Malaysian Sign Language Mobile Application Using Deep Learning

This research paper presents the development of a Malaysian Sign Language (MSL) application. The project aims to bridge the communication gap between the deaf community and other Malaysians. The application utilizes the Yolov5 algorithm, chosen for its real-time object detection capabilities and low memory requirements. The waterfall model has been implemented and has been divided into 8 phases. The American Sign Language (ASL) dataset from Roboflow was adapted for MSL due to similar hand gestures. The model achieved a high mAp score of 87.23% after 500 iterations, successfully classifying MSL alphabets with the exception of “J” and “Z”.