Skip to content

mylepaklab/client-app

Repository files navigation

BIMTranslator

BIMTranslator is an AI-powered web application built for the Pan-SEA AI Developer Challenge 2025. Our goal is to bridge the communication gap by enabling real-time translation of BIM (Bahasa Isyarat Malaysia - Malaysian Sign Language) gestures into text, making digital services accessible to the Malaysian deaf community.

What Judges Can Try

API sample:

curl -X GET "https://your-api.example.com/match_animation_sequence?sentence=apa%20nama" \
    -H "Accept: application/json"

Features

  • Real-time BIM gesture recognition: Live webcam capture with MediaPipe hand tracking
  • 21-point hand landmark detection: Precise tracking of both hands with smoothing filters
  • Instant gesture translation: Convert BIM gestures to clear text output
  • Interactive phrase matching: Support for common Malaysian sign language expressions
  • Visual feedback system: Animated playback of gesture sequences with confidence scoring
  • Privacy-first design: All processing happens in the browser, no video data stored
  • Responsive web interface: Works on desktop and mobile devices

How It Works

  1. Capture: Use your webcam to capture hand movements in real-time
  2. Track: MediaPipe extracts 21 landmarks per hand with smoothing filters for accuracy
  3. Recognize: TensorFlow.js model classifies BIM gestures with confidence scoring
  4. Translate: Convert recognized gestures to text and provide visual feedback through animation playback

Technology Stack

  • Frontend: React 18 with TypeScript
  • Build Tool: Farm (Vite-compatible bundler)
  • Hand Tracking: MediaPipe Tasks Vision
  • Machine Learning: TensorFlow.js with custom gesture recognition model
  • Styling: Tailwind CSS with custom brand theming
  • Animation: Framer Motion for smooth interactions
  • Routing: React Router for navigation

Getting Started

Prerequisites

  • Node.js version 18 or newer
  • pnpm (recommended), npm, or yarn
  • Modern browser with WebRTC support for camera access

Installation

Clone the repository:

git clone https://github.com/TraFost/bim-translator.git
cd bim-translator

Install dependencies:

pnpm install
# or
npm install

Start the development server:

pnpm dev
# or
npm run dev

Open your browser and go to http://localhost:3000 to see BIMTranslator.

Available Scripts

  • pnpm dev - Start development server
  • pnpm build - Build for production
  • pnpm preview - Preview production build
  • pnpm clean - Clear persistent cache files

Project Structure

├── public/           # Static assets
│   ├── models/       # TensorFlow.js gesture recognition models
│   └── assets/       # Images and other static files
├── src/              # Source code
│   ├── components/   # React components
│   │   ├── demo/     # Hand sign recognition demo
│   │   └── home/     # Landing page components
│   ├── configs/      # Configuration files
│   ├── constants/    # App constants and content
│   ├── hooks/        # Custom React hooks
│   ├── lib/          # Utility libraries (axios, utils)
│   ├── pages/        # App pages (main, demo)
│   └── types/        # TypeScript type definitions
├── farm.config.ts    # Farm bundler configuration
├── package.json      # Project metadata and dependencies
└── README.md         # This file

For Judges and Reviewers

Impact for SEA Region

  • Removes communication barriers for deaf and hard of hearing users in Malaysia
  • Supports BIM users in accessing digital services and AI assistants
  • Reduces dependency on human interpreters for basic digital interactions
  • Empowers independence in the deaf community through direct AI communication

Technical Innovation

  • Real-time hand tracking with 21-point precision per hand using MediaPipe
  • Client-side processing for privacy and performance
  • Custom gesture recognition model trained for BIM gestures
  • Smooth filtering algorithms for stable landmark detection
  • Interactive visual feedback system with confidence scoring

Feasibility and Performance

  • Live demo available with real-time gesture recognition
  • Browser-based solution requiring no downloads or installations
  • Responsive design working on desktop and mobile devices
  • Privacy-first approach with no video data storage

User Experience

  • One-click access to live demo
  • Intuitive interface designed for accessibility
  • Real-time visual feedback with color-coded hand tracking
  • Clear navigation between landing page and demo

Usage

Demo Access

Visit /demo to try the live gesture recognition demo:

  1. Allow camera access when prompted
  2. Position your hands in view of the webcam
  3. Make BIM gestures and see real-time hand tracking
  4. Select phrases to match and see animation playback
  5. View confidence scores for gesture recognition accuracy

Supported Features

  • Real-time hand landmark detection with visual overlay
  • Gesture recognition with confidence scoring
  • Phrase matching for common BIM expressions
  • Interactive animation playback system
  • Support for both left and right hand tracking

What is BIM?

BIM (Bahasa Isyarat Malaysia) is the official sign language of Malaysia, used by the deaf and hard of hearing community. It's a rich visual language that uses:

  • Hand shapes and movements to convey meaning
  • Facial expressions for grammatical structure
  • Spatial relationships for complex concepts
  • Cultural context specific to Malaysian deaf community

BIMTranslator bridges the gap between this visual language and digital text-based systems.

Legal

  • Privacy Policy
  • Terms of Service

Contact

License and Use

This project is developed for the Pan-SEA AI Developer Challenge 2025. All rights reserved. Please contact the team for any reuse outside the challenge context.


Farm and React

This template helps you start a React and TypeScript app with Farm.

Setup

Install the dependencies:

pnpm install

Get Started

Start the dev server:

pnpm start

Build the app for production:

pnpm build

Preview the production build:

pnpm preview

Clear persistent cache files:

pnpm clean

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages