {"image_id": 16, "category_id": 1, "bbox": [380.5443420410156, 162.265380859375, 633.5134582519531, 557.734619140625], "score": 0.04437420889735222}, {"image_id": 16, "category_id": 1, "bbox": [630.3564453125, 0.0, 580.2840576171875, 509.85748291015625], "score": 0.04364008828997612}, {"image_id": 16, "category_id": 1, "bbox": [897.9171752929688, 169.39280700683594, 382.08282470703125, 550.6071929931641], "score": 0.04103400558233261}, {"image_id": 16, "category_id": 1, "bbox": [532.2042236328125, 0.0, 615.1090087890625, 362.6470947265625], "score": 0.04097892716526985}, {"image_id": 16, "category_id": 1, "bbox": [0.0, 176.7051239013672, 622.8216552734375, 543.2948760986328], "score": 0.040484994649887085}, {"image_id": 16, "category_id": 1, "bbox": [384.74835205078125, 540.9713134765625, 229.1282958984375, 162.43731689453125], "score": 0.03996006399393082}, {"image_id": 16, "category_id": 1, "bbox": [134.569091796875, 0.0, 627.1005859375, 375.66448974609375], "score": 0.03877488523721695}, {"image_id": 16, "category_id": 1, "bbox": [356.1250305175781, 426.20916748046875, 314.5539245605469, 293.79083251953125], "score": 0.036838095635175705}, {"image_id": 16, "category_id": 1, "bbox": [790.0411376953125, 0.0, 489.9588623046875, 576.9779663085938], "score": 0.03545915335416794}, {"image_id": 16, "category_id": 2, "bbox": [374.6915588378906, 376.95355224609375, 362.2582702636719, 333.26007080078125], "score": 0.40823251008987427}, {"image_id": 16, "category_id": 2, "bbox": [510.4267578125, 416.3811340332031, 291.88934326171875, 303.6188659667969], "score": 0.1628989428281784}, {"image_id": 16, "category_id": 2, "bbox": [464.1882629394531, 252.70892333984375, 271.3666076660156, 393.460205078125], "score": 0.0721372663974762}, {"image_id": 16, "category_id": 2, "bbox": [362.6244812011719, 299.6726989746094, 205.22348022460938, 371.9792175292969], "score": 0.06583435088396072}, {"image_id": 16, "category_id": 2, "bbox": [900.43505859375, 193.4424285888672, 222.790283203125, 409.82411193847656], "score": 0.06368616223335266}, {"image_id": 16, "category_id": 2, "bbox": [342.2330322265625, 224.3411102294922, 196.91412353515625, 388.96644592285156], "score": 0.06261888891458511}, {"image_id": 16, "category_id": 2, "bbox": [201.50938415527344, 433.6234436035156, 334.57154846191406, 286.3765563964844], "score": 0.059675030410289764}, {"image_id": 16, "category_id": 2, "bbox": [741.4872436523438, 320.49310302734375, 266.03656005859375, 389.92034912109375], "score": 0.05816587433218956}, {"image_id": 16, "category_id": 2, "bbox": [904.086181640625, 116.78359985351562, 208.246826171875, 384.7676696777344], "score": 0.05695266276597977}, {"image_id": 16, "category_id": 2, "bbox": [590.281982421875, 288.16461181640625, 281.69720458984375, 392.32293701171875], "score": 0.05417177081108093}, {"image_id": 16, "category_id": 2, "bbox": [800.674560546875, 278.0810241699219, 266.830322265625, 398.8716735839844], "score": 0.053575389087200165}, {"image_id": 16, "category_id": 2, "bbox": [654.4322509765625, 356.0010070800781, 307.76568603515625, 363.9989929199219], "score": 0.052051231265068054}, {"image_id": 16, "category_id": 2, "bbox": [381.912353515625, 216.9392547607422, 271.4776611328125, 400.02284240722656], "score": 0.049821220338344574}, {"image_id": 16, "category_id": 2, "bbox": [747.88720703125, 208.172119140625, 257.34228515625, 401.63580322265625], "score": 0.04964092746376991}, {"image_id": 16, "category_id": 2, "bbox": [1008.7347412109375, 212.14476013183594, 236.3936767578125, 372.65370178222656], "score": 0.04958510771393776}, {"image_id": 16, "category_id": 2, "bbox": [927.4014282226562, 272.383544921875, 243.11065673828125, 408.31378173828125], "score": 0.049196965992450714}, {"image_id": 16, "category_id": 2, "bbox": [535.3876953125, 212.96585083007812, 253.74530029296875, 402.7171936035156], "score": 0.04918587580323219}, {"image_id": 16, "category_id": 2, "bbox": [826.5387573242188, 165.60958862304688, 254.59857177734375, 398.0992736816406], "score": 0.04882694408297539}, {"image_id": 16, "category_id": 2, "bbox": [256.838623046875, 208.4617462158203, 259.85198974609375, 408.2347869873047], "score": 0.04751313105225563}, {"image_id": 16, "category_id": 2, "bbox": [673.7640991210938, 248.00155639648438, 272.78790283203125, 395.0783996582031], "score": 0.04608523100614548}, {"image_id": 16, "category_id": 2, "bbox": [1070.961669921875, 177.84765625, 209.038330078125, 386.8983154296875], "score": 0.04577619209885597}, {"image_id": 16, "category_id": 2, "bbox": [1025.5101318359375, 133.16525268554688, 201.0634765625, 352.4986877441406], "score": 0.04225904121994972}, {"image_id": 16, "category_id": 2, "bbox": [1073.748291015625, 81.47222137451172, 206.251708984375, 388.50479888916016], "score": 0.041746124625205994}, {"image_id": 16, "category_id": 2, "bbox": [154.67506408691406, 312.0383605957031, 303.82456970214844, 407.9616394042969], "score": 0.04020458832383156}, {"image_id": 16, "category_id": 2, "bbox": [130.55560302734375, 84.07486724853516, 600.1583862304688, 635.9251327514648], "score": 0.039116017520427704}, {"image_id": 16, "category_id": 2, "bbox": [582.5391845703125, 460.6905212402344, 332.5740966796875, 259.3094787597656], "score": 0.03909450024366379}, {"image_id": 16, "category_id": 2, "bbox": [881.3582763671875, 345.85284423828125, 247.264892578125, 374.14715576171875], "score": 0.03856780380010605}, {"image_id": 16, "category_id": 2, "bbox": [190.73626708984375, 0.0, 330.726806640625, 228.35089111328125], "score": 0.03853188827633858}, {"image_id": 16, "category_id": 2, "bbox": [162.0272674560547, 215.4930877685547, 290.28941345214844, 405.2661895751953], "score": 0.03824985772371292}, {"image_id": 16, "category_id": 2, "bbox": [324.1534423828125, 280.1866149902344, 313.3809814453125, 439.8133850097656], "score": 0.03821541368961334}, {"image_id": 16, "category_id": 2, "bbox": [928.5352783203125, 23.192533493041992, 220.8953857421875, 437.0414447784424], "score": 0.03739107400178909}, {"image_id": 16, "category_id": 2, "bbox": [780.98046875, 430.8081970214844, 313.95166015625, 289.1918029785156], "score": 0.03504335507750511}, {"image_id": 16, "category_id": 2, "bbox": [160.67730712890625, 0.0, 289.50445556640625, 360.2318420410156], "score": 0.03491990640759468}, {"image_id": 16, "category_id": 2, "bbox": [134.569091796875, 0.0, 627.1005859375, 375.66448974609375], "score": 0.03480954095721245}, {"image_id": 16, "category_id": 2, "bbox": [328.9412536621094, 0.0, 316.5586242675781, 271.8346252441406], "score": 0.03469555824995041}, {"image_id": 16, "category_id": 3, "bbox": [374.6915588378906, 376.95355224609375, 362.2582702636719, 333.26007080078125], "score": 0.5387378334999084}, {"image_id": 16, "category_id": 3, "bbox": [510.3959655761719, 376.9273986816406, 295.3989562988281, 339.6742248535156], "score": 0.2397172451019287}, {"image_id": 16, "category_id": 3, "bbox": [362.6244812011719, 299.6726989746094, 205.22348022460938, 371.9792175292969], "score": 0.2000344842672348}, {"image_id": 16, "category_id": 3, "bbox": [349.681396484375, 175.08908081054688, 227.1343994140625, 435.5223693847656], "score": 0.1607332080602646}, {"image_id": 16, "category_id": 3, "bbox": [908.3897705078125, 170.68675231933594, 210.3905029296875, 387.3239288330078], "score": 0.14116838574409485}, {"image_id": 16, "category_id": 3, "bbox": [324.1534423828125, 280.1866149902344, 313.3809814453125, 439.8133850097656], "score": 0.1339721828699112}, {"image_id": 16, "category_id": 3, "bbox": [926.6473388671875, 235.84046936035156, 232.6497802734375, 417.4474334716797], "score": 0.10526320338249207}, {"image_id": 16, "category_id": 3, "bbox": [424.9783630371094, 278.9742126464844, 371.9918518066406, 441.0257873535156], "score": 0.10098551958799362}, {"image_id": 16, "category_id": 3, "bbox": [381.912353515625, 216.9392547607422, 271.4776611328125, 400.02284240722656], "score": 0.09634407609701157}, {"image_id": 16, "category_id": 3, "bbox": [922.7388305664062, 61.649505615234375, 229.36932373046875, 429.0456848144531], "score": 0.09247448295354843}, {"image_id": 16, "category_id": 3, "bbox": [464.1882629394531, 252.70892333984375, 271.3666076660156, 393.460205078125], "score": 0.08747979998588562}, {"image_id": 16, "category_id": 3, "bbox": [176.94850158691406, 0.0, 275.32530212402344, 360.2836608886719], "score": 0.07500462979078293}, {"image_id": 16, "category_id": 3, "bbox": [1025.8707275390625, 168.45895385742188, 207.7408447265625, 340.8694152832031], "score": 0.06891115754842758}, {"image_id": 16, "category_id": 3, "bbox": [826.5387573242188, 165.60958862304688, 254.59857177734375, 398.0992736816406], "score": 0.0668478012084961}, {"image_id": 16, "category_id": 3, "bbox": [652.6432495117188, 16.462806701660156, 555.1187133789062, 647.6706161499023], "score": 0.06487735360860825}, {"image_id": 16, "category_id": 3, "bbox": [226.284423828125, 5.047187805175781, 654.1099853515625, 662.524284362793], "score": 0.061036452651023865}, {"image_id": 16, "category_id": 3, "bbox": [568.3052978515625, 330.9673156738281, 337.3135986328125, 389.0326843261719], "score": 0.056971173733472824}, {"image_id": 16, "category_id": 3, "bbox": [535.3876953125, 212.96585083007812, 253.74530029296875, 402.7171936035156], "score": 0.05513722449541092}, {"image_id": 16, "category_id": 3, "bbox": [504.2130126953125, 161.31239318847656, 637.9130859375, 558.6876068115234], "score": 0.05499034374952316}, {"image_id": 16, "category_id": 3, "bbox": [130.55560302734375, 84.07486724853516, 600.1583862304688, 635.9251327514648], "score": 0.05390508845448494}, {"image_id": 16, "category_id": 3, "bbox": [256.838623046875, 208.4617462158203, 259.85198974609375, 408.2347869873047], "score": 0.05376997962594032}, {"image_id": 16, "category_id": 3, "bbox": [800.674560546875, 278.0810241699219, 266.830322265625, 398.8716735839844], "score": 0.05218616500496864}, {"image_id": 16, "category_id": 3, "bbox": [0.0, 11.810302734375, 595.8836669921875, 649.3958740234375], "score": 0.04942874237895012}, {"image_id": 16, "category_id": 3, "bbox": [357.4493103027344, 240.18296813964844, 664.8404235839844, 479.81703186035156], "score": 0.04893581569194794}, {"image_id": 16, "category_id": 3, "bbox": [888.44873046875, 307.0274963378906, 238.362060546875, 411.1767883300781], "score": 0.04624510183930397}, {"image_id": 16, "category_id": 3, "bbox": [796.6165771484375, 174.51051330566406, 483.3834228515625, 545.4894866943359], "score": 0.04518451914191246}, {"image_id": 16, "category_id": 3, "bbox": [1022.3482055664062, 43.18114471435547, 192.48663330078125, 409.8479995727539], "score": 0.043276216834783554}, {"image_id": 16, "category_id": 3, "bbox": [385.13409423828125, 558.0647583007812, 232.6156005859375, 161.93524169921875], "score": 0.042906444519758224}, {"image_id": 16, "category_id": 3, "bbox": [338.5818176269531, 471.57196044921875, 320.5621032714844, 248.42803955078125], "score": 0.039614252746105194}, {"image_id": 16, "category_id": 3, "bbox": [186.42694091796875, 48.5874137878418, 304.403564453125, 413.3409309387207], "score": 0.03960668668150902}, {"image_id": 16, "category_id": 3, "bbox": [271.910888671875, 405.84466552734375, 323.2139892578125, 314.15533447265625], "score": 0.0395544096827507}, {"image_id": 16, "category_id": 3, "bbox": [748.4271850585938, 250.6399383544922, 260.8472900390625, 387.2527618408203], "score": 0.03748486191034317}, {"image_id": 16, "category_id": 3, "bbox": [385.1438293457031, 0.0, 650.6490173339844, 586.452392578125], "score": 0.03702947124838829}, {"image_id": 16, "category_id": 3, "bbox": [0.0, 250.64659118652344, 653.8670043945312, 469.35340881347656], "score": 0.03537863492965698}, {"image_id": 16, "category_id": 3, "bbox": [0.0, 89.0743637084961, 367.0728759765625, 630.9256362915039], "score": 0.035141654312610626}, {"image_id": 16, "category_id": 3, "bbox": [134.569091796875, 0.0, 627.1005859375, 375.66448974609375], "score": 0.0347442701458931}, {"image_id": 17, "category_id": 1, "bbox": [642.640380859375, 155.69898986816406, 225.71331787109375, 380.4238739013672], "score": 0.7648375034332275}, {"image_id": 17, "category_id": 1, "bbox": [709.8895874023438, 13.42254638671875, 177.9163818359375, 440.2854309082031], "score": 0.458734393119812}, {"image_id": 17, "category_id": 1, "bbox": [627.34912109375, 18.419300079345703, 333.43218994140625, 498.51606369018555], "score": 0.37424737215042114}, {"image_id": 17, "category_id": 1, "bbox": [620.2101440429688, 226.9580535888672, 265.8079833984375, 423.08729553222656], "score": 0.28743839263916016}, {"image_id": 17, "category_id": 1, "bbox": [754.619384765625, 17.712295532226562, 213.39239501953125, 433.1717987060547], "score": 0.21018463373184204}, {"image_id": 17, "category_id": 1, "bbox": [248.20370483398438, 2.4769363403320312, 619.7076110839844, 678.7368087768555], "score": 0.19410593807697296}, {"image_id": 17, "category_id": 1, "bbox": [541.73193359375, 170.29168701171875, 314.62603759765625, 458.75823974609375], "score": 0.18471553921699524}, {"image_id": 17, "category_id": 1, "bbox": [714.63427734375, 0.0, 170.64453125, 288.231201171875], "score": 0.16654491424560547}, {"image_id": 17, "category_id": 