🖐️ ASL Detection System (29 Classes)

A real-time deep learning-based American Sign Language (ASL) recognition system.
Detects and classifies 29 ASL alphabet signs (A-Z, SPACE, DEL, NOTHING) from images or webcam input.

---

🚀 Overview

ASL Detection System is a robust, easy-to-use tool for recognizing all 26 ASL alphabets plus SPACE, DEL, and NOTHING gestures. Featuring real-time detection, high accuracy, and a user-friendly interface.

---

📑 Table of Contents

1. Features
2. Demo
3. Installation
4. Requirements
5. Quick Start
6. Usage
7. Project Structure
8. Training the Model
9. Evaluation
10. Dataset
11. Contributing
12. Acknowledgements
13. Live Demo
14. License
15. Troubleshooting
16. Credits
17. Contact / Support

---

✨ Features

• 🔤 Recognizes all ASL alphabets (A-Z) plus SPACE, DEL, and NOTHING
• 🎥 Real-time webcam and image input support
• 💯 State-of-the-art test accuracy
• 🖼️ Confusion matrix visualization
• 🛠️ Easy-to-use scripts for training, detection, and evaluation

---

🎥 Demo

🎬 Live Demo — Try it in your browser!

---

⚙️ Installation

Prerequisites:

• Python 3.8+
• NVIDIA GPU (recommended for training)

Install dependencies:

pip install -r requirements.txt

---

⚙️ Requirements

• Python 3.8+
• TensorFlow 2.x
• OpenCV
• NumPy
• Matplotlib
• scikit-learn

---

🏁 Quick Start

# 1. Clone the repository
git clone https://github.com/PrabalJay/asl-detection-29-classes.git
cd asl-detection-29-classes

# 2. Install dependencies
pip install -r requirements.txt

# 3. Train the model
python train.py

# 4. Evaluate the model
python evaluate.py

# 5. Run real-time detection
python detect.py

---

💡 Usage

• Train the Model:
  python train.py
  Trains the ASL recognition model using the dataset in data/train/

• Evaluate the Model:
  python evaluate.py
  Evaluates the trained model and generates a confusion matrix

• Real-Time Detection:
  python detect.py
  Launches webcam-based or image-based ASL detection

• Check Dataset Images:
  python check_images.py
  Verifies and cleans dataset images

• Configuration:
  python config.py
  Adjusts configuration settings for training and detection

Examples

Detect ASL in an Image

python detect.py
# Example output:
# Predicted: Sign Y (1.00)

Real-time Webcam Detection

python detect.py --webcam
# (Press 'Q' to quit)

---

📁 Project Structure

asl-detection-29-classes/
├── data/
│   ├── train/              # Training images
│   └── test/               # Test images
├── docs/                   # GitHub Pages folder
│   ├── index.html          # Your HTML file
│   ├── asl_demo.mp4        # Video (compressed)
│   ├── sample_1.png        # Demo images
│   ├── sample_2.png
│   └── sample_3.png
├── models/
│   └── model.keras         # Trained model
├── check_images.py         # Dataset verification utility
├── config.py               # Configuration script
├── detect.py               # Real-time/image detection script
├── evaluate.py             # Model evaluation script
├── train.py                # Model training script
├── requirements.txt        # Python dependencies
├── confusion_matrix.png    # Model performance visualization
└── README.md

---

🎯 Training the Model

Organize your dataset as follows:

data/train/
  ├── A/     # Images for 'A'
  ├── B/
  └── ... (all 29 classes)

Train:

python train.py

• Saves best model to models/model.keras
• Generates confusion_matrix.png

---

📊 Evaluation

Evaluate the model:

python evaluate.py

Sample output:

Test Accuracy: 96.3%
Class-wise Precision:
A: 0.98 | B: 0.95 | ...

---

📂 Dataset

• Source: ASL Alphabet Dataset on Kaggle
• Requirements:
  • 3,000+ images
  • ~100 images per class
  • PNG/JPG format

---

🤝 Contributing

Contributions, issues, and feature requests are welcome!
Feel free to fork the repo and submit a pull request.

---

🙏 Acknowledgements

• TensorFlow — Deep learning framework
• OpenCV — Computer vision library
• ASL Alphabet Dataset — Dataset used for training and testing
• scikit-learn — Evaluation and metrics
• Matplotlib — Visualization

Special thanks to the open-source community and contributors for their resources and support.

---

🟢 Live Demo

Try it here

Features Demonstrated:

• ✅ Webcam hand tracking
• ✅ 29-class prediction (A-Z + SPACE/DELETE/NOTHING)
• ✅ Confidence score display

---

📜 License

MIT License — See LICENSE for details.

---

🛠️ Troubleshooting

Issue	Solution
CUDA errors	Install correct TensorFlow-GPU
Missing dependencies	Run pip install -r requirements.txt
Low accuracy	Increase epochs or add more training data

---

🙌 Credits

Built with TensorFlow/Keras

---

📬 Contact / Support

For questions, open an issue.

---