SLD-TENSORFLOW: INNOVATIONS IN SIGN LANGUAGE RECOGNITION

Abstract

Sign language is an essential mode of communication for individuals with hearing and
speech impairments. This project focuses on developing a Sign Language Detection System that
leverages deep learning techniques to recognize and interpret Indian Sign Language (ISL) gestures
in real time. By using the TensorFlow Object Detection API, the system is trained on a dataset
consisting of labeled images to accurately classify different hand gestures. The trained model
processes live video input from a webcam, enabling seamless recognition and translation of
gestures into corresponding text or speech. To ensure accessibility and ease of use, a web-based
application has been designed with multiple modules. The system includes a Live Gesture
Detection module that captures and processes real-time video input, providing instant recognition
of sign language gestures. A Dataset Management module allows users to manage and update the
dataset, ensuring adaptability to new gestures. Additionally, the Model Training & Testing module
enables users to retrain and evaluate the model for improved accuracy. The Translation module
converts recognized gestures into text or speech output, facilitating effective communication. This
project aims to bridge the communication gap between sign language users and nonsign language
users by offering an intuitive and efficient solution. The system has potential applications in
various domains, including education, healthcare, and accessibility services