README.md

metadata

title: Sign Language Recognition
emoji: 🤟
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.0
app_file: app.py
pinned: false

🤟 Sign Language Recognition

An AI-powered application that recognizes American Sign Language (ASL) alphabet letters (A-Z) from images using AutoGluon multimodal deep learning.

🎯 Features

• Real-time prediction: Upload images or use webcam to capture hand signs
• High accuracy: Powered by AutoGluon's state-of-the-art multimodal predictor
• 26 ASL letters: Recognizes all letters from A to Z
• Top-5 predictions: Shows confidence scores for the most likely letters
• Image preprocessing: Displays how the model processes your input (224x224 resize)

🚀 How to Use

1. Upload an image of an ASL hand sign or use your webcam to capture one
2. The model will automatically analyze the image
3. View the top 5 predictions with confidence scores
4. See the preprocessed image that the model actually processes

🧠 Model Details

• Framework: AutoGluon MultiModalPredictor
• Task: Image Classification
• Classes: 26 (A-Z letters of ASL alphabet)
• Input size: 224x224 RGB images
• Model size: ~41.5 MB

📋 Examples

The app includes example ASL signs to help you get started:

• Letter A
• Letter B
• Letter C

🛠️ Technical Stack

• AutoGluon: For multimodal deep learning predictions
• Gradio: For the interactive web interface
• Hugging Face Spaces: For hosting and deployment
• PIL/Pillow: For image processing
• Pandas: For data handling

📦 Repository Structure

.
├── app.py                              # Main application code
├── requirements.txt                     # Python dependencies
├── autogluon_image_predictor_dir.zip   # Trained model (zipped)
└── README.md                           # This file

🔧 Local Development

To run this app locally:

# Clone the repository
git clone https://huggingface.co/spaces/Anyuhhh/sign-language-recognition
cd sign-language-recognition

# Install dependencies
pip install -r requirements.txt

# Run the app
python app.py

📝 Notes

• This model recognizes static ASL letters only (not dynamic signs or words)
• For best results, use images with:
  • Clear hand visibility
  • Good lighting conditions
  • Neutral background
  • Hand positioned centrally in frame

👨‍💻 Author

Created by Anyuhhh