1, "bbox": [0.0, 10.041023254394531, 614.572998046875, 655.2190475463867], "score": 0.1395576298236847}, {"image_id": 17, "category_id": 1, "bbox": [562.1514892578125, 4.9571685791015625, 269.9791259765625, 456.39512634277344], "score": 0.11050999909639359}, {"image_id": 17, "category_id": 1, "bbox": [413.08575439453125, 0.0, 546.5186767578125, 595.09326171875], "score": 0.10670723021030426}, {"image_id": 17, "category_id": 1, "bbox": [752.6062622070312, 0.0, 214.60028076171875, 284.84356689453125], "score": 0.1051381528377533}, {"image_id": 17, "category_id": 1, "bbox": [625.20751953125, 358.7413635253906, 272.4921875, 361.2586364746094], "score": 0.09037618339061737}, {"image_id": 17, "category_id": 1, "bbox": [710.1547241210938, 73.19877624511719, 298.57452392578125, 456.23854064941406], "score": 0.06896702945232391}, {"image_id": 17, "category_id": 1, "bbox": [123.26718139648438, 0.0, 629.2522888183594, 577.5272827148438], "score": 0.06626073271036148}, {"image_id": 17, "category_id": 1, "bbox": [176.06105041503906, 6.988197326660156, 270.39991760253906, 371.98155975341797], "score": 0.05491214245557785}, {"image_id": 17, "category_id": 1, "bbox": [471.49468994140625, 148.0978240966797, 314.69976806640625, 435.9379425048828], "score": 0.0538911372423172}, {"image_id": 17, "category_id": 1, "bbox": [927.9784545898438, 169.49703979492188, 226.39202880859375, 426.4973449707031], "score": 0.05239735543727875}, {"image_id": 17, "category_id": 1, "bbox": [559.3236083984375, 0.0, 519.92529296875, 586.4776000976562], "score": 0.0479855090379715}, {"image_id": 17, "category_id": 1, "bbox": [375.63616943359375, 165.5070343017578, 598.5910034179688, 554.4929656982422], "score": 0.046177081763744354}, {"image_id": 17, "category_id": 1, "bbox": [827.1849975585938, 0.0, 207.22430419921875, 271.7910461425781], "score": 0.04441769793629646}, {"image_id": 17, "category_id": 1, "bbox": [513.6051635742188, 355.4906005859375, 339.42755126953125, 364.5093994140625], "score": 0.0431843101978302}, {"image_id": 17, "category_id": 1, "bbox": [81.54586791992188, 173.18812561035156, 704.6744079589844, 546.8118743896484], "score": 0.03941582888364792}, {"image_id": 17, "category_id": 2, "bbox": [627.2880249023438, 303.3180847167969, 266.27679443359375, 341.3589172363281], "score": 0.2694793939590454}, {"image_id": 17, "category_id": 2, "bbox": [557.4025268554688, 346.1223449707031, 308.894775390625, 373.8776550292969], "score": 0.14485512673854828}, {"image_id": 17, "category_id": 2, "bbox": [560.3780517578125, 234.26344299316406, 292.88714599609375, 406.70225524902344], "score": 0.11618058383464813}, {"image_id": 17, "category_id": 2, "bbox": [456.5608825683594, 276.8895568847656, 333.2786560058594, 391.6845397949219], "score": 0.09265735745429993}, {"image_id": 17, "category_id": 2, "bbox": [176.06105041503906, 6.988197326660156, 270.39991760253906, 371.98155975341797], "score": 0.08762674033641815}, {"image_id": 17, "category_id": 2, "bbox": [777.381591796875, 319.29669189453125, 304.945556640625, 386.52154541015625], "score": 0.07967693358659744}, {"image_id": 17, "category_id": 2, "bbox": [636.3483276367188, 112.46546173095703, 240.86932373046875, 409.54772186279297], "score": 0.07723695039749146}, {"image_id": 17, "category_id": 2, "bbox": [710.54052734375, 259.0818176269531, 292.6044921875, 370.0212707519531], "score": 0.07368699461221695}, {"image_id": 17, "category_id": 2, "bbox": [888.1851806640625, 191.5070037841797, 256.00048828125, 433.1315460205078], "score": 0.06643594801425934}, {"image_id": 17, "category_id": 2, "bbox": [707.6183471679688, 351.13037109375, 309.5440673828125, 368.86962890625], "score": 0.061486709862947464}, {"image_id": 17, "category_id": 2, "bbox": [852.9856567382812, 436.72845458984375, 302.03375244140625, 283.27154541015625], "score": 0.05872819945216179}, {"image_id": 17, "category_id": 2, "bbox": [787.1588745117188, 205.80575561523438, 277.44415283203125, 406.5547180175781], "score": 0.05764651298522949}, {"image_id": 17, "category_id": 2, "bbox": [384.1004943847656, 327.140380859375, 341.2298889160156, 360.0958251953125], "score": 0.054869819432497025}, {"image_id": 17, "category_id": 2, "bbox": [239.642578125, 77.51788330078125, 641.6549072265625, 642.4821166992188], "score": 0.0511140301823616}, {"image_id": 17, "category_id": 2, "bbox": [761.0565185546875, 0.0, 196.028564453125, 343.58642578125], "score": 0.049717165529727936}, {"image_id": 17, "category_id": 2, "bbox": [261.4793395996094, 13.59781265258789, 315.5679626464844, 361.13540267944336], "score": 0.04895368590950966}, {"image_id": 17, "category_id": 2, "bbox": [882.2146606445312, 72.02662658691406, 239.12445068359375, 405.4775848388672], "score": 0.0469292588531971}, {"image_id": 17, "category_id": 2, "bbox": [718.8453979492188, 0.0, 161.7257080078125, 319.530517578125], "score": 0.04671063646674156}, {"image_id": 17, "category_id": 2, "bbox": [798.7554931640625, 104.35291290283203, 263.0467529296875, 403.11898040771484], "score": 0.04662745073437691}, {"image_id": 17, "category_id": 2, "bbox": [562.1514892578125, 4.9571685791015625, 269.9791259765625, 456.39512634277344], "score": 0.046216342598199844}, {"image_id": 17, "category_id": 2, "bbox": [531.407470703125, 90.97057342529297, 541.0201416015625, 629.029426574707], "score": 0.04612142592668533}, {"image_id": 17, "category_id": 2, "bbox": [399.5692443847656, 4.198940277099609, 560.3531799316406, 666.7764015197754], "score": 0.04540373757481575}, {"image_id": 17, "category_id": 2, "bbox": [802.0678100585938, 6.163261413574219, 230.70526123046875, 386.36637115478516], "score": 0.04321713745594025}, {"image_id": 17, "category_id": 2, "bbox": [0.0, 9.433719635009766, 324.7626953125, 376.1458396911621], "score": 0.04228012263774872}, {"image_id": 17, "category_id": 2, "bbox": [941.8444213867188, 120.73287963867188, 212.88714599609375, 380.136962890625], "score": 0.04207183048129082}, {"image_id": 17, "category_id": 2, "bbox": [913.3199462890625, 315.23919677734375, 269.2445068359375, 377.40972900390625], "score": 0.041885875165462494}, {"image_id": 17, "category_id": 2, "bbox": [103.86776733398438, 3.766284942626953, 650.8767395019531, 666.1833000183105], "score": 0.04067213088274002}, {"image_id": 17, "category_id": 2, "bbox": [867.0841674804688, 0.0, 232.04888916015625, 352.4938049316406], "score": 0.04011482000350952}, {"image_id": 17, "category_id": 2, "bbox": [778.3263549804688, 465.1954040527344, 317.91082763671875, 254.80459594726562], "score": 0.03932567313313484}, {"image_id": 17, "category_id": 2, "bbox": [658.2764282226562, 90.51498413085938, 572.1337280273438, 629.4850158691406], "score": 0.03889114409685135}, {"image_id": 17, "category_id": 2, "bbox": [191.97482299804688, 0.0, 315.9578552246094, 230.13912963867188], "score": 0.03879811614751816}, {"image_id": 17, "category_id": 2, "bbox": [546.93994140625, 0.0, 266.18017578125, 286.6252746582031], "score": 0.038391221314668655}, {"image_id": 17, "category_id": 2, "bbox": [1014.8348388671875, 87.3218765258789, 217.29443359375, 374.9229965209961], "score": 0.03828512132167816}, {"image_id": 17, "category_id": 3, "bbox": [627.2880249023438, 303.3180847167969, 266.27679443359375, 341.3589172363281], "score": 0.5789417028427124}, {"image_id": 17, "category_id": 3, "bbox": [642.640380859375, 155.69898986816406, 225.71331787109375, 380.4238739013672], "score": 0.3278292715549469}, {"image_id": 17, "category_id": 3, "bbox": [560.3780517578125, 234.26344299316406, 292.88714599609375, 406.70225524902344], "score": 0.25536417961120605}, {"image_id": 17, "category_id": 3, "bbox": [627.34912109375, 18.419300079345703, 333.43218994140625, 498.51606369018555], "score": 0.17199085652828217}, {"image_id": 17, "category_id": 3, "bbox": [557.4025268554688, 346.1223449707031, 308.894775390625, 373.8776550292969], "score": 0.15819358825683594}, {"image_id": 17, "category_id": 3, "bbox": [188.04348754882812, 6.398540496826172, 275.52239990234375, 374.3571662902832], "score": 0.154961958527565}, {"image_id": 17, "category_id": 3, "bbox": [399.5692443847656, 4.198940277099609, 560.3531799316406, 666.7764015197754], "score": 0.14968079328536987}, {"image_id": 17, "category_id": 3, "bbox": [709.8895874023438, 13.42254638671875, 177.9163818359375, 440.2854309082031], "score": 0.13697436451911926}, {"image_id": 17, "category_id": 3, "bbox": [754.619384765625, 17.712295532226562, 213.39239501953125, 433.1717987060547], "score": 0.11727626621723175}, {"image_id": 17, "category_id": 3, "bbox": [714.63427734375, 0.0, 170.64453125, 288.231201171875], "score": 0.11073663830757141}, {"image_id": 17, "category_id": 3, "bbox": [708.415283203125, 283.960693359375, 298.839111328125, 381.5162353515625], "score": 0.10457336902618408}, {"image_id": 17, "category_id": 3, "bbox": [239.642578125, 77.51788330078125, 641.6549072265625, 642.4821166992188], "score": 0.10010140389204025}, {"image_id": 17, "category_id": 3, "bbox": [456.5608825683594, 276.8895568847656, 333.2786560058594, 391.6845397949219], "score": 0.09778010100126266}, {"image_id": 17, "category_id": 3, "bbox": [710.1547241210938, 73.19877624511719, 298.57452392578125, 456.23854064941406], "score": 0.09644732624292374}, {"image_id": 17, "category_id": 3, "bbox": [752.6062622070312, 0.0, 214.60028076171875, 284.84356689453125], "score": 0.09482651948928833}, {"image_id": 17, "category_id": 3, "bbox": [927.9784545898438, 169.49703979492188, 226.39202880859375, 426.4973449707031], "score": 0.08590667694807053}, {"image_id": 17, "category_id": 3, "bbox": [515.4315795898438, 164.00503540039062, 570.2935180664062, 555.9949645996094], "score": 0.0858064517378807}, {"image_id": 17, "category_id": 3, "bbox": [103.86776733398438, 3.766284942626953, 650.8767395019531, 666.1833000183105], "score": 0.08361556380987167}, {"image_id": 17, "category_id": 3, "bbox": [777.381591796875, 319.29669189453125, 304.945556640625, 386.52154541015625], "score": 0.07980667799711227}, {"image_id": 17, "category_id": 3, "bbox": [359.8238525390625, 244.33580017089844, 638.1507568359375, 475.66419982910156], "score": 0.07391756027936935}, {"image_id": 17, "category_id": 3, "bbox": [559.3236083984375, 0.0, 519.92529296875, 586.4776000976562], "score": 0.07355862855911255}, {"image_id": 17, "category_id": 3, "bbox": [457.28759765625, 105.8379898071289, 331.62744140625, 452.5794906616211], "score": 0.07215924561023712}, {"image_id": 17, "category_id": 3, "bbox": [658.2764282226562, 90.51498413085938, 572.1337280273438, 629.4850158691406], "score": 0.07176879048347473}, {"image_id": 17, "category_id": 3, "bbox": [629.640869140625, 360.28125, 325.51025390625, 359.71875], "score": 0.06696338206529617}, {"image_id": 17, "category_id": 3, "bbox": [562.1514892578125, 4.9571685791015625, 269.9791259765625, 456.39512634277344], "score": 0.06498369574546814}, {"image_id": 17, "category_id": 3, "bbox": [795.4771728515625, 142.04312133789062, 269.571044921875, 394.7601013183594], "score": 0.06374162435531616}, {"image_id": 17, "category_id": 3, "bbox": [0.0, 82.62295532226562, 623.1898803710938, 637.3770446777344], "score": 0.06221986189484596}, {"image_id": 17, "category_id": 3, "bbox": [936.7857666015625, 77.63832092285156, 228.3360595703125, 393.3939971923828], "score": 0.060388896614313126}, {"image_id": 17, "category_id": 3, "bbox": [133.6034698486328, 34.106849670410156, 265.5951385498047, 372.94795989990234], "score": 0.059565313160419464}, {"image_id": 17, "category_id": 3, "bbox": [833.68310546875, 0.0, 186.2125244140625, 300.0516357421875], "score": 0.057351432740688324}, {"image_id": 17, "category_id": 3, "bbox": [881.1314086914062, 275.140869140625, 267.21331787109375, 392.84820556640625], "score": 0.05715303122997284}, {"image_id": 17, "category_id": 3, "bbox": [884.6882934570312, 109.40875244140625, 241.95269775390625, 400.7122802734375], "score": 0.055558882653713226}, {"image_id": 17, "category_id": 3, "bbox": [262.99261474609375, 25.98273468017578, 304.71624755859375, 388.0404281616211], "score": 0.054053790867328644}, {"image_id": 17, "category_id": 3, "bbox": [776.2408447265625, 19.730518341064453, 503.7591552734375, 621.8473014831543], "score": 0.04952942207455635}, {"image_id": 17, "category_id": 3, "bbox": [232.09915161132812, 0.0, 643.8135070800781, 498.2525634765625], "score": 0.04799845069646835}, {"image_id": 17, "category_id": 3, "bbox": [882.1927490234375, 156.61203002929688, 397.8072509765625, 563.3879699707031], "score": 0.04705141857266426}, {"image_id": 17, "category_id": 3, "bbox": [71.58273315429688, 236.45701599121094, 729.8620300292969, 483.54298400878906], "score": 0.04608164727687836}, {"image_id": 17, "category_id": 3, "bbox": [384.1004943847656, 327.140380859375, 341.2298889160156, 360.0958251953125], "score": 0.04371112585067749}, {"image_id": 17, "category_id": 3, "bbox": [648.4258422851562, 0.0, 227.63140869140625, 330.24798583984375], "score": 0.04368837922811508}, {"image_id": 17, "category_id": 3, "bbox": [328.321044921875, 6.271081924438477, 317.582275390625, 377.4578609466553], "score": 0.04149802774190903}, {"image_id": 17, "category_id": 3, "bbox": [0.0, 0.0, 636.8208618164062, 486.11431884765625], "score": 