Handwritten Digit Classifier

A modern web application for classifying handwritten digits using a Convolutional Neural Network (CNN). This project provides both a user-friendly web interface and a REST API for digit classification.

Features

• Modern Web Interface

  • Responsive design with Bootstrap 5
  • Drag and drop image upload
  • Real-time image preview
  • Animated results display
  • Mobile-friendly interface

• Advanced Model

  • CNN architecture with Batch Normalization
  • Data augmentation for better generalization
  • Early stopping and learning rate reduction
  • High accuracy on MNIST dataset

• API Support

  • RESTful API endpoint for predictions
  • JSON response format
  • Error handling and status codes

Requirements

• Python 3.8+
• TensorFlow 2.x
• Flask
• NumPy
• Pillow
• scikit-learn
• joblib

Installation

1. Clone the repository:

git clone https://github.com/yourusername/handwritten-digit-classifier.git
cd handwritten-digit-classifier

2. Create a virtual environment and activate it:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the required packages:

pip install -r requirements.txt

Usage

1. Train the model:

python Model2.0.py

2. Start the web application:

python app.py

3. Open your browser and navigate to http://localhost:5000

API Usage

The application provides a REST API endpoint for predictions:

curl -X POST -F "image=@path/to/image.png" http://localhost:5000/api/predict

Example response:

{
    "prediction": 7,
    "confidence": 0.987654321
}

Project Structure

handwritten-digit-classifier/
├── app.py                 # Flask application
├── Model2.0.py           # Model training script
├── model.pkl             # Trained model
├── static/
│   ├── style.css         # Custom styles
│   ├── script.js         # Client-side JavaScript
│   └── uploads/          # Uploaded images
├── templates/
│   └── index.html        # Web interface template
└── README.md             # Project documentation

Model Architecture

The CNN model consists of:

• Two convolutional layers with Batch Normalization
• Max pooling layers
• Dropout for regularization
• Dense layers with ReLU activation
• Softmax output layer

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

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

Acknowledgments

• MNIST dataset
• TensorFlow and Keras
• Flask web framework
• Bootstrap 5