Logo Lanfrica

Ayoub-ABIDI/E-learNIT

Domaine:

natural language processingeducation

Type de record:

softwareproject
Créateur:
Ayo
Hôte:
AI-powered Tunisian Sign Language translator for online education | Real-time recognition using RAG (CLIP + LLaMA) | Chrome Extension for Google Meet, Teams & Zoom | Herotopia Challenge 2025 # E-learNIT: AI-Powered Tunisian Sign Language Translator **Real-time Tunisian Sign Language Translation for Online Education** Demo Video • Documentation • Quick Start • Architecture --- ## Table of Contents - Problem Statement - Our Solution - Innovation: RAG vs Classical CV - Technical Architecture - System Workflow - Installation & Setup - Usage Guide - Supported Vocabulary - Performance Metrics - Future Roadmap - Team & Acknowledgments --- ## Problem Statement ### The Challenge In Tunisia, **over 250,000 deaf individuals** face significant barriers in accessing online education: | Challenge | Impact | |-----------|--------| | **No real-time translation** | Deaf students cannot participate in live discussions | | **Limited accessibility tools** | Existing tools don't support Tunisian Sign Language (TSL) | | **Communication barriers** | Unable to ask questions or contribute during virtual classes | | **Language isolation** | Tunisian Sign Language differs from international standards | | **Educational inequality** | 70% lower graduation rates compared to hearing peers | ### Our Target Users - **Deaf and hard-of-hearing students** in Tunisia - **Teachers** with deaf students in their online classes - **Educational institutions** seeking inclusive platforms - **Online meeting participants** who use Tunisian Sign Language --- ## 💡 Our Solution **E-learNIT** is a Chrome extension that provides **real-time Tunisian Sign Language translation** for online education platforms, enabling deaf students to: ### ✅ Core Features ``` ┌────────────────────────────────────────────────────────────┐ │ 🎥 WEBCAM → 🤟 SIGN → 📝 TEXT → 🔊 SPEECH → 💬 CHAT │ └────────────────────────────────────────────────────────────┘ ``` 1. **🎥 Real-time Sign Recognition** - Captures signs from webcam every 3-4 seconds - Works seamlessly during video calls - No manual intervention needed 2. **📝 Multi-language Translation** - **Tunisian Arabic** (Darija) - Native …