0.04147428646683693}, {"image_id": 17, "category_id": 3, "bbox": [371.3867492675781, 0.0, 644.7054138183594, 424.3362731933594], "score": 0.04110519587993622}, {"image_id": 17, "category_id": 3, "bbox": [652.0845947265625, 0.0, 600.3082275390625, 496.5477600097656], "score": 0.04083681479096413}, {"image_id": 17, "category_id": 3, "bbox": [866.6705932617188, 3.4014015197753906, 246.23004150390625, 390.69167709350586], "score": 0.038429342210292816}, {"image_id": 18, "category_id": 1, "bbox": [399.3203125, 208.83363342285156, 182.7967529296875, 416.86949157714844], "score": 0.8311252593994141}, {"image_id": 18, "category_id": 1, "bbox": [365.54168701171875, 238.3986053466797, 243.45892333984375, 474.71498107910156], "score": 0.6822131872177124}, {"image_id": 18, "category_id": 1, "bbox": [646.476806640625, 340.6021728515625, 171.64959716796875, 345.20819091796875], "score": 0.24593405425548553}, {"image_id": 18, "category_id": 1, "bbox": [722.4418334960938, 343.31280517578125, 178.3448486328125, 344.2802734375], "score": 0.2243434190750122}, {"image_id": 18, "category_id": 1, "bbox": [316.36541748046875, 135.3047637939453, 271.07562255859375, 435.6820526123047], "score": 0.21753700077533722}, {"image_id": 18, "category_id": 1, "bbox": [7.19024658203125, 15.80826187133789, 591.340576171875, 647.6103172302246], "score": 0.1466938555240631}, {"image_id": 18, "category_id": 1, "bbox": [441.40289306640625, 179.20196533203125, 326.45159912109375, 462.87567138671875], "score": 0.11606841534376144}, {"image_id": 18, "category_id": 1, "bbox": [620.6363525390625, 366.08489990234375, 251.36920166015625, 353.91510009765625], "score": 0.10744303464889526}, {"image_id": 18, "category_id": 1, "bbox": [625.2373657226562, 232.4219512939453, 235.98431396484375, 424.23057556152344], "score": 0.10626339912414551}, {"image_id": 18, "category_id": 1, "bbox": [273.296875, 8.289321899414062, 549.80859375, 663.2640838623047], "score": 0.1059536561369896}, {"image_id": 18, "category_id": 1, "bbox": [698.5021362304688, 390.3028259277344, 243.13519287109375, 329.6971740722656], "score": 0.09309118986129761}, {"image_id": 18, "category_id": 1, "bbox": [211.33445739746094, 176.27047729492188, 313.41627502441406, 461.6811218261719], "score": 0.08671552687883377}, {"image_id": 18, "category_id": 1, "bbox": [374.04400634765625, 63.0802001953125, 224.32049560546875, 432.2757568359375], "score": 0.07621091604232788}, {"image_id": 18, "category_id": 1, "bbox": [150.67730712890625, 87.24994659423828, 543.9344482421875, 632.7500534057617], "score": 0.07576047629117966}, {"image_id": 18, "category_id": 1, "bbox": [469.2803955078125, 312.87628173828125, 254.69354248046875, 405.8546142578125], "score": 0.07135026156902313}, {"image_id": 18, "category_id": 1, "bbox": [926.2571411132812, 165.7624053955078, 225.25531005859375, 430.77915954589844], "score": 0.0690477043390274}, {"image_id": 18, "category_id": 1, "bbox": [378.72991943359375, 4.6676788330078125, 626.1039428710938, 653.0517425537109], "score": 0.06764914840459824}, {"image_id": 18, "category_id": 1, "bbox": [1058.138671875, 0.0, 221.861328125, 371.155517578125], "score": 0.06072165444493294}, {"image_id": 18, "category_id": 1, "bbox": [806.2705078125, 171.50828552246094, 473.7294921875, 548.4917144775391], "score": 0.05551741272211075}, {"image_id": 18, "category_id": 1, "bbox": [293.9404296875, 371.70257568359375, 315.2745361328125, 346.0548095703125], "score": 0.05381467193365097}, {"image_id": 18, "category_id": 1, "bbox": [0.0, 10.865409851074219, 374.0946350097656, 641.0012283325195], "score": 0.05233713611960411}, {"image_id": 18, "category_id": 1, "bbox": [0.0, 235.07273864746094, 645.2622680664062, 484.92726135253906], "score": 0.05222594738006592}, {"image_id": 18, "category_id": 1, "bbox": [535.991455078125, 0.0, 590.5478515625, 582.2379150390625], "score": 0.0489974208176136}, {"image_id": 18, "category_id": 1, "bbox": [546.1051025390625, 327.04864501953125, 256.87158203125, 377.16180419921875], "score": 0.048086073249578476}, {"image_id": 18, "category_id": 1, "bbox": [367.6881103515625, 95.8426284790039, 360.782958984375, 489.6484603881836], "score": 0.04687792807817459}, {"image_id": 18, "category_id": 1, "bbox": [892.5545654296875, 256.63653564453125, 253.4176025390625, 424.99078369140625], "score": 0.046792056411504745}, {"image_id": 18, "category_id": 1, "bbox": [495.47747802734375, 161.63047790527344, 642.3250122070312, 558.3695220947266], "score": 0.045307327061891556}, {"image_id": 18, "category_id": 1, "bbox": [202.5981903076172, 34.071006774902344, 240.05918884277344, 388.4756546020508], "score": 0.04360231012105942}, {"image_id": 18, "category_id": 1, "bbox": [14.1793212890625, 0.0, 610.4825439453125, 365.2908020019531], "score": 0.042837247252464294}, {"image_id": 18, "category_id": 1, "bbox": [682.4840087890625, 186.42723083496094, 265.7386474609375, 451.67677307128906], "score": 0.04089988023042679}, {"image_id": 18, "category_id": 1, "bbox": [719.1195678710938, 270.8828430175781, 271.4766845703125, 416.6634216308594], "score": 0.040136415511369705}, {"image_id": 18, "category_id": 1, "bbox": [303.7578430175781, 476.8208923339844, 366.8321838378906, 243.17910766601562], "score": 0.03972431644797325}, {"image_id": 18, "category_id": 1, "bbox": [261.5706787109375, 232.38951110839844, 597.423828125, 487.61048889160156], "score": 0.03971507400274277}, {"image_id": 18, "category_id": 1, "bbox": [647.2861328125, 0.0, 598.2288818359375, 362.5942687988281], "score": 0.038032177835702896}, {"image_id": 18, "category_id": 2, "bbox": [302.3157958984375, 414.4477844238281, 370.93206787109375, 305.5522155761719], "score": 0.20930245518684387}, {"image_id": 18, "category_id": 2, "bbox": [546.1051025390625, 327.04864501953125, 256.87158203125, 377.16180419921875], "score": 0.08744240552186966}, {"image_id": 18, "category_id": 2, "bbox": [381.5999755859375, 476.9804382324219, 347.48406982421875, 243.01956176757812], "score": 0.07954926043748856}, {"image_id": 18, "category_id": 2, "bbox": [469.2803955078125, 312.87628173828125, 254.69354248046875, 405.8546142578125], "score": 0.07507217675447464}, {"image_id": 18, "category_id": 2, "bbox": [722.4418334960938, 343.31280517578125, 178.3448486328125, 344.2802734375], "score": 0.07198373228311539}, {"image_id": 18, "category_id": 2, "bbox": [646.476806640625, 340.6021728515625, 171.64959716796875, 345.20819091796875], "score": 0.07030263543128967}, {"image_id": 18, "category_id": 2, "bbox": [516.07177734375, 476.368408203125, 338.5997314453125, 243.631591796875], "score": 0.06532739847898483}, {"image_id": 18, "category_id": 2, "bbox": [283.5845947265625, 282.197998046875, 306.40130615234375, 437.802001953125], "score": 0.06179151311516762}, {"image_id": 18, "category_id": 2, "bbox": [399.3203125, 208.83363342285156, 182.7967529296875, 416.86949157714844], "score": 0.06143924593925476}, {"image_id": 18, "category_id": 2, "bbox": [176.12423706054688, 62.01991653442383, 260.8713684082031, 396.5207633972168], "score": 0.054708462208509445}, {"image_id": 18, "category_id": 2, "bbox": [724.3902587890625, 430.767578125, 275.361572265625, 289.232421875], "score": 0.054218851029872894}, {"image_id": 18, "category_id": 2, "bbox": [811.0269775390625, 440.5391845703125, 262.1390380859375, 279.4608154296875], "score": 0.0493231900036335}, {"image_id": 18, "category_id": 2, "bbox": [136.8541717529297, 445.76678466796875, 322.35292053222656, 274.23321533203125], "score": 0.04711971431970596}, {"image_id": 18, "category_id": 2, "bbox": [656.072265625, 474.6877136230469, 293.760009765625, 245.31228637695312], "score": 0.045958537608385086}, {"image_id": 18, "category_id": 2, "bbox": [900.1009521484375, 194.52024841308594, 242.98193359375, 431.92152404785156], "score": 0.04459964856505394}, {"image_id": 18, "category_id": 2, "bbox": [127.78926849365234, 32.41390609741211, 264.49292755126953, 385.48080825805664], "score": 0.04278760403394699}, {"image_id": 18, "category_id": 2, "bbox": [721.1678466796875, 234.2992706298828, 263.803466796875, 421.04002380371094], "score": 0.04277132824063301}, {"image_id": 18, "category_id": 2, "bbox": [865.000732421875, 398.81805419921875, 267.694091796875, 321.18194580078125], "score": 0.04218815267086029}, {"image_id": 18, "category_id": 2, "bbox": [803.3971557617188, 229.92938232421875, 288.75714111328125, 423.7611083984375], "score": 0.03997331112623215}, {"image_id": 18, "category_id": 2, "bbox": [221.5021514892578, 122.79947662353516, 288.9256134033203, 441.89608001708984], "score": 0.03965191915631294}, {"image_id": 18, "category_id": 2, "bbox": [17.797882080078125, 69.74903869628906, 294.515625, 394.51893615722656], "score": 0.039341673254966736}, {"image_id": 18, "category_id": 2, "bbox": [927.3861083984375, 80.94548034667969, 232.7408447265625, 383.9580841064453], "score": 0.03900161758065224}, {"image_id": 18, "category_id": 2, "bbox": [286.16064453125, 158.5311737060547, 534.67529296875, 561.4688262939453], "score": 0.03815891221165657}, {"image_id": 18, "category_id": 2, "bbox": [209.35533142089844, 467.666748046875, 323.96925354003906, 252.333251953125], "score": 0.03812236338853836}, {"image_id": 18, "category_id": 2, "bbox": [533.1817016601562, 201.4303741455078, 288.11578369140625, 424.6631317138672], "score": 0.03806813061237335}, {"image_id": 18, "category_id": 2, "bbox": [203.09947204589844, 0.0, 244.0326690673828, 363.9478759765625], "score": 0.03801187500357628}, {"image_id": 18, "category_id": 2, "bbox": [1056.2508544921875, 80.65677642822266, 223.7491455078125, 383.92501068115234], "score": 0.03753840550780296}, {"image_id": 18, "category_id": 2, "bbox": [316.36541748046875, 135.3047637939453, 271.07562255859375, 435.6820526123047], "score": 0.03720267489552498}, {"image_id": 18, "category_id": 3, "bbox": [321.35113525390625, 275.3378601074219, 320.51458740234375, 444.6621398925781], "score": 0.3923160135746002}, {"image_id": 18, "category_id": 3, "bbox": [378.07537841796875, 373.43023681640625, 356.179443359375, 342.35174560546875], "score": 0.31821033358573914}, {"image_id": 18, "category_id": 3, "bbox": [301.011474609375, 454.5102233886719, 373.0806884765625, 265.4897766113281], "score": 0.25587278604507446}, {"image_id": 18, "category_id": 3, "bbox": [399.3203125, 208.83363342285156, 182.7967529296875, 416.86949157714844], "score": 0.25195708870887756}, {"image_id": 18, "category_id": 3, "bbox": [647.7015991210938, 299.2430725097656, 178.05328369140625, 357.9086608886719], "score": 0.16788813471794128}, {"image_id": 18, "category_id": 3, "bbox": [467.2480773925781, 273.23577880859375, 262.4407653808594, 415.36444091796875], "score": 0.13269028067588806}, {"image_id": 18, "category_id": 3, "bbox": [723.7334594726562, 301.4986267089844, 183.9490966796875, 357.7221374511719], "score": 0.12863761186599731}, {"image_id": 18, "category_id": 3, "bbox": [286.16064453125, 158.5311737060547, 534.67529296875, 561.4688262939453], "score": 0.11346122622489929}, {"image_id": 18, "category_id": 3, "bbox": [520.19580078125, 394.83563232421875, 296.5428466796875, 325.16436767578125], "score": 0.10685143619775772}, {"image_id": 18, "category_id": 3, "bbox": [150.67730712890625, 87.24994659423828, 543.9344482421875, 632.7500534057617], "score": 0.09723170846700668}, {"image_id": 18, "category_id": 3, "bbox": [202.5981903076172, 34.071006774902344, 240.05918884277344, 388.4756546020508], "score": 0.09371877461671829}, {"image_id": 18, "category_id": 3, "bbox": [211.33445739746094, 176.27047729492188, 313.41627502441406, 461.6811218261719], "score": 0.08003179728984833}, {"image_id": 18, "category_id": 3, "bbox": [441.40289306640625, 179.20196533203125, 326.45159912109375, 462.87567138671875], "score": 0.07789149135351181}, {"image_id": 18, "category_id": 3, "bbox": [0.0, 235.07273864746094, 645.2622680664062, 484.92726135253906], "score": 0.07733529061079025}, {"image_id": 18, "category_id": 3, "bbox": [450.97222900390625, 474.2694091796875, 301.54693603515625, 245.7305908203125], "score": 0.0763150006532669}, {"image_id": 18, "category_id": 3, "bbox": [7.19024658203125, 15.80826187133789, 591.340576171875, 647.6103172302246], "score": 0.07627503573894501}, {"image_id": 18, "category_id": 3, "bbox": [354.5708923339844, 226.02536010742188, 656.2649230957031, 493.9746398925781], "score": 0.0760548859834671}, {"image_id": 18, "category_id": 3, "bbox": [316.36541748046875, 135.3047637939453, 271.07562255859375, 435.6820526123047], "score": 0.07517086714506149}, {"image_id": 18, "category_id": 3, "bbox": [551.7643432617188, 280.83978271484375, 263.37603759765625, 391.72967529296875], "score": 0.06793973594903946}, {"image_id": 18, "category_id": 3, "bbox": [926.2571411132812, 165.7624053955078, 225.25531005859375, 430.77915954589844], "score": 0.06740720570087433}, {"image_id": 18, "category_id": 3, "bbox": [495.47747802734375, 161.63047790527344, 642.3250122070312, 558.3695220947266], "score": 0.0636921301484108}, {"image_id": 18, "category_id": 3, "bbox": [613.4796142578125, 386.6827087402344, 259.681640625, 333.3172912597656], "score": 0.0630265399813652}, {"image_id": 18, "category_id": 3, "bbox": [720.7098388671875, 390.5624084472656, 275.11053466796875, 