Abstract
The implementation of Convolutional Neural Networks (CNNs), particularly YOLOv8s (You Only Look Once version 8 small), can significantly advance the real-time conversion of American Sign Language (ASL) gestures into text. ASL is a primary communication method for the hearing-impaired community, yet converting it to written text remains challenging. This project addresses the need for an efficient ASL-to-text system, aiming to enhance communication between deaf and hearing individuals. YOLOv8s, known for its superior object detection capabilities, enables the proposed system to identify and interpret ASL gestures in live video feeds, providing instant and accurate text translations. The use of CNNs, especially YOLOv8s, ensures real-time processing, maintaining accuracy without sacrificing speed. The research motivation is to bridge the communication gap between the deaf community and those relying on written or verbal communication. This paper outlines the employed methodology, including the training process and model optimization, and discusses the results and potential applications. The implications of this ASL-to-text conversion system extend to inclusive technology, fostering improved accessibility and communication for individuals with hearing impairments in various contexts