Real-Time Vision-Based Sign Language Recognition Using Mediapipe and CNN

Communication barriers between hearing and speech-impaired individuals and the general population often hinder effective interaction and social inclusion. Traditional approaches to sign language interpretation rely heavily on human translators, which can be costly, inconsistent, and unavailable in real-time scenarios. With advancements in computer vision and deep learning, automated sign language recognition offers a promising solution to bridge this communication gap. This project presents a vision-based Sign Language Recognition (SLR) system that interprets hand gestures captured through a standard webcam into corresponding text or speech output. The system utilizes OpenCV for image preprocessing, Mediapipe for real-time hand landmark detection, and a Convolutional Neural Network (CNN) model for gesture classification. The framework is designed to recognize alphabets and commonly used signs from Indian Sign Language (ISL) with high accuracy and responsiveness. Experimental results show that the proposed system achieves reliable gesture recognition under varying lighting and background conditions. It provides an efficient, low-cost, and accessible solution that promotes inclusive communication, with potential applications in assistive technology, education, and human–computer interaction.