Real-Time ASL Detection

About

A real-time computer vision project that recognizes American Sign Language letters and digits from a webcam feed. It combines hand tracking with a machine learning classifier to turn detected hand positions into an on-screen prediction.

Recognizes 36 ASL letters and digits in real time on consumer hardware.

How it works

1. OpenCV captures frames from the webcam and makes them available for processing.
2. MediaPipe tracks the hand and extracts landmarks from each frame.
3. A Random Forest classifier uses the landmark data to predict an ASL letter or digit, and the result appears with the live camera feed.

Built with

The pipeline uses Python and OpenCV for camera capture, MediaPipe for hand tracking, NumPy for working with landmark data, and Scikit-learn for the Random Forest classifier. It recognizes 36 letters and digits on consumer hardware.

Python · OpenCV · MediaPipe · Scikit-learn · Numpy

Links

• Source code on GitHub