329.4375915527344], "score": 0.05813354253768921}, {"image_id": 18, "category_id": 3, "bbox": [892.5545654296875, 256.63653564453125, 253.4176025390625, 424.99078369140625], "score": 0.05630258843302727}, {"image_id": 18, "category_id": 3, "bbox": [806.2705078125, 171.50828552246094, 473.7294921875, 548.4917144775391], "score": 0.05583582818508148}, {"image_id": 18, "category_id": 3, "bbox": [378.72991943359375, 4.6676788330078125, 626.1039428710938, 653.0517425537109], "score": 0.04974007233977318}, {"image_id": 18, "category_id": 3, "bbox": [0.0, 87.19134521484375, 363.3289794921875, 632.8086547851562], "score": 0.04743045195937157}, {"image_id": 18, "category_id": 3, "bbox": [209.1873321533203, 331.3917236328125, 320.51939392089844, 388.6082763671875], "score": 0.04735464230179787}, {"image_id": 18, "category_id": 3, "bbox": [261.3817443847656, 0.0, 601.8935241699219, 580.4398193359375], "score": 0.046188417822122574}, {"image_id": 18, "category_id": 3, "bbox": [535.991455078125, 0.0, 590.5478515625, 582.2379150390625], "score": 0.04467049241065979}, {"image_id": 18, "category_id": 3, "bbox": [127.78926849365234, 32.41390609741211, 264.49292755126953, 385.48080825805664], "score": 0.04389302805066109}, {"image_id": 18, "category_id": 3, "bbox": [927.3861083984375, 80.94548034667969, 232.7408447265625, 383.9580841064453], "score": 0.04347451031208038}, {"image_id": 18, "category_id": 3, "bbox": [374.04400634765625, 63.0802001953125, 224.32049560546875, 432.2757568359375], "score": 0.04309418052434921}, {"image_id": 18, "category_id": 3, "bbox": [721.1678466796875, 234.2992706298828, 263.803466796875, 421.04002380371094], "score": 0.042947448790073395}, {"image_id": 18, "category_id": 3, "bbox": [208.94036865234375, 89.63260650634766, 279.2989501953125, 432.12459564208984], "score": 0.04247027263045311}, {"image_id": 18, "category_id": 3, "bbox": [367.6881103515625, 95.8426284790039, 360.782958984375, 489.6484603881836], "score": 0.041150644421577454}, {"image_id": 18, "category_id": 3, "bbox": [642.7740478515625, 25.01525115966797, 612.30712890625, 621.1850051879883], "score": 0.040001191198825836}, {"image_id": 18, "category_id": 3, "bbox": [629.7042846679688, 0.0, 612.2349243164062, 424.857421875], "score": 0.03994756191968918}, {"image_id": 19, "category_id": 1, "bbox": [1083.5552978515625, 121.27059936523438, 228.735595703125, 624.4814147949219], "score": 0.5378648638725281}, {"image_id": 19, "category_id": 1, "bbox": [1095.45751953125, 124.42922973632812, 348.4569091796875, 621.9758605957031], "score": 0.3355593979358673}, {"image_id": 19, "category_id": 1, "bbox": [1058.4752197265625, 276.05023193359375, 290.092041015625, 604.0613403320312], "score": 0.21437472105026245}, {"image_id": 19, "category_id": 1, "bbox": [972.6149291992188, 214.06674194335938, 357.96490478515625, 582.3255310058594], "score": 0.196515753865242}, {"image_id": 19, "category_id": 1, "bbox": [384.75030517578125, 8.526128768920898, 919.6034545898438, 998.1067447662354], "score": 0.17202228307724}, {"image_id": 19, "category_id": 1, "bbox": [10.40277099609375, 16.570232391357422, 919.6292724609375, 977.7473335266113], "score": 0.1427760273218155}, {"image_id": 19, "category_id": 1, "bbox": [172.7029266357422, 101.75118255615234, 956.0100860595703, 978.2488174438477], "score": 0.12031728774309158}, {"image_id": 19, "category_id": 1, "bbox": [951.8260498046875, 309.17303466796875, 369.1380615234375, 664.3521118164062], "score": 0.08533389866352081}, {"image_id": 19, "category_id": 1, "bbox": [1209.674072265625, 235.85670471191406, 710.325927734375, 844.1432952880859], "score": 0.0849035233259201}, {"image_id": 19, "category_id": 1, "bbox": [232.0223846435547, 154.03482055664062, 354.2984161376953, 579.1134948730469], "score": 0.08398036658763885}, {"image_id": 19, "category_id": 1, "bbox": [564.4915161132812, 234.95079040527344, 907.3417358398438, 845.0492095947266], "score": 0.08222981542348862}, {"image_id": 19, "category_id": 1, "bbox": [209.3423309326172, 13.704036712646484, 374.99415588378906, 550.5466957092285], "score": 0.07037705928087234}, {"image_id": 19, "category_id": 1, "bbox": [1054.5245361328125, 0.0, 309.7841796875, 588.5333862304688], "score": 0.06612008064985275}, {"image_id": 19, "category_id": 1, "bbox": [971.2454223632812, 43.616939544677734, 308.84051513671875, 650.9736976623535], "score": 0.06537369638681412}, {"image_id": 19, "category_id": 1, "bbox": [606.6021728515625, 0.0, 836.741455078125, 877.9691162109375], "score": 0.06344739347696304}, {"image_id": 19, "category_id": 1, "bbox": [1609.6180419921875, 0.0, 310.3819580078125, 637.5390625], "score": 0.06304096430540085}, {"image_id": 19, "category_id": 1, "bbox": [1360.6820068359375, 11.725622177124023, 559.3179931640625, 973.8843631744385], "score": 0.0547214038670063}, {"image_id": 19, "category_id": 1, "bbox": [852.2474365234375, 201.65789794921875, 413.0628662109375, 609.2301635742188], "score": 0.05065874755382538}, {"image_id": 19, "category_id": 1, "bbox": [0.0, 231.03955078125, 747.4063720703125, 848.96044921875], "score": 0.04972207546234131}, {"image_id": 19, "category_id": 1, "bbox": [984.03369140625, 5.130117416381836, 889.6597900390625, 965.2625217437744], "score": 0.049543242901563644}, {"image_id": 19, "category_id": 1, "bbox": [0.0, 24.19112205505371, 557.1464233398438, 958.66526222229], "score": 0.04752582684159279}, {"image_id": 19, "category_id": 1, "bbox": [1169.9854736328125, 135.29786682128906, 385.6878662109375, 641.7587738037109], "score": 0.043798353523015976}, {"image_id": 19, "category_id": 1, "bbox": [332.64019775390625, 446.6051940917969, 999.8425903320312, 633.3948059082031], "score": 0.04351605102419853}, {"image_id": 19, "category_id": 1, "bbox": [0.0, 465.058837890625, 977.2416381835938, 614.941162109375], "score": 0.04312947392463684}, {"image_id": 19, "category_id": 1, "bbox": [338.8112487792969, 0.0, 1006.6994934082031, 632.661376953125], "score": 0.041987013071775436}, {"image_id": 19, "category_id": 1, "bbox": [1175.132568359375, 0.0, 744.867431640625, 748.4298706054688], "score": 0.039251185953617096}, {"image_id": 19, "category_id": 1, "bbox": [800.4139404296875, 324.8336486816406, 864.024658203125, 755.1663513183594], "score": 0.03872949257493019}, {"image_id": 19, "category_id": 1, "bbox": [828.9464721679688, 0.0, 371.56866455078125, 566.5751342773438], "score": 0.038453709334135056}, {"image_id": 19, "category_id": 1, "bbox": [5.962005615234375, 0.0, 973.1551208496094, 618.959716796875], "score": 0.0383325032889843}, {"image_id": 19, "category_id": 1, "bbox": [1583.55908203125, 170.4002685546875, 336.44091796875, 683.7091674804688], "score": 0.03760361298918724}, {"image_id": 19, "category_id": 2, "bbox": [972.6149291992188, 214.06674194335938, 357.96490478515625, 582.3255310058594], "score": 0.13344503939151764}, {"image_id": 19, "category_id": 2, "bbox": [1061.555419921875, 177.15061950683594, 402.3941650390625, 626.0648956298828], "score": 0.07584280520677567}, {"image_id": 19, "category_id": 2, "bbox": [906.649658203125, 490.4180603027344, 414.161376953125, 562.4248352050781], "score": 0.07568567991256714}, {"image_id": 19, "category_id": 2, "bbox": [1058.4752197265625, 276.05023193359375, 290.092041015625, 604.0613403320312], "score": 0.0735924169421196}, {"image_id": 19, "category_id": 2, "bbox": [1083.5552978515625, 121.27059936523438, 228.735595703125, 624.4814147949219], "score": 0.07207722961902618}, {"image_id": 19, "category_id": 2, "bbox": [861.1846313476562, 153.22219848632812, 391.92437744140625, 617.7627868652344], "score": 0.06676746159791946}, {"image_id": 19, "category_id": 2, "bbox": [823.4144897460938, 429.1307373046875, 431.06524658203125, 568.1822509765625], "score": 0.06295453011989594}, {"image_id": 19, "category_id": 2, "bbox": [182.9867706298828, 192.42616271972656, 406.2402801513672, 603.0377044677734], "score": 0.062301456928253174}, {"image_id": 19, "category_id": 2, "bbox": [992.4437255859375, 554.4766235351562, 392.9923095703125, 525.5233764648438], "score": 0.05489707738161087}, {"image_id": 19, "category_id": 2, "bbox": [49.389793395996094, 204.04562377929688, 444.58521270751953, 596.2515563964844], "score": 0.05435497313737869}, {"image_id": 19, "category_id": 2, "bbox": [204.5184783935547, 61.10201644897461, 374.19044494628906, 572.5913429260254], "score": 0.0534067302942276}, {"image_id": 19, "category_id": 2, "bbox": [744.5196533203125, 99.6021499633789, 396.0238037109375, 613.7987899780273], "score": 0.050507158041000366}, {"image_id": 19, "category_id": 2, "bbox": [1091.6768798828125, 477.3342590332031, 396.825927734375, 584.3298034667969], "score": 0.05017963424324989}, {"image_id": 19, "category_id": 2, "bbox": [22.055831909179688, 329.1246032714844, 426.83363342285156, 585.8310241699219], "score": 0.048516646027565}, {"image_id": 19, "category_id": 2, "bbox": [951.8260498046875, 309.17303466796875, 369.1380615234375, 664.3521118164062], "score": 0.04581855237483978}, {"image_id": 19, "category_id": 2, "bbox": [1169.9854736328125, 135.29786682128906, 385.6878662109375, 641.7587738037109], "score": 0.04551633819937706}, {"image_id": 19, "category_id": 2, "bbox": [130.42144775390625, 321.74322509765625, 425.14727783203125, 594.6079711914062], "score": 0.04543738439679146}, {"image_id": 19, "category_id": 2, "bbox": [311.60504150390625, 172.343017578125, 382.371337890625, 556.8453369140625], "score": 0.04534950107336044}, {"image_id": 19, "category_id": 2, "bbox": [734.2424926757812, 235.48228454589844, 433.84942626953125, 632.9399566650391], "score": 0.044453464448451996}, {"image_id": 19, "category_id": 2, "bbox": [689.9584350585938, 495.2972106933594, 489.53497314453125, 563.9757385253906], "score": 0.04359890893101692}, {"image_id": 19, "category_id": 2, "bbox": [0.0, 274.5169372558594, 339.6925354003906, 582.7585754394531], "score": 0.043590739369392395}, {"image_id": 19, "category_id": 2, "bbox": [384.75030517578125, 8.526128768920898, 919.6034545898438, 998.1067447662354], "score": 0.04220017045736313}, {"image_id": 19, "category_id": 2, "bbox": [971.2454223632812, 43.616939544677734, 308.84051513671875, 650.9736976623535], "score": 0.04172388091683388}, {"image_id": 19, "category_id": 2, "bbox": [714.184326171875, 0.0, 412.9075927734375, 472.53460693359375], "score": 0.04151882231235504}, {"image_id": 19, "category_id": 2, "bbox": [172.7029266357422, 101.75118255615234, 956.0100860595703, 978.2488174438477], "score": 0.041330285370349884}, {"image_id": 19, "category_id": 2, "bbox": [630.5449829101562, 115.80489349365234, 392.96240234375, 581.8255996704102], "score": 0.040975429117679596}, {"image_id": 19, "category_id": 2, "bbox": [844.8473510742188, 19.57660675048828, 768.7935180664062, 952.7991256713867], "score": 0.0408250130712986}, {"image_id": 19, "category_id": 2, "bbox": [416.5718994140625, 203.90652465820312, 410.5155029296875, 607.7856750488281], "score": 0.0408090241253376}, {"image_id": 19, "category_id": 2, "bbox": [593.6343383789062, 119.41899871826172, 853.6694946289062, 960.5810012817383], "score": 0.039982955902814865}, {"image_id": 19, "category_id": 2, "bbox": [62.652889251708984, 71.5713882446289, 444.43994522094727, 558.7232894897461], "score": 0.03956343233585358}, {"image_id": 19, "category_id": 2, "bbox": [803.3406372070312, 0.0, 409.33721923828125, 426.28814697265625], "score": 0.039108842611312866}, {"image_id": 19, "category_id": 2, "bbox": [298.03607177734375, 20.91777992248535, 406.2025146484375, 542.6248226165771], "score": 0.03902973234653473}, {"image_id": 19, "category_id": 2, "bbox": [7.951789855957031, 440.16748046875, 317.23033905029297, 597.554931640625], "score": 0.038969069719314575}, {"image_id": 19, "category_id": 2, "bbox": [1190.696533203125, 657.8096313476562, 429.1717529296875, 422.19036865234375], "score": 0.038737472146749496}, {"image_id": 19, "category_id": 2, "bbox": [996.1561889648438, 0.0, 399.83258056640625, 361.66510009765625], "score": 0.038505084812641144}, {"image_id": 19, "category_id": 2, "bbox": [1084.5267333984375, 0.0, 428.9188232421875, 363.90313720703125], "score": 0.03837126865983009}, {"image_id": 19, "category_id": 2, "bbox": [1017.2208862304688, 0.0, 844.3344116210938, 871.92919921875], "score": 0.03763663023710251}, {"image_id": 19, "category_id": 2, "bbox": [523.068115234375, 112.6292953491211, 385.1661376953125, 590.5076675415039], "score": 0.03694787621498108}, {"image_id": 19, "category_id": 3, "bbox": [1061.439453125, 240.22496032714844, 274.5799560546875, 597.1305084228516], "score": 0.606549859046936}, {"image_id": 19, "category_id": 3, "bbox": [974.9205322265625, 162.74070739746094, 345.029052734375, 604.3134918212891], "score": 0.34585994482040405}, {"image_id": 19, "category_id": 3, "bbox": [1061.555419921875, 177.15061950683594, 402.3941650390625, 626.0648956298828], "score": 0.31565138697624207}, {"image_id": 19, "category_id": 3, "bbox": [951.8260498046875, 309.17303466796875, 369.1380615234375, 664.3521118164062], "score": 0.25405964255332947}, {"image_id": 19, "category_id": 3, "bbox": [1026.9862060546875, 384.30938720703125, 343.874755859375, 636.9789428710938], "score": 0.1600458323955536}, {"image_id": 19, "category_id": 