Important
Demo Repository: This is a demonstration repository showcasing the UI/UX and core technical capabilities of the SignBridge application. The current version is actively being updated.
SignBridge is a modern web application designed to break communication barriers by providing real-time American Sign Language (ASL) translation. Utilizing computer vision and machine learning, it translates hand gestures into text and speech, facilitating seamless interactions between the deaf and hearing communities.
- Frontend: React (v18.3) with TypeScript
- Build Tool: Vite (v6.3)
- Styling: Tailwind CSS
- UI Components: Radix UI & Lucide React
- AI/ML: TensorFlow.js & @tensorflow-models/handpose
- Animation: Framer Motion (implied via design system)
The core of SignBridge is built on the MediaPipe Handpose model. The application captures video frames from the user's camera and processes them locally using TensorFlow.js to detect 21 distinct hand landmarks (3D coordinates).
Recognition is implemented via a heuristic-based engine that analyzes the relationships between landmarks:
- Finger Extension: Calculating the relative distance between finger tips and MCP joints.
- Palm Orientation: Determining the vector of the hand relative to the camera.
- Pattern Matching: Comparing current hand states against pre-defined ASL gesture patterns (e.g., 'Hello', 'Thank You', 'Yes/No').
Integrated with the Web Speech API, the application provides immediate auditory feedback for recognized signs, allowing for a two-way communication flow.
- Live Recognition Feed: AI-powered overlay showing tracked hand points and real-time translation.
- Demo Mode: A simulated environment for testing features without requiring a physical camera.
- Learn Section: Interactive guides for users to practice and expand their sign language vocabulary.
- History & Analytics: Tracking of recent translations for quick reference.
- Responsive Mockup: Housed within a premium mobile viewport design to demonstrate mobile application potential.
- Node.js (v18 or higher)
- npm or yarn
npm installnpm run devnpm run buildSignBridge — Breaking barriers, one sign at a time.