sarahkhudze/sign-language-detector-tensorflow

A real-time sign language recognition prototype for Chichewa (Malawi national language), combining **MediaPipe hand landmark detection** with a **MNIST-trained CNN model** using TensorFlow. # sign-language-translator-tensorflow 🧾 Chichewa Sign Language Recognition System A real-time sign language recognition prototype for Chichewa , Malawi’s national language. This system detects hand gestures from a webcam and translates them into text — supporting both rule-based classification (MediaPipe) and deep learning classification (TensorFlow CNN model) . 🎯 Goal : Enable inclusive communication for the deaf and hard-of-hearing community in Malawi using AI and computer vision. 🔍 Features ✅ Real-time gesture recognition from webcam 🤖 Rule-based recognition using MediaPipe landmarks 🧠 Deep learning model trained on SL-MNIST (adaptable to CSL data) 📺 Web interface with sentence accumulation 🟣 Purple-themed UI for cultural relevance 🚫 Video feed disabled by default to improve recognition accuracy 📁 Clean folder structure for easy extension 🧩 Technologies Used MediaPipe Hand landmark detection TensorFlow Deep learning model for gesture classification Flask Backend web server OpenCV Image processing and webcam handling HTML/CSS/JavaScript Frontend interface Python 3.9+ Programming language and runtime.