taashatrinita/isl

This is a college demo prototype that recognizes Indian Sign Language (ISL) gestures in real time from a webcam feed and translates them into on-screen text and spoken audio. The goal is to bridge communication between people who use sign language and those who do not. The application runs in a browser using Streamlit, a Python tool for building simple web interfaces. When you start it, it opens your webcam and uses MediaPipe, a library that can detect and track hand positions in video, to identify the locations of your finger joints and knuckle points in each frame. It then applies simple geometric rules to those positions to classify which gesture you are making: whether your fingers are spread apart, close together, or in a thumbs-up position corresponds to "HELLO", "THANK YOU", or "HELP" respectively. When a gesture is recognized, the label appears on screen alongside a confidence indicator, and it can optionally read the word aloud using text-to-speech. The readme is upfront that this is a limited demo: it only recognizes three gestures and uses hand-coded rules rather than a trained machine learning model, which means it is not a general-purpose sign language recognizer. A full system would need a proper dataset and a trained classifier. The project is written in Python and runs on Windows. It is intended as a starting point or demonstration for a college project rather than a production tool.