Recognize Vietnamese Sign Language Using Deep Neural Network

Abstract

World Health Organization published an article called ‘Deafness and hearing loss' in March 2020, it said that more than 466 million people in the world lost their hearing ability, and 34 million of them were children. Sign Language has been born and developed for a long time, but its application to communicate has met with many inadequacies and difficulties. Many methods of Computer Vision-based approach gave good results on Sign Language Alphabet Recognition but all of them require the perfect result from background removing step. However, when it comes to real life, removing a complex background is too difficult for any simple background removing algorithms. In this work, our main purpose is to build a model based on deep learning that can recognize Vietnamese Sign Language Alphabet in a complex environment. Results obtained show a robust accuracy of this model in recognizing Vietnamese Sign Language Alphabet.