3, "bbox": [852.2474365234375, 201.65789794921875, 413.0628662109375, 609.2301635742188], "score": 0.1350499838590622}, {"image_id": 19, "category_id": 3, "bbox": [222.45452880859375, 107.30513000488281, 358.33551025390625, 578.2366790771484], "score": 0.12839359045028687}, {"image_id": 19, "category_id": 3, "bbox": [609.53466796875, 14.435468673706055, 824.3419189453125, 975.857988357544], "score": 0.12174081057310104}, {"image_id": 19, "category_id": 3, "bbox": [1074.3028564453125, 60.24495315551758, 248.936279296875, 643.4259452819824], "score": 0.12147335708141327}, {"image_id": 19, "category_id": 3, "bbox": [369.1385498046875, 108.63579559326172, 945.6484375, 971.3642044067383], "score": 0.1072656586766243}, {"image_id": 19, "category_id": 3, "bbox": [1080.687255859375, 319.0575866699219, 405.7197265625, 654.0152282714844], "score": 0.10543368756771088}, {"image_id": 19, "category_id": 3, "bbox": [14.940399169921875, 102.16429901123047, 908.7486877441406, 977.8357009887695], "score": 0.08965212106704712}, {"image_id": 19, "category_id": 3, "bbox": [204.9681854248047, 0.0, 388.2613067626953, 510.0557861328125], "score": 0.08299317210912704}, {"image_id": 19, "category_id": 3, "bbox": [831.1487426757812, 130.16346740722656, 795.6423950195312, 949.8365325927734], "score": 0.08285055309534073}, {"image_id": 19, "category_id": 3, "bbox": [964.263427734375, 234.05503845214844, 914.1812744140625, 845.9449615478516], "score": 0.07750314474105835}, {"image_id": 19, "category_id": 3, "bbox": [1221.2388916015625, 6.906349182128906, 698.7611083984375, 974.9222030639648], "score": 0.07689044624567032}, {"image_id": 19, "category_id": 3, "bbox": [178.90072631835938, 11.81363296508789, 958.0703430175781, 984.2643699645996], "score": 0.06495741754770279}, {"image_id": 19, "category_id": 3, "bbox": [1105.939697265625, 0.0, 351.2701416015625, 639.305419921875], "score": 0.06210622936487198}, {"image_id": 19, "category_id": 3, "bbox": [1363.134765625, 230.79869079589844, 556.865234375, 849.2013092041016], "score": 0.05937302112579346}, {"image_id": 19, "category_id": 3, "bbox": [534.384521484375, 331.02789306640625, 977.738525390625, 748.9721069335938], "score": 0.055273693054914474}, {"image_id": 19, "category_id": 3, "bbox": [965.5433349609375, 0.0, 308.699951171875, 633.299072265625], "score": 0.05506341531872749}, {"image_id": 19, "category_id": 3, "bbox": [906.649658203125, 490.4180603027344, 414.161376953125, 562.4248352050781], "score": 0.05341381952166557}, {"image_id": 19, "category_id": 3, "bbox": [1169.9854736328125, 135.29786682128906, 385.6878662109375, 641.7587738037109], "score": 0.05341159924864769}, {"image_id": 19, "category_id": 3, "bbox": [871.9500732421875, 54.54948806762695, 332.63232421875, 594.874828338623], "score": 0.051953934133052826}, {"image_id": 19, "category_id": 3, "bbox": [309.0577392578125, 125.52936553955078, 381.44183349609375, 558.5551681518555], "score": 0.05152403935790062}, {"image_id": 19, "category_id": 3, "bbox": [248.37368774414062, 245.9715118408203, 362.0437316894531, 619.9862518310547], "score": 0.051376838237047195}, {"image_id": 19, "category_id": 3, "bbox": [141.2560272216797, 332.6877746582031, 1039.2582550048828, 747.3122253417969], "score": 0.05011116713285446}, {"image_id": 19, "category_id": 3, "bbox": [356.83734130859375, 0.0, 967.5060424804688, 748.9248046875], "score": 0.04369395226240158}, {"image_id": 19, "category_id": 3, "bbox": [823.4144897460938, 429.1307373046875, 431.06524658203125, 568.1822509765625], "score": 0.04016021266579628}, {"image_id": 19, "category_id": 3, "bbox": [10.906402587890625, 0.0, 936.0325012207031, 725.3652954101562], "score": 0.039944734424352646}, {"image_id": 19, "category_id": 3, "bbox": [300.4727783203125, 0.0, 412.12359619140625, 508.9482116699219], "score": 0.03989148512482643}, {"image_id": 19, "category_id": 3, "bbox": [811.5675659179688, 0.0, 383.35906982421875, 503.3565673828125], "score": 0.03696984797716141}] \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/coco_detection_main.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/coco_detection_main.yml new file mode 100644 index 000000000..69af912b5 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/coco_detection_main.yml @@ -0,0 +1,19 @@ +metric: COCO +num_classes: 3 + +TrainDataset: + !COCODataSet + image_dir: picodet_motorcycle/JPEGImages/ + anno_path: voc_train.json + dataset_dir: /home/aistudio/data/data128282/ + data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd'] + +EvalDataset: + !COCODataSet + image_dir: picodet_motorcycle/JPEGImages/ + anno_path: voc_test.json + dataset_dir: /home/aistudio/data/data128282/ + +TestDataset: + !ImageFolder + anno_path: voc_test.json diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/eval.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/eval.py new file mode 100644 index 000000000..67f5383ff --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/eval.py @@ -0,0 +1,144 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys + +# add python path of PadleDetection to sys.path +parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2))) +sys.path.insert(0, parent_path) + +# ignore warning log +import warnings +warnings.filterwarnings('ignore') + +import paddle + +from ppdet.core.workspace import load_config, merge_config +from ppdet.utils.check import check_gpu, check_npu, check_version, check_config +from ppdet.utils.cli import ArgsParser +from ppdet.engine import Trainer, init_parallel_env +from ppdet.metrics.coco_utils import json_eval_results +from ppdet.slim import build_slim_model + +from ppdet.utils.logger import setup_logger +logger = setup_logger('eval') + + +def parse_args(): + parser = ArgsParser() + parser.add_argument( + "--output_eval", + default=None, + type=str, + help="Evaluation directory, default is current directory.") + + parser.add_argument( + '--json_eval', + action='store_true', + default=False, + help='Whether to re eval with already exists bbox.json or mask.json') + + parser.add_argument( + "--slim_config", + default=None, + type=str, + help="Configuration file of slim method.") + + # TODO: bias should be unified + parser.add_argument( + "--bias", + action="store_true", + help="whether add bias or not while getting w and h") + + parser.add_argument( + "--classwise", + action="store_true", + help="whether per-category AP and draw P-R Curve or not.") + + parser.add_argument( + '--save_prediction_only', + action='store_true', + default=False, + help='Whether to save the evaluation results only') + + args = parser.parse_args() + return args + + +def run(FLAGS, cfg): + if FLAGS.json_eval: + logger.info( + "In json_eval mode, PaddleDetection will evaluate json files in " + "output_eval directly. And proposal.json, bbox.json and mask.json " + "will be detected by default.") + json_eval_results( + cfg.metric, + json_directory=FLAGS.output_eval, + dataset=cfg['EvalDataset']) + return + + # init parallel environment if nranks > 1 + init_parallel_env() + + # build trainer + trainer = Trainer(cfg, mode='eval') + + # load weights + trainer.load_weights(cfg.weights) + + # training + trainer.evaluate() + + +def main(): + FLAGS = parse_args() + cfg = load_config(FLAGS.config) + # TODO: bias should be unified + cfg['bias'] = 1 if FLAGS.bias else 0 + cfg['classwise'] = True if FLAGS.classwise else False + cfg['output_eval'] = FLAGS.output_eval + cfg['save_prediction_only'] = FLAGS.save_prediction_only + merge_config(FLAGS.opt) + + # disable npu in config by default + if 'use_npu' not in cfg: + cfg.use_npu = False + + if cfg.use_gpu: + place = paddle.set_device('gpu') + elif cfg.use_npu: + place = paddle.set_device('npu') + else: + place = paddle.set_device('cpu') + + if 'norm_type' in cfg and cfg['norm_type'] == 'sync_bn' and not cfg.use_gpu: + cfg['norm_type'] = 'bn' + + if FLAGS.slim_config: + cfg = build_slim_model(cfg, FLAGS.slim_config, mode='eval') + check_config(cfg) + check_gpu(cfg.use_gpu) + check_npu(cfg.use_npu) + check_version() + + run(FLAGS, cfg) + + +if __name__ == '__main__': + main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/export_model.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/export_model.py new file mode 100644 index 000000000..deac2ea12 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/export_model.py @@ -0,0 +1,115 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys + +# add python path of PadleDetection to sys.path +parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2))) +sys.path.insert(0, parent_path) + +# ignore warning log +import warnings +warnings.filterwarnings('ignore') + +import paddle + +from ppdet.core.workspace import load_config, merge_config +from ppdet.utils.check import check_gpu, check_version, check_config +from ppdet.utils.cli import ArgsParser +from ppdet.engine import Trainer +from ppdet.slim import build_slim_model + +from ppdet.utils.logger import setup_logger +logger = setup_logger('export_model') + + +def parse_args(): + parser = ArgsParser() + parser.add_argument( + "--output_dir", + type=str, + default="output_inference", + help="Directory for storing the output model files.") + parser.add_argument( + "--export_serving_model", + type=bool, + default=False, + help="Whether to export serving model or not.") + parser.add_argument( + "--slim_config", + default=None, + type=str, + help="Configuration file of slim method.") + args = parser.parse_args() + return args + + +def run(FLAGS, cfg): + # build detector + trainer = Trainer(cfg, mode='test') + + # load weights + if cfg.architecture in ['DeepSORT']: + if cfg.det_weights != 'None': + trainer.load_weights_sde(cfg.det_weights, cfg.reid_weights) + else: + trainer.load_weights_sde(None, cfg.reid_weights) + else: + trainer.load_weights(cfg.weights) + + # export model + trainer.export(FLAGS.output_dir) + + if FLAGS.export_serving_model: + from paddle_serving_client.io import inference_model_to_serving + model_name = os.path.splitext(os.path.split(cfg.filename)[-1])[0] + + inference_model_to_serving( + dirname="{}/{}".format(FLAGS.output_dir, model_name), + serving_server="{}/{}/serving_server".format(FLAGS.output_dir, + model_name), + serving_client="{}/{}/serving_client".format(FLAGS.output_dir, + model_name), + model_filename="model.pdmodel", + params_filename="model.pdiparams") + + +def main(): + paddle.set_device("cpu") + FLAGS = parse_args() + cfg = load_config(FLAGS.config) + # TODO: to be refined in the future + if 'norm_type' in cfg and cfg['norm_type'] == 'sync_bn': + FLAGS.opt['norm_type'] = 'bn' + merge_config(FLAGS.opt) + + if FLAGS.slim_config: + cfg = build_slim_model(cfg, FLAGS.slim_config, mode='test') + + # FIXME: Temporarily solve the priority problem of FLAGS.opt + merge_config(FLAGS.opt) + check_config(cfg) + check_gpu(cfg.use_gpu) + check_version() + + run(FLAGS, cfg) + + +if __name__ == '__main__': + main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/optimizer_300e.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/optimizer_300e.yml new file mode 100644 index 000000000..5a89bbbce --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/optimizer_300e.yml @@ -0,0 +1,18 @@ +epoch: 300 + +LearningRate: + base_lr: 0.4 + schedulers: + - !CosineDecay + max_epochs: 300 + - !LinearWarmup + start_factor: 0.1 + steps: 300 + +OptimizerBuilder: + optimizer: + momentum: 0.9 + type: Momentum + regularizer: + factor: 0.00004 + type: L2 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdopt b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdopt new file mode 100644 index 000000000..0869488a1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdopt differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdparams b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdparams new file mode 100644 index 000000000..dddcbaa91 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/best_model.pdparams differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdopt b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdopt new file mode 100644 index 000000000..0869488a1 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdopt differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdparams b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdparams new file mode 100644 index 000000000..dddcbaa91 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output/picodet_lcnet_1_5x_416_coco/model_final.pdparams differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/infer_cfg.