LARGE SCALE IMAGE CLASSIFICATION FOR REAL-TIME SIGN LANGUAGE RECOGNITION

Interaction and communication for a normal human being is simpler compared to a disabled person withconditions such as speaking and hearing who can encounter communication issues with other individuals. SignLanguage simplifies this communication barrier between a disabled and normal person. Increased focus oninclusion and accessibility in communication highlights the importance of effective sign language recognitionsystems. This project overcomes the problem of connecting the communication gap between sign language usersand non-sign language users, by creating a strong real-time sign language recognition system. Based onimprovements in deep learning and large-scale image classification, the suggested system interprets signlanguage gestures captured through a camera and converts them into text or speech in real-time. The model istrained on a large dataset covering a wide range of hand gestures, providing coverage over a majority of signlanguages and reducing the impact of biases. Mechanisms like convolutional neural networks (CNNs), transferlearning, and real-time optimization algorithms are used to provide high accuracy and low latency. The projecthas uses in accessibility technology, increasing communication for the hearing and speech disabled, andincreasing the inclusivity of society. Its scope lies in integration in education tools, virtual assistants, and publicservices. By merging innovation in machine learning with emphasis on social impact, this project significantlyadds to the dismantling of communication barriers and enabling the differently-abled population.