AMERICAN SIGN LANGUAGE ALPHABET RECOGNITION

Abstract

Building and testing an ASL letter and number identification system is the primary 
goal of the "Real-Time American Sign Language Recognition" project. By using 
advancements in computer vision and machine learning, this project seeks to meet the 
urgent demand for improved communication tools for the deaf and hard-of-hearing 
population. The suggested solution uses OpenCV to collect images in real-time, a 
CNN to extract features, and a random forest classifier to recognise American Sign 
Language hand motions. The models were trained using a large dataset that included 
pictures of the American Sign Language alphabet and the numbers 0–9. After 
extensive testing, the model successfully identified static hand motions that represent 
American Sign Language letters with high rates of accuracy. Findings from this 
research suggest that educational resources, assistive technology, and communication 
aids might all benefit from incorporating AI-driven recognition systems, making them 
more accessible and inclusive for those who use American Sign Language (ASL). 
More advanced features of American Sign Language (ASL) and dynamic gesture 
detection are areas that will be investigated in future research.