Sign Language Recognition Using Deep Learning

Abstract

Sign language is an essential means of communication for individuals who are speech- or hearing-impaired. It uses structured hand gestures, movements, and facial expressions to convey meaning. However, most people are not familiar with sign language, which creates a significant communication barrier between speech-impaired individuals and the rest of society.
With advancements in Artificial Intelligence and Deep Learning, it has become possible to automatically recognize hand gestures and translate them into understandable text or speech. Sign language recognition systems aim to bridge this communication gap by converting hand gestures into natural language in real time.
This project focuses on developing a real-time sign language recognition system using deep learning techniques. The system captures live hand gestures using a smartphone or webcam and converts them into corresponding text and speech outputs. By enabling real-time processing, the system supports faster and more intuitive communication.
Traditional gesture recognition systems rely on static images or offline video datasets, which are ineffective for real-time gesture interpretation. These systems often fail to capture the temporal dynamics of gestures. To overcome these challenges, the proposed system integrates Media Pipe for real-time hand landmark detection and a LSTM hybrid deep learning model for gesture classification. The LSTM models temporal dependencies across gesture sequences.