Sign Language Gesture Recognition Using Hand Landmark Detection and Machine Learning

Project Synopsis & Overview

This project develops an automated sign language gesture recognition system for recognizing hand gestures from images and real-time webcam input. The system uses image preprocessing techniques such as resizing, color conversion, noise reduction, and normalization to prepare the input data. Hand detection and tracking are performed to identify the hand region, followed by extraction of 21 hand landmarks and finger joint coordinates. The extracted landmarks are converted into numerical feature vectors and used to train a machine learning classification model for recognizing different sign language gesture classes. During real-time operation, webcam frames are captured and processed to detect the hand and extract landmarks. The trained model predicts the corresponding gesture along with a confidence score. The recognized gesture is finally converted into a character or text output. Model performance is evaluated using accuracy, precision, recall, and F1-score.

Complete Technical Specifications