Skip to content

Latest commit

 

History

History
44 lines (36 loc) · 2.86 KB

README.md

File metadata and controls

44 lines (36 loc) · 2.86 KB

DL-NumRecognizer

A deep learning-based project for real-time digit recognition and visualization using PyTorch and OpenCV. This application trains a neural network to recognize handwritten digits with simple live visualization of predictions, allowing users to observe model performance as it improves over training epochs.

Features

  • Real-Time Training Visualization: Watch the model's predictions live as it learns to classify handwritten digits. Visualization can be enabled or disabled with a command-line flag.
  • Interactive Digit Recognition: Test the trained model with your own input for real-time feedback on digit classification accuracy.
  • Customizable Training: Easily adjust training parameters, including batch size, learning rate, and epoch count, to experiment with different setups.

Screenshots


Real-time visualization of model predictions during training.


Test the model's digit recognition capabilities interactively.

How to Run

  1. Install Dependencies: Install the required libraries:

    pip install torch torchvision opencv-python matplotlib

    Note: If you have an NVIDIA GPU, installing CUDA is recommended for faster training. To enable CUDA, install the CUDA toolkit and cuDNN (optional, but recommended). PyTorch will automatically detect CUDA if it is properly configured. You can find more instructions in the PyTorch installation guide. Find the suitable PyTorch version for your CUDA version, and install it using a command like the following:

    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu117

  2. Train the Model: You can either train your own model or download a pre-trained model. Visualization is optional and can be enabled by adding the --visualize flag:

    python train.py --visualize

  3. Evaluate the Model: After training (or downloading) the model, you can interact with it to classify handwritten digits. Run the following command:

    python predict.py
    • In the drawing window:
      • Press p or Enter to classify the drawn digit. The predicted digit, along with the model's confidence, will be displayed in the console window.
      • Press q or Esc to close the drawing window and exit the program.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributions

Contributions are welcome! Please open an issue or submit a pull request if you have any suggestions or improvements.