// project

04 / 06

ASL Recognition

Real-time American Sign Language gesture recognition using MediaPipe and LSTM layers.

This project implements a real-time sign language action recognition system using a Long Short-Term Memory neural network. It classifies five ASL gestures — hello, thanks, iloveyou, yes, and no — using MediaPipe Holistic keypoints, stacked LSTM layers, webcam-based prediction, and evaluation across an 80/20 train-test split.