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/infer_cfg.yml new file mode 100644 index 000000000..e29f9298f --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/infer_cfg.yml @@ -0,0 +1,118 @@ +mode: fluid +draw_threshold: 0.5 +metric: COCO +use_dynamic_shape: false +arch: PicoDet +min_subgraph_size: 3 +Preprocess: +- interp: 2 + keep_ratio: false + target_size: + - 640 + - 640 + type: Resize +- is_scale: true + mean: + - 0.485 + - 0.456 + - 0.406 + std: + - 0.229 + - 0.224 + - 0.225 + type: NormalizeImage +- type: Permute +- stride: 32 + type: PadStride +label_list: +- person +- bicycle +- car +- motorcycle +- airplane +- bus +- train +- truck +- boat +- traffic light +- fire hydrant +- stop sign +- parking meter +- bench +- bird +- cat +- dog +- horse +- sheep +- cow +- elephant +- bear +- zebra +- giraffe +- backpack +- umbrella +- handbag +- tie +- suitcase +- frisbee +- skis +- snowboard +- sports ball +- kite +- baseball bat +- baseball glove +- skateboard +- surfboard +- tennis racket +- bottle +- wine glass +- cup +- fork +- knife +- spoon +- bowl +- banana +- apple +- sandwich +- orange +- broccoli +- carrot +- hot dog +- pizza +- donut +- cake +- chair +- couch +- potted plant +- bed +- dining table +- toilet +- tv +- laptop +- mouse +- remote +- keyboard +- cell phone +- microwave +- oven +- toaster +- sink +- refrigerator +- book +- clock +- vase +- scissors +- teddy bear +- hair drier +- toothbrush +NMS: + keep_top_k: 100 + name: MultiClassNMS + nms_threshold: 0.5 + nms_top_k: 1000 + score_threshold: 0.3 +fpn_stride: +- 8 +- 16 +- 32 +- 64 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams new file mode 100644 index 000000000..6dcfada5f Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams.info b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams.info new file mode 100644 index 000000000..c505a92f6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdiparams.info differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdmodel b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdmodel new file mode 100644 index 000000000..dfcfe5115 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/model.pdmodel differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/serving_server/model.pdmodel b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/output_inference/picodet_lcnet_1_5x_416_coco/serving_server/model.pdmodel new file mode 100644 index 000000000..e69de29bb diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_640_reader.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_640_reader.yml new file mode 100644 index 000000000..a931f2a76 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_640_reader.yml @@ -0,0 +1,41 @@ +worker_num: 6 +TrainReader: + sample_transforms: + - Decode: {} + - RandomCrop: {} + - RandomFlip: {prob: 0.5} + - RandomDistort: {} + batch_transforms: + - BatchRandomResize: {target_size: [576, 608, 640, 672, 704], random_size: True, random_interp: True, keep_ratio: False} + - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]} + - Permute: {} + batch_size: 56 + shuffle: true + drop_last: true + collate_batch: false + + +EvalReader: + sample_transforms: + - Decode: {} + - Resize: {interp: 2, target_size: [640, 640], keep_ratio: False} + - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]} + - Permute: {} + batch_transforms: + - PadBatch: {pad_to_stride: 32} + batch_size: 8 + shuffle: false + + +TestReader: + inputs_def: + image_shape: [1, 3, 640, 640] + sample_transforms: + - Decode: {} + - Resize: {interp: 2, target_size: [640, 640], keep_ratio: False} + - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]} + - Permute: {} + batch_transforms: + - PadBatch: {pad_to_stride: 32} + batch_size: 1 + shuffle: false diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_esnet.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_esnet.yml new file mode 100644 index 000000000..aa099fca1 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_esnet.yml @@ -0,0 +1,55 @@ +architecture: PicoDet +pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ESNet_x1_0_pretrained.pdparams + +PicoDet: + backbone: ESNet + neck: CSPPAN + head: PicoHead + +ESNet: + scale: 1.0 + feature_maps: [4, 11, 14] + act: hard_swish + channel_ratio: [0.875, 0.5, 1.0, 0.625, 0.5, 0.75, 0.625, 0.625, 0.5, 0.625, 1.0, 0.625, 0.75] + +CSPPAN: + out_channels: 128 + use_depthwise: True + num_csp_blocks: 1 + num_features: 4 + +PicoHead: + conv_feat: + name: PicoFeat + feat_in: 128 + feat_out: 128 + num_convs: 4 + num_fpn_stride: 4 + norm_type: bn + share_cls_reg: True + fpn_stride: [8, 16, 32, 64] + feat_in_chan: 128 + prior_prob: 0.01 + reg_max: 7 + cell_offset: 0.5 + loss_class: + name: VarifocalLoss + use_sigmoid: True + iou_weighted: True + loss_weight: 1.0 + loss_dfl: + name: DistributionFocalLoss + loss_weight: 0.25 + loss_bbox: + name: GIoULoss + loss_weight: 2.0 + assigner: + name: SimOTAAssigner + candidate_topk: 10 + iou_weight: 6 + nms: + name: MultiClassNMS + nms_top_k: 1000 + keep_top_k: 100 + score_threshold: 0.025 + nms_threshold: 0.6 diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_lcnet_1_5x_416_coco.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_lcnet_1_5x_416_coco.yml new file mode 100644 index 000000000..302137468 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/picodet_lcnet_1_5x_416_coco.yml @@ -0,0 +1,36 @@ +_BASE_: [ + './coco_detection_main.yml', + './runtime.yml', + './picodet_esnet.yml', + './optimizer_300e.yml', + './picodet_640_reader.yml', +] + +pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/LCNet_x1_5_pretrained.pdparams +weights: output/picodet_lcnet_1_5x_416_coco/best_model +find_unused_parameters: True +use_ema: true +cycle_epoch: 40 +snapshot_epoch: 10 +epoch: 10 + +PicoDet: + backbone: LCNet + neck: CSPPAN + head: PicoHead + +LCNet: + scale: 1.5 + feature_maps: [3, 4, 5] + +TrainReader: + batch_size: 20 + +LearningRate: + base_lr: 0.1 + schedulers: + - !CosineDecay + max_epochs: 300 + - !LinearWarmup + start_factor: 0.1 + steps: 300 \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/requirements.txt b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/requirements.txt new file mode 100644 index 000000000..8b184e905 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/requirements.txt @@ -0,0 +1,17 @@ +tqdm +typeguard ; python_version >= '3.4' +visualdl>=2.1.0 ; python_version <= '3.7' +opencv-python +PyYAML +shapely +scipy +terminaltables +Cython +pycocotools +#xtcocotools==1.6 #only for crowdpose +setuptools>=42.0.0 +lap +sklearn +motmetrics +openpyxl +cython_bbox \ No newline at end of file diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/runtime.yml b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/runtime.yml new file mode 100644 index 000000000..c502ddabe --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/runtime.yml @@ -0,0 +1,5 @@ +use_gpu: true +log_iter: 20 +save_dir: output +snapshot_epoch: 1 +print_flops: false diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/train.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/train.py new file mode 100644 index 000000000..878aa60fa --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/train/train.py @@ -0,0 +1,171 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +from __future__ import absolute_import +from __future__ import division +from __future__ import print_function + +import os +import sys + +# add python path of PadleDetection to sys.path +parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 2))) +sys.path.insert(0, parent_path) + +# ignore warning log +import warnings +warnings.filterwarnings('ignore') + +import paddle + +from ppdet.core.workspace import load_config, merge_config +from ppdet.engine import Trainer, init_parallel_env, set_random_seed, init_fleet_env +from ppdet.slim import build_slim_model + +import ppdet.utils.cli as cli +import ppdet.utils.check as check +from ppdet.utils.logger import setup_logger +logger = setup_logger('train') + + +def parse_args(): + parser = cli.ArgsParser() + parser.add_argument( + "--eval", + action='store_true', + default=False, + help="Whether to perform evaluation in train") + parser.add_argument( + "-r", "--resume", default=None, help="weights path for resume") + parser.add_argument( + "--slim_config", + default=None, + type=str, + help="Configuration file of slim method.") + parser.add_argument( + "--enable_ce", + type=bool, + default=False, + help="If set True, enable continuous evaluation job." + "This flag is only used for internal test.") + parser.add_argument( + "--fp16", + action='store_true', + default=False, + help="Enable mixed precision training.") + parser.add_argument( + "--fleet", action='store_true', default=False, help="Use fleet or not") + parser.add_argument( + "--use_vdl", + type=bool, + default=False, + help="whether to record the data to VisualDL.") + parser.add_argument( + '--vdl_log_dir', + type=str, + default="vdl_log_dir/scalar", + help='VisualDL logging directory for scalar.') + parser.add_argument( + '--save_prediction_only', + action='store_true', + default=False, + help='Whether to save the evaluation results only') + parser.add_argument( + '--profiler_options', + type=str, + default=None, + help="The option of profiler, which should be in " + "format \"key1=value1;key2=value2;key3=value3\"." + "please see ppdet/utils/profiler.py for detail.") + parser.add_argument( + '--save_proposals', + action='store_true', + default=False, + help='Whether to save the train proposals') + parser.add_argument( + '--proposals_path', + type=str, + default="sniper/proposals.json", + help='Train proposals directory') + + args = parser.parse_args() + return args + + +def run(FLAGS, cfg): + # init fleet environment + if cfg.fleet: + init_fleet_env(cfg.get('find_unused_parameters', False)) + else: + # init parallel environment if nranks > 1 + init_parallel_env() + + if FLAGS.enable_ce: + set_random_seed(0) + + # build trainer + trainer = Trainer(cfg, mode='train') + + # load weights + if FLAGS.resume is not None: + trainer.resume_weights(FLAGS.resume) + elif 'pretrain_weights' in cfg and cfg.pretrain_weights: + trainer.load_weights(cfg.pretrain_weights) + + # training + trainer.train(FLAGS.eval) + + +def main(): + FLAGS = parse_args() + cfg = load_config(FLAGS.config) + cfg['fp16'] = FLAGS.fp16 + cfg['fleet'] = FLAGS.fleet + cfg['use_vdl'] = FLAGS.use_vdl + cfg['vdl_log_dir'] = FLAGS.vdl_log_dir + cfg['save_prediction_only'] = FLAGS.save_prediction_only + cfg['profiler_options'] = FLAGS.profiler_options + cfg['save_proposals'] = FLAGS.save_proposals + cfg['proposals_path'] = FLAGS.proposals_path + merge_config(FLAGS.opt) + + # disable npu in config by default + if 'use_npu' not in cfg: + cfg.use_npu = False + + if cfg.use_gpu: + place = paddle.set_device('gpu') + elif cfg.use_npu: + place = paddle.set_device('npu') + else: + place = paddle.set_device('cpu') + + if 'norm_type' in cfg and cfg['norm_type'] == 'sync_bn' and not cfg.use_gpu: + cfg['norm_type'] = 'bn' + + if FLAGS.slim_config: + cfg = build_slim_model(cfg, FLAGS.slim_config) + + # FIXME: Temporarily solve the priority problem of FLAGS.opt + merge_config(FLAGS.opt) + check.check_config(cfg) + check.check_gpu(cfg.use_gpu) + check.check_npu(cfg.use_npu) + check.check_version() + + run(FLAGS, cfg) + + +if __name__ == "__main__": + main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/config.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/config.cpython-37.pyc new file mode 100644 index 000000000..34c5168eb Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/config.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/get_image_list.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/get_image_list.cpython-37.pyc new file mode 100644 index 000000000..c2ca48112 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/get_image_list.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/logger.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/logger.cpython-37.pyc new file mode 100644 index 000000000..fdde1c9da Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/logger.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/predictor.cpython-37.pyc b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/predictor.cpython-37.pyc new file mode 100644 index 000000000..a54b304db Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/__pycache__/predictor.cpython-37.pyc differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/config.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/config.py new file mode 100644 index 000000000..eb7914806 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/config.py @@ -0,0 +1,197 @@ +# copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import copy +import argparse +import yaml + +from utils import logger + +__all__ = ['get_config'] + + +class AttrDict(dict): + def __getattr__(self, key): + return self[key] + + def __setattr__(self, key, value): + if key in self.__dict__: + self.