Sign Language Recognition for Deaf and Dumb

• This sign language recognition project involved creating a custom dataset, preprocessing images, training a model, integrating with React, and hosting with Firebase.

• The result is a real-time sign language recognition application that recognizes a variety of sign language gestures.

• Our Model is trained for 26 alphabets and 16 words of ASL and which are commonly used in general communication.

Features

• Real-Time Recognition

• Easy-to-Use Interface

• Adaptive Learning

• High Accuracy

• Real Time User Progress Data

Tech Stack

Front-end:

• React
• Redux
• Tailwind

Back-end:

• NodeJS
• SQLite

Machine Learning Framework:

• MediaPipe

NPM Packages:

• @mediapipe/drawing_utils
• @mediapipe/hands
• @mediapipe/tasks-vision
• @redux-devtools/extension
• @reduxjs/toolkit
• framer-motion
• lucide-react
• react-dom
• react-redux
• react-router-dom
• react-webcam
• tailwind-css
• redux
• axios
• redux-thunk
• uuid

Project Details

• Research Documentation: View .docx

• Research Paper: View .docx

• Dataset Link: Sign_Dataset

• Gesture Recogition Documentation: Mediapipe

• A Copy of Trained Model is located in root folder

Steps to set up the project

• Clone this git repository
• Open terminal, type cd backend then npm install followed by npm run start
• Now open another terminal, type cd project then npm install followed by npm run dev
• Wait for the react development server to start, if not opened automatically, manually visit localhost:5173

Acknowledgements

• React
• mediapipe
• NPM

Screenshots

🤝 Contributing

• Fork the repo and create a feature branch.
• Write clean, modular, well-commented code.
• Submit a pull request with a clear description.

📄 License

This project is licensed under the GPL 3.0 License.