Real-Time Myanmar Sign Language Recognition Using Deep Learning

Abstract

Language is the main communication tool between human beings. People who have suffered from deaf and dumb use sign language for communication. To communicate with deaf and dumb people, we need to know sign language well. Researchers are seeking to explicit the meaning of sign language and to develop communication among deaf people and normal people. The significance of sign language recognition is hand detection, sign classification, and language translation. The main focus of this work is to develop a vision-based sign language recognition system, to transfer a tiny YOLO object detection model, a CNN-based classification model, and to develop a graph-based recognition model. Experimental results show that our proposed system can recognize Myanmar sign language not only with an accuracy of over 98% but also in real-time. In this paper, we have used 29 different hand signs for the proposed model in MATLAB 2020a.