__dict__[key] = value + else: + self[key] = value + + def __deepcopy__(self, content): + return copy.deepcopy(dict(self)) + + +def create_attr_dict(yaml_config): + from ast import literal_eval + for key, value in yaml_config.items(): + if type(value) is dict: + yaml_config[key] = value = AttrDict(value) + if isinstance(value, str): + try: + value = literal_eval(value) + except BaseException: + pass + if isinstance(value, AttrDict): + create_attr_dict(yaml_config[key]) + else: + yaml_config[key] = value + + +def parse_config(cfg_file): + """Load a config file into AttrDict""" + with open(cfg_file, 'r') as fopen: + yaml_config = AttrDict(yaml.load(fopen, Loader=yaml.SafeLoader)) + create_attr_dict(yaml_config) + return yaml_config + + +def print_dict(d, delimiter=0): + """ + Recursively visualize a dict and + indenting acrrording by the relationship of keys. + """ + placeholder = "-" * 60 + for k, v in sorted(d.items()): + if isinstance(v, dict): + logger.info("{}{} : ".format(delimiter * " ", + logger.coloring(k, "HEADER"))) + print_dict(v, delimiter + 4) + elif isinstance(v, list) and len(v) >= 1 and isinstance(v[0], dict): + logger.info("{}{} : ".format(delimiter * " ", + logger.coloring(str(k), "HEADER"))) + for value in v: + print_dict(value, delimiter + 4) + else: + logger.info("{}{} : {}".format(delimiter * " ", + logger.coloring(k, "HEADER"), + logger.coloring(v, "OKGREEN"))) + if k.isupper(): + logger.info(placeholder) + + +def print_config(config): + """ + visualize configs + Arguments: + config: configs + """ + logger.advertise() + print_dict(config) + + +def override(dl, ks, v): + """ + Recursively replace dict of list + Args: + dl(dict or list): dict or list to be replaced + ks(list): list of keys + v(str): value to be replaced + """ + + def str2num(v): + try: + return eval(v) + except Exception: + return v + + assert isinstance(dl, (list, dict)), ("{} should be a list or a dict") + assert len(ks) > 0, ('lenght of keys should larger than 0') + if isinstance(dl, list): + k = str2num(ks[0]) + if len(ks) == 1: + assert k < len(dl), ('index({}) out of range({})'.format(k, dl)) + dl[k] = str2num(v) + else: + override(dl[k], ks[1:], v) + else: + if len(ks) == 1: + # assert ks[0] in dl, ('{} is not exist in {}'.format(ks[0], dl)) + if not ks[0] in dl: + logger.warning('A new filed ({}) detected!'.format(ks[0], dl)) + dl[ks[0]] = str2num(v) + else: + override(dl[ks[0]], ks[1:], v) + + +def override_config(config, options=None): + """ + Recursively override the config + Args: + config(dict): dict to be replaced + options(list): list of pairs(key0.key1.idx.key2=value) + such as: [ + 'topk=2', + 'VALID.transforms.1.ResizeImage.resize_short=300' + ] + Returns: + config(dict): replaced config + """ + if options is not None: + for opt in options: + assert isinstance(opt, str), ( + "option({}) should be a str".format(opt)) + assert "=" in opt, ( + "option({}) should contain a =" + "to distinguish between key and value".format(opt)) + pair = opt.split('=') + assert len(pair) == 2, ("there can be only a = in the option") + key, value = pair + keys = key.split('.') + override(config, keys, value) + return config + + +def get_config(fname, overrides=None, show=True): + """ + Read config from file + """ + assert os.path.exists(fname), ( + 'config file({}) is not exist'.format(fname)) + config = parse_config(fname) + override_config(config, overrides) + if show: + print_config(config) + # check_config(config) + return config + + +def parser(): + parser = argparse.ArgumentParser("generic-image-rec train script") + parser.add_argument( + '-c', + '--config', + type=str, + default='configs/config.yaml', + help='config file path') + parser.add_argument( + '-o', + '--override', + action='append', + default=[], + help='config options to be overridden') + parser.add_argument( + '-v', + '--verbose', + action='store_true', + help='wheather print the config info') + return parser + + +def parse_args(): + args = parser().parse_args() + return args diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/get_image_list.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/get_image_list.py new file mode 100644 index 000000000..6f10935ad --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/get_image_list.py @@ -0,0 +1,49 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import os +import argparse +import base64 +import numpy as np + + +def get_image_list(img_file): + imgs_lists = [] + if img_file is None or not os.path.exists(img_file): + raise Exception("not found any img file in {}".format(img_file)) + + img_end = ['jpg', 'png', 'jpeg', 'JPEG', 'JPG', 'bmp'] + if os.path.isfile(img_file) and img_file.split('.')[-1] in img_end: + imgs_lists.append(img_file) + elif os.path.isdir(img_file): + for single_file in os.listdir(img_file): + if single_file.split('.')[-1] in img_end: + imgs_lists.append(os.path.join(img_file, single_file)) + if len(imgs_lists) == 0: + raise Exception("not found any img file in {}".format(img_file)) + imgs_lists = sorted(imgs_lists) + return imgs_lists + + +def get_image_list_from_label_file(image_path, label_file_path): + imgs_lists = [] + gt_labels = [] + with open(label_file_path, "r") as fin: + lines = fin.readlines() + for line in lines: + image_name, label = line.strip("\n").split() + label = int(label) + imgs_lists.append(os.path.join(image_path, image_name)) + gt_labels.append(int(label)) + return imgs_lists, gt_labels diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/logger.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/logger.py new file mode 100644 index 000000000..ece852624 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/logger.py @@ -0,0 +1,120 @@ +# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import logging +import os +import datetime + +logging.basicConfig( + level=logging.INFO, + format="%(asctime)s %(levelname)s: %(message)s", + datefmt="%Y-%m-%d %H:%M:%S") + + +def time_zone(sec, fmt): + real_time = datetime.datetime.now() + return real_time.timetuple() + + +logging.Formatter.converter = time_zone +_logger = logging.getLogger(__name__) + +Color = { + 'RED': '\033[31m', + 'HEADER': '\033[35m', # deep purple + 'PURPLE': '\033[95m', # purple + 'OKBLUE': '\033[94m', + 'OKGREEN': '\033[92m', + 'WARNING': '\033[93m', + 'FAIL': '\033[91m', + 'ENDC': '\033[0m' +} + + +def coloring(message, color="OKGREEN"): + assert color in Color.keys() + if os.environ.get('PADDLECLAS_COLORING', False): + return Color[color] + str(message) + Color["ENDC"] + else: + return message + + +def anti_fleet(log): + """ + logs will print multi-times when calling Fleet API. + Only display single log and ignore the others. + """ + + def wrapper(fmt, *args): + if int(os.getenv("PADDLE_TRAINER_ID", 0)) == 0: + log(fmt, *args) + + return wrapper + + +@anti_fleet +def info(fmt, *args): + _logger.info(fmt, *args) + + +@anti_fleet +def warning(fmt, *args): + _logger.warning(coloring(fmt, "RED"), *args) + + +@anti_fleet +def error(fmt, *args): + _logger.error(coloring(fmt, "FAIL"), *args) + + +def scaler(name, value, step, writer): + """ + This function will draw a scalar curve generated by the visualdl. + Usage: Install visualdl: pip3 install visualdl==2.0.0b4 + and then: + visualdl --logdir ./scalar --host 0.0.0.0 --port 8830 + to preview loss corve in real time. + """ + writer.add_scalar(tag=name, step=step, value=value) + + +def advertise(): + """ + Show the advertising message like the following: + + =========================================================== + == PaddleClas is powered by PaddlePaddle ! == + =========================================================== + == == + == For more info please go to the following website. == + == == + == https://github.com/PaddlePaddle/PaddleClas == + =========================================================== + + """ + copyright = "PaddleClas is powered by PaddlePaddle !" + ad = "For more info please go to the following website." + website = "https://github.com/PaddlePaddle/PaddleClas" + AD_LEN = 6 + len(max([copyright, ad, website], key=len)) + + info( + coloring("\n{0}\n{1}\n{2}\n{3}\n{4}\n{5}\n{6}\n{7}\n".format( + "=" * (AD_LEN + 4), + "=={}==".format(copyright.center(AD_LEN)), + "=" * (AD_LEN + 4), + "=={}==".format(' ' * AD_LEN), + "=={}==".format(ad.center(AD_LEN)), + "=={}==".format(' ' * AD_LEN), + "=={}==".format(website.center(AD_LEN)), + "=" * (AD_LEN + 4), ), "RED")) diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/predictor.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/predictor.py new file mode 100644 index 000000000..11f153071 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/utils/predictor.py @@ -0,0 +1,70 @@ +# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +import os +import argparse +import base64 +import shutil +import cv2 +import numpy as np + +from paddle.inference import Config +from paddle.inference import create_predictor + + +class Predictor(object): + def __init__(self, args, inference_model_dir=None): + # HALF precission predict only work when using tensorrt + if args.use_fp16 is True: + assert args.use_tensorrt is True + self.args = args + self.paddle_predictor, self.config = self.create_paddle_predictor( + args, inference_model_dir) + + def predict(self, image): + raise NotImplementedError + + def create_paddle_predictor(self, args, inference_model_dir=None): + if inference_model_dir is None: + inference_model_dir = args.inference_model_dir + params_file = os.path.join(inference_model_dir, "inference.pdiparams") + model_file = os.path.join(inference_model_dir, "inference.pdmodel") + config = Config(model_file, params_file) + + if args.use_gpu: + config.enable_use_gpu(args.gpu_mem, 0) + else: + config.disable_gpu() + if args.enable_mkldnn: + # cache 10 different shapes for mkldnn to avoid memory leak + config.set_mkldnn_cache_capacity(10) + config.enable_mkldnn() + config.set_cpu_math_library_num_threads(args.cpu_num_threads) + + if args.enable_profile: + config.enable_profile() + config.disable_glog_info() + config.switch_ir_optim(args.ir_optim) # default true + if args.use_tensorrt: + config.enable_tensorrt_engine( + precision_mode=Config.Precision.Half + if args.use_fp16 else Config.Precision.Float32, + max_batch_size=args.batch_size, + min_subgraph_size=30) + + config.enable_memory_optim() + # use zero copy + config.switch_use_feed_fetch_ops(False) + predictor = create_predictor(config) + + return predictor, config diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/x2coco.py b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/x2coco.py new file mode 100644 index 000000000..2d0e64e64 --- /dev/null +++ b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/code/x2coco.py @@ -0,0 +1,450 @@ +#!/usr/bin/env python +# coding: utf-8 +# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + +import argparse +import glob +import json +import os +import os.path as osp +import shutil +import xml.etree.ElementTree as ET +from tqdm import tqdm + +import numpy as np +import PIL.ImageDraw + +label_to_num = {} +categories_list = [] +labels_list = [] + + +class MyEncoder(json.JSONEncoder): + def default(self, obj): + if isinstance(obj, np.integer): + return int(obj) + elif isinstance(obj, np.floating): + return float(obj) + elif isinstance(obj, np.ndarray): + return obj.tolist() + else: + return super(MyEncoder, self).default(obj) + + +def images_labelme(data, num): + image = {} + image['height'] = data['imageHeight'] + image['width'] = data['imageWidth'] + image['id'] = num + 1 + if '\\' in data['imagePath']: + image['file_name'] = data['imagePath'].split('\\')[-1] + else: + image['file_name'] = data['imagePath'].split('/')[-1] + image['file_name'] = image['file_name'].rstrip() + return image + + +def images_cityscape(data, num, img_file): + image = {} + image['height'] = data['imgHeight'] + image['width'] = data['imgWidth'] + image['id'] = num + 1 + image['file_name'] = img_file + image['file_name'] = image['file_name'].rstrip() + return image + + +def categories(label, labels_list): + category = {} + category['supercategory'] = 'component' + category['id'] = len(labels_list) + 1 + category['name'] = label + return category + + +def annotations_rectangle(points, label, image_num, object_num, label_to_num): + annotation = {} + seg_points = np.asarray(points).copy() + seg_points[1, :] = np.asarray(points)[2, :] + seg_points[2, :] = np.asarray(points)[1, :] + annotation['segmentation'] = [list(seg_points.flatten())] + annotation['iscrowd'] = 0 + annotation['image_id'] = image_num + 1 + annotation['bbox'] = list( + map(float, [ + points[0][0], points[0][1], points[1][0] - points[0][0], points[1][ + 1] - points[0][1] + ])) + annotation['area'] = annotation['bbox'][2] * annotation['bbox'][3] + annotation['category_id'] = label_to_num[label] + annotation['id'] = object_num + 1 + return annotation + + +def annotations_polygon(height, width, points, label, image_num, object_num, + label_to_num): + annotation = {} + annotation['segmentation'] = [list(np.asarray(points).flatten())] + annotation['iscrowd'] = 0 + annotation['image_id'] = image_num + 1 + annotation['bbox'] = list(map(float, get_bbox(height, width, points))) + annotation['area'] = annotation['bbox'][2] * annotation['bbox'][3] + annotation['category_id'] = label_to_num[label] + annotation['id'] = object_num + 1 + return annotation + + +def get_bbox(height, width, points): + polygons = points + mask = np.zeros([height, width], dtype=np.uint8) + mask = PIL.Image.fromarray(mask) + xy = list(map(tuple, polygons)) + PIL.ImageDraw.Draw(mask).polygon(xy=xy, outline=1, fill=1) + mask = np.array(mask, dtype=bool) + index = np.argwhere(mask == 1) + rows = index[:, 0] + clos = index[:, 1] + left_top_r = np.min(rows) + left_top_c = np.min(clos) + right_bottom_r = np.max(rows) + right_bottom_c = np.max(clos) + return [ + left_top_c, left_top_r, right_bottom_c - left_top_c, + right_bottom_r - left_top_r + ] + + +def deal_json(ds_type, img_path, json_path): + data_coco = {} + images_list = [] + annotations_list = [] + image_num = -1 + object_num = -1 + for img_file in os.listdir(img_path): + img_label = os.path.splitext(img_file)[0] + if img_file.split('.')