IndicSignAI - Indian Sign Language Recognition & Translation

A real-time Indian Sign Language (ISL) recognition and translation web application supporting 13 regional languages.

🌟 Features

• Real-time Sign Detection: 178+ ISL signs recognized using deep learning
• Multi-language Translation: Translate to 13 regional languages
  • Assamese, Hindi, Manipuri, Nepali, Marathi, Odia, Mizorami
  • Gujarati, Tamil, Telugu, Bengali, Meitei Lon, Dzongkha
• Voice Input: Speech-to-text for text-to-sign conversion
• Text-to-Speech: Audio output for detected signs
• Web Interface: Responsive design for desktop and mobile

📁 Project Structure

PROTOTYPE 5/
├── src/                    # Main application source
│   ├── app.py             # Flask web application
│   └── models/            # ML model modules
│       ├── translation.py
│       ├── isl_image_model.py
│       └── meitei_lon_fallback.py
├── models/                # Trained model files
│   └── ISL_IMAGE.keras/   # Primary ISL model
├── templates/             # HTML templates
├── static/                # CSS, JavaScript
├── training/              # Training scripts & data
│   ├── scripts/
│   └── data/
├── scripts/               # Utility scripts
│   ├── run.bat
│   └── install_dependencies.py
├── archive/               # Legacy files (preserved)
│   ├── legacy_models/
│   ├── experiments/
│   └── prototypes/
├── assets/                # 3D characters, images
└── requirements.txt       # Python dependencies

🚀 Quick Start

1. Install Dependencies

# Using the provided script
python scripts/install_dependencies.py

# Or manually
pip install -r requirements.txt

2. Run the Application

# Using the batch script (Windows)
scripts\run.bat

# Or directly
cd src
python app.py

3. Open in Browser

Navigate to: http://localhost:5000

🎯 Usage

1. Sign to Text: Show hand signs to the camera
2. Text to Sign: Type or speak text to see sign representations
3. Translation: Convert between English and regional languages
4. Voice Input: Use microphone for speech-to-sign

🛠️ System Requirements

• Python 3.8+
• Webcam (for sign detection)
• 4GB+ RAM recommended
• Windows/Linux/macOS

📦 Dependencies

• TensorFlow 2.13.0
• OpenCV 4.8.1
• Flask 2.3.3
• MediaPipe 0.10.21
• PyTorch 2.0.1
• Transformers 4.30.2

🔧 Configuration

Key settings in src/app.py:

• PROCESS_EVERY_N_FRAMES = 3 - Detection frequency (higher = less CPU)
• CAMERA_WIDTH = 480 - Camera resolution
• CONFIDENCE_THRESHOLD = 0.15 - Detection sensitivity

📝 License

This project is for educational and research purposes.

🙏 Acknowledgments

• NLLB-200 for translation capabilities
• MediaPipe for hand tracking
• TensorFlow for deep learning framework