Machine Learning based Sign language Detection

Abstract

Communication is essential for human interaction, but for people with hearing and speech impairments, expressing thoughts through spoken language is difficult. Sign Language serves as their main communication medium, yet most individuals cannot understand it, creating a social barrier. With the growth of Artificial Intelligence (AI) and Computer Vision (CV), several systems have been developed to recognize and translate sign gestures into text or speech.This review paper discusses recent advancements in Indian Sign Language (ISL) recognition, focusing on two major approaches: the Web-Based ISL Converter, which uses Convolutional Neural Networks (CNN) for gesture-to-speech translation, and the SignFlow model proposed in IEEE Access (2025), which combines CNN and Transformer architectures for real-time continuous ISL recognition. The analysis shows that while CNN-based systems are lightweight and easy to deploy, hybrid CNN–Transformer models achieve higher accuracy and better real-time performance. The study concludes that integrating these methods can lead to intelligent, multilingual, and inclusive tools that bridge the communication gap for the deaf and mute community.