Sign Language Interpreter
Real-time web application that translates hand gestures into text using Mediapipe skeleton tracking and a CNN classifier, built for the Youth For Challenge association.
Python 
Flask 
OpenCV 
Built as a volunteer project for the Youth For Challenge association, this application makes communication more accessible for deaf and mute users by translating hand signs into text in real time.
How it works
Mediapipe Hands extracts a 21-point skeleton from the webcam feed on every frame — wrist, finger joints, fingertips. These normalised landmarks (robust to hand size, skin tone, and lighting) feed into a CNN classifier trained on recorded gesture sequences. The prediction is streamed to a detection history panel alongside the live video.
The backend is a lightweight Flask server; the frontend renders the video feed and detection log in the browser with no additional dependencies.
Scope
The current model covers common expressions and phrases: greetings, affirmations, negations, and basic social phrases. The architecture is designed for incremental expansion — adding new signs requires only recording new samples and retraining the classifier.