This is a PYQT-based demonstration. If you wish to use the UniSPAC tool directly, we recommend using napari-UniSPAC. To reproduce the training process, please refer to the the training folder.
It is recommended to deploy the software on a Linux system. Pre-install PyQt5 (Qt) and PyTorch. Devices that support cuda allow for smoother software usage.
Set up the software environment:
conda create -n UniSPAC python=3.9
conda activate UniSPAC
git clone https://github.com/ddd9898/UniSPAC.git
cd UniSPAC
pip install -r requirements.txtDownload test data and checkpoints：
bash ./download.shThe total files after data and model decompression take up 9.3GB of storage, so please make sure you have enough capacity. See the downloaded model weights in the checkpoints folder and the Hemi-Brain-ROI-1 test data in the data folder.
Finally, launch the software:
python demo.pyBrief tutorial: Click the left mouse button to add a positive point prompt, and the right mouse button to add a negative point prompt. Press Q to undo the previous point prompt, press E to clear all prompts.
If you want to apply UniSPAC to your own data, the napari plugin for UniSPAC might come in handy. Assuming you are a veteran napari user, installing napari-UniSPAC with the following command is sufficient.
pip install napari-UniSPACThe installation should take a few minutes, depending on your network conditions.
Feel free to contact djt20@mails.tsinghua.edu.cn if you have issues for any questions.