Deep Learning-Based Sign Language Recognition: A Comprehensive Analysis

In an increasingly interconnected world, fostering inclusivity is important, especially for individuals with disabilities. For the community with the people having hearing issues, sign language is a vital communication medium. However, a significant communication barrier arises when sign language users interact with those unfamiliar with it. This results in them having barriers to education, services, and social engagement. This study introduces a real time translation system leveraging multiple ML and Deep Learning algorithms to convert live sign language gestures into grammatically coherent English sentences unlike existing systems that focus on isolated gestures or letters, our approach captures word-level and sentence-level nuances. For this goal currently we leveraged well renowned American Sign Language (ASL) dataset to train our models. A stop signal gesture enables users to indicate the end of a sentence, thus ensuring seamless, fluid communication. By modelling spatial and temporal visual patterns in video datasets, the system aspires to redefine accessibility, empowering the community of people which have hearing issues to communicate more effectively and independently.