[ + -1] not in ['bmp', 'jpg', 'jpeg', 'png', 'JPEG', 'JPG', 'PNG']: + continue + label_file = osp.join(json_path, img_label + '.json') + print('Generating dataset from:', label_file) + image_num = image_num + 1 + with open(label_file) as f: + data = json.load(f) + if ds_type == 'labelme': + images_list.append(images_labelme(data, image_num)) + elif ds_type == 'cityscape': + images_list.append(images_cityscape(data, image_num, img_file)) + if ds_type == 'labelme': + for shapes in data['shapes']: + object_num = object_num + 1 + label = shapes['label'] + if label not in labels_list: + categories_list.append(categories(label, labels_list)) + labels_list.append(label) + label_to_num[label] = len(labels_list) + p_type = shapes['shape_type'] + if p_type == 'polygon': + points = shapes['points'] + annotations_list.append( + annotations_polygon(data['imageHeight'], data[ + 'imageWidth'], points, label, image_num, + object_num, label_to_num)) + + if p_type == 'rectangle': + (x1, y1), (x2, y2) = shapes['points'] + x1, x2 = sorted([x1, x2]) + y1, y2 = sorted([y1, y2]) + points = [[x1, y1], [x2, y2], [x1, y2], [x2, y1]] + annotations_list.append( + annotations_rectangle(points, label, image_num, + object_num, label_to_num)) + elif ds_type == 'cityscape': + for shapes in data['objects']: + object_num = object_num + 1 + label = shapes['label'] + if label not in labels_list: + categories_list.append(categories(label, labels_list)) + labels_list.append(label) + label_to_num[label] = len(labels_list) + points = shapes['polygon'] + annotations_list.append( + annotations_polygon(data['imgHeight'], data[ + 'imgWidth'], points, label, image_num, object_num, + label_to_num)) + data_coco['images'] = images_list + data_coco['categories'] = categories_list + data_coco['annotations'] = annotations_list + return data_coco + + +def voc_get_label_anno(ann_dir_path, ann_ids_path, labels_path): + with open(labels_path, 'r') as f: + labels_str = f.read().split() + labels_ids = list(range(1, len(labels_str) + 1)) + + with open(ann_ids_path, 'r') as f: + ann_ids = [lin.strip().split(' ')[-1] for lin in f.readlines()] + + ann_paths = [] + for aid in ann_ids: + if aid.endswith('xml'): + ann_path = os.path.join(ann_dir_path, aid) + else: + ann_path = os.path.join(ann_dir_path, aid + '.xml') + ann_paths.append(ann_path) + + return dict(zip(labels_str, labels_ids)), ann_paths + + +def voc_get_image_info(annotation_root, im_id): + filename = annotation_root.findtext('filename') + assert filename is not None + img_name = os.path.basename(filename) + + size = annotation_root.find('size') + width = float(size.findtext('width')) + height = float(size.findtext('height')) + + image_info = { + 'file_name': filename, + 'height': height, + 'width': width, + 'id': im_id + } + return image_info + + +def voc_get_coco_annotation(obj, label2id): + label = obj.findtext('name') + assert label in label2id, "label is not in label2id." + category_id = label2id[label] + bndbox = obj.find('bndbox') + xmin = float(bndbox.findtext('xmin')) + ymin = float(bndbox.findtext('ymin')) + xmax = float(bndbox.findtext('xmax')) + ymax = float(bndbox.findtext('ymax')) + assert xmax > xmin and ymax > ymin, "Box size error." + o_width = xmax - xmin + o_height = ymax - ymin + anno = { + 'area': o_width * o_height, + 'iscrowd': 0, + 'bbox': [xmin, ymin, o_width, o_height], + 'category_id': category_id, + 'ignore': 0, + } + return anno + + +def voc_xmls_to_cocojson(annotation_paths, label2id, output_dir, output_file): + output_json_dict = { + "images": [], + "type": "instances", + "annotations": [], + "categories": [] + } + bnd_id = 1 # bounding box start id + im_id = 0 + print('Start converting !') + for a_path in tqdm(annotation_paths): + # Read annotation xml + ann_tree = ET.parse(a_path) + ann_root = ann_tree.getroot() + + img_info = voc_get_image_info(ann_root, im_id) + output_json_dict['images'].append(img_info) + print(a_path) + for obj in ann_root.findall('object'): + ann = voc_get_coco_annotation(obj=obj, label2id=label2id) + ann.update({'image_id': im_id, 'id': bnd_id}) + output_json_dict['annotations'].append(ann) + bnd_id = bnd_id + 1 + im_id += 1 + + for label, label_id in label2id.items(): + category_info = {'supercategory': 'none', 'id': label_id, 'name': label} + output_json_dict['categories'].append(category_info) + output_file = os.path.join(output_dir, output_file) + with open(output_file, 'w') as f: + output_json = json.dumps(output_json_dict) + f.write(output_json) + + +def main(): + parser = argparse.ArgumentParser( + formatter_class=argparse.ArgumentDefaultsHelpFormatter) + parser.add_argument( + '--dataset_type', + help='the type of dataset, can be `voc`, `labelme` or `cityscape`') + parser.add_argument('--json_input_dir', help='input annotated directory') + parser.add_argument('--image_input_dir', help='image directory') + parser.add_argument( + '--output_dir', help='output dataset directory', default='./') + parser.add_argument( + '--train_proportion', + help='the proportion of train dataset', + type=float, + default=1.0) + parser.add_argument( + '--val_proportion', + help='the proportion of validation dataset', + type=float, + default=0.0) + parser.add_argument( + '--test_proportion', + help='the proportion of test dataset', + type=float, + default=0.0) + parser.add_argument( + '--voc_anno_dir', + help='In Voc format dataset, path to annotation files directory.', + type=str, + default=None) + parser.add_argument( + '--voc_anno_list', + help='In Voc format dataset, path to annotation files ids list.', + type=str, + default=None) + parser.add_argument( + '--voc_label_list', + help='In Voc format dataset, path to label list. The content of each line is a category.', + type=str, + default=None) + parser.add_argument( + '--voc_out_name', + type=str, + default='voc.json', + help='In Voc format dataset, path to output json file') + args = parser.parse_args() + try: + assert args.dataset_type in ['voc', 'labelme', 'cityscape'] + except AssertionError as e: + print( + 'Now only support the voc, cityscape dataset and labelme dataset!!') + os._exit(0) + + if args.dataset_type == 'voc': + assert args.voc_anno_dir and args.voc_anno_list and args.voc_label_list + label2id, ann_paths = voc_get_label_anno( + args.voc_anno_dir, args.voc_anno_list, args.voc_label_list) + voc_xmls_to_cocojson( + annotation_paths=ann_paths, + label2id=label2id, + output_dir=args.output_dir, + output_file=args.voc_out_name) + else: + try: + assert os.path.exists(args.json_input_dir) + except AssertionError as e: + print('The json folder does not exist!') + os._exit(0) + try: + assert os.path.exists(args.image_input_dir) + except AssertionError as e: + print('The image folder does not exist!') + os._exit(0) + try: + assert abs(args.train_proportion + args.val_proportion \ + + args.test_proportion - 1.0) < 1e-5 + except AssertionError as e: + print( + 'The sum of pqoportion of training, validation and test datase must be 1!' + ) + os._exit(0) + + # Allocate the dataset. + total_num = len(glob.glob(osp.join(args.json_input_dir, '*.json'))) + if args.train_proportion != 0: + train_num = int(total_num * args.train_proportion) + out_dir = args.output_dir + '/train' + if not os.path.exists(out_dir): + os.makedirs(out_dir) + else: + train_num = 0 + if args.val_proportion == 0.0: + val_num = 0 + test_num = total_num - train_num + out_dir = args.output_dir + '/test' + if args.test_proportion != 0.0 and not os.path.exists(out_dir): + os.makedirs(out_dir) + else: + val_num = int(total_num * args.val_proportion) + test_num = total_num - train_num - val_num + val_out_dir = args.output_dir + '/val' + if not os.path.exists(val_out_dir): + os.makedirs(val_out_dir) + test_out_dir = args.output_dir + '/test' + if args.test_proportion != 0.0 and not os.path.exists(test_out_dir): + os.makedirs(test_out_dir) + count = 1 + for img_name in os.listdir(args.image_input_dir): + if count <= train_num: + if osp.exists(args.output_dir + '/train/'): + shutil.copyfile( + osp.join(args.image_input_dir, img_name), + osp.join(args.output_dir + '/train/', img_name)) + else: + if count <= train_num + val_num: + if osp.exists(args.output_dir + '/val/'): + shutil.copyfile( + osp.join(args.image_input_dir, img_name), + osp.join(args.output_dir + '/val/', img_name)) + else: + if osp.exists(args.output_dir + '/test/'): + shutil.copyfile( + osp.join(args.image_input_dir, img_name), + osp.join(args.output_dir + '/test/', img_name)) + count = count + 1 + + # Deal with the json files. + if not os.path.exists(args.output_dir + '/annotations'): + os.makedirs(args.output_dir + '/annotations') + if args.train_proportion != 0: + train_data_coco = deal_json(args.dataset_type, + args.output_dir + '/train', + args.json_input_dir) + train_json_path = osp.join(args.output_dir + '/annotations', + 'instance_train.json') + json.dump( + train_data_coco, + open(train_json_path, 'w'), + indent=4, + cls=MyEncoder) + if args.val_proportion != 0: + val_data_coco = deal_json(args.dataset_type, + args.output_dir + '/val', + args.json_input_dir) + val_json_path = osp.join(args.output_dir + '/annotations', + 'instance_val.json') + json.dump( + val_data_coco, + open(val_json_path, 'w'), + indent=4, + cls=MyEncoder) + if args.test_proportion != 0: + test_data_coco = deal_json(args.dataset_type, + args.output_dir + '/test', + args.json_input_dir) + test_json_path = osp.join(args.output_dir + '/annotations', + 'instance_test.json') + json.dump( + test_data_coco, + open(test_json_path, 'w'), + indent=4, + cls=MyEncoder) + + +if __name__ == '__main__': + main() diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/index_infer_result.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/index_infer_result.png new file mode 100644 index 000000000..a419074b7 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/index_infer_result.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/infer_result.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/infer_result.png new file mode 100644 index 000000000..bf299fd11 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/infer_result.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/label_img.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/label_img.png new file mode 100644 index 000000000..46aef3f11 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/label_img.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/result_5.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/result_5.png new file mode 100644 index 000000000..511156661 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/result_5.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/xml_content.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/xml_content.png new file mode 100644 index 000000000..f4db8d49c Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/docs/images/xml_content.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_1.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_1.png new file mode 100644 index 000000000..ebce35597 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_1.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_2.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_2.png new file mode 100644 index 000000000..466689918 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_2.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_3.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_3.png new file mode 100644 index 000000000..6ada31633 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_3.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_4.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_4.png new file mode 100644 index 000000000..ad7e495c6 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_4.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_5.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_5.png new file mode 100644 index 000000000..8585dd300 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/pic_5.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_1.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_1.png new file mode 100644 index 000000000..fb1b850f4 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_1.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_2.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_2.png new file mode 100644 index 000000000..87c36dd61 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_2.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_3.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_3.png new file mode 100644 index 000000000..050aebd3a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_3.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_4.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_4.png new file mode 100644 index 000000000..a0a34bf6a Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_4.png differ diff --git a/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_5.png b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_5.png new file mode 100644 index 000000000..511156661 Binary files /dev/null and b/Paddle_Industry_Practice_Sample_Library/Electromobile_In_Elevator_Detection/result/result_5.png differ