saintdannyyy/Unicef-Startup-Lab-Hackathon

Wote - AI-powered Ghana Sign Language learning & competitive gaming (Telegram bot + web frontend + ML pipeline) — built for UNICEF Startup Lab (3rd Place, Best AI Implementation). # Wote — Ghana Sign Language (GSL) Learning Platform **Wote** is a comprehensive open-source platform for learning and practicing Ghanaian Sign Language (GSL). It provides an integrated suite of tools for beginners to learn signs, practice with AI-powered feedback, play competitive multiplayer games, and contribute to a growing community dictionary. 🏆 **Built for UNICEF Startup Lab — Awarded 3rd Place & Best AI Implementation** ## Key Features ✨ **Telegram Bot** (`wote_gslbot/`) - Solo practice mode (3-question quick quizzes with instant feedback) - 2-player synchronized multiplayer matches with real-time leaderboard - Integrated GSL dictionary with 100+ signs - Works via polling (easy demo setup) or webhooks 🌐 **Web Frontend** (`wote_web/` — React + TypeScript) - Learn module with structured lessons (alphabets, numbers, common words) - Practice modes (quiz, freestyle, challenge) - Live webcam sign detection using MediaPipe + TensorFlow.js - Beautiful Duolingo-style UI with progress tracking and star rewards - Mock detection mode for reliable demos 🤖 **ML Training Pipeline** (`model_training/`) - Browser-based data collection UI for landmark annotation - Hybrid neural network training (dense + sequence models) - Export to TensorFlow.js for browser inference - Dataset analysis and quality validation tools 📱 **Supports Multiple Media Formats** - Videos: `.mp4`, `.mov`, `.avi` - Images: `.png`, `.jpg`, `.jpeg` ## Repository Structure ``` codeworks/ ├── wote_gslbot/ # Telegram bot + game engine │ ├── bot_enhanced.py # Main bot (polling + webhook support) │ ├── game_database.py # Game rooms, scoring, leaderboard logic │ ├── database.py # Media and dictionary scanner │ ├── config.py # Configuration with env vars │ ├── requirements.txt # Python dependencies │ ├── data/ # Media assets │ │ ├── dictionary.json # GSL dictionary entri …