Image Classifier Toolkit

A Django-based web application with multiple image classification projects including flower, animal, fruit, emotion, vehicle, and digit classifiers. Each classifier has its own model and training pipeline using TensorFlow/Keras.

Project Structure

image_classifier_toolkit/
├── common/                    # Common utilities, e.g., training helper classes
├── config/                    # Project configuration files (settings, urls, etc.)
├── datasets/                  # Dataset folders for different classifiers
│ ├── flowers/
│ │ ├── daisy/
│ │ ├── dandelion/
│ │ ├── rose/
│ │ ├── sunflower/
│ │ └── tulip/
│ ├── animals/
│ ├── fruits/
│ ├── emotions/
│ ├── vehicles/
│ └── digits/
├── models/                    # Saved trained model files (.h5)
│ ├── flower_model.h5
│ ├── animal_model.h5
│ ├── fruit_model.h5
│ ├── emotion_model.h5
│ ├── vehicle_model.h5
│ └── digit_model.h5
├── media/                     # Uploaded media files (images for prediction)
├── flower_classifier/         # Django app for flower classification
├── animal_classifier/         # Django app for animal classification
├── fruit_classifier/          # Django app for fruit classification
├── emotion_classifier/        # Django app for emotion classification
├── vehicle_classifier/        # Django app for vehicle classification
├── digit_classifier/          # Django app for digit classification
├── manage.py                  # Django management script
└── README.md

Dataset Downloads

You need to download datasets for each classifier and place them inside the datasets folder as shown above.

Installation & Setup

1. Clone the repository:

   git clone https://github.com/maruf-pfc/image_classifier_toolkit.git
   cd image_classifier_toolkit

2. Create and activate a Python virtual environment:

   python3 -m venv env
   source env/bin/activate   # Linux/macOS
   # .\env\Scripts\activate  # Windows

3. Install dependencies:

   pip3 install -r requirements.txt

4. Make sure you have TensorFlow installed and compatible with your environment.

Training Models

Each classifier has a custom management command to train its model.

Example for Flower Classifier:

python3 manage.py train_flower_model

Similarly, run commands for other classifiers, e.g.:

python3 manage.py train_animal_model
python3 manage.py train_fruit_model
python3 manage.py train_human_emotion_model
python3 manage.py train_vehicle_model
python3 manage.py train_digit_model

Note:

• Ensure the corresponding dataset is available inside datasets/<classifier_name>/ folder before training.
• Models will be saved in the models/ directory.

Running the Project

Start the Django development server:

python3 manage.py runserver

Open your browser and visit: http://127.0.0.1:8000/

The home page lists all available classifiers with links to their prediction pages.

Using the Web App

• On the classifier page (e.g., Flower Classifier), upload an image.
• The app will show the uploaded image, run prediction using the trained model, and display the predicted class with confidence.
• You can navigate back to home anytime.

Flower Prediction Demo

Animal Prediction Demo

Notes

• Uploaded images are stored in the media/ folder.
• Static files should be collected or served correctly in production.
• For large datasets or training on GPU, consider using Google Colab or a powerful machine.
• Keep your dataset folders organized to avoid flow_from_directory errors during training.

Contributing

Contributions are welcome! Feel free to submit issues or pull requests.

License

MIT License © Md. Maruf Sarker

Contact

If you have questions or want to collaborate, reach out:

• Email: mdmarufsarker.mms@gmail.com
• GitHub: https://github.com/maruf-pfc
• Twitter: @md_marufsarker