Real-Time Communication for Disabled People

This project aims to train a neural network model and develop a computer application for the actual time interpretation of hand movements in American Sign Language (ASL), with a particular emphasis on fingerspelling. The motivation stems from the historical significance and expressive nature of sign language, recognizing the communication challenges faced by the deaf community due to limited sign language knowledge and interpreter availability. The proposed research approach involves capturing and filtering hand gestures, followed by a classifier to predict the corresponding ASL letter or number. Rigorous testing demonstrates an impressive 98% accuracy for the 26 alphabet letters. The system captures real-time video, processes it through the neural network, and displays the output as text, facilitating seamless communication for both deaf individuals and those unfamiliar with sign language. Beyond technological innovation, this research addresses a societal need for accessible and inclusive communication, showcasing the potential of technology in breaking down communication barriers and fostering understanding in diverse communities.