A Deep Learning based Recognition System for Yemeni Sign Language

Abstract

There are more than 466 million people with hearing disabilities in the world. Those people need to communicate with others, get learning and interact with activities around them. Sign language is the bridge to eliminate the gap between them and other people. Developing an automatic system to recognize sign language has a lot of challenges, especially for Yemeni sign language, as there are very few researches touching on this language. In this paper, we propose a new Convolution Neural Network based model for classifying the sign language of Yemen. The System was trained and tested using a dataset that includes 16,192 images gathered from 40 people with different distances and variations. The proposed model uses pre-processing methods to remove noises and reposition the images, etc. The results display that the proposed model achieved 94% accuracy.