Smart Application for Thai and English Sign Language Translation

Abstract

Currently, the sign language is used as a means of communication for individuals with hearing impairments. Since most people do not have knowledge to use sign language, as a result, miscommunication may occur. The aim of this project is to develop a program for translating Thai and English sign language. It can translate Thai sign language, English sign language, Arabic numbers, and English words. MediaPipe algorithm is used for hand detection and find 21 Keypoints on each hand. This research is able to detect a maximum of two hands, allowing a total of 42 Keypoints to be used to make decisions for the classification of sign language, which encompasses a total of 68 symbols in the test. A total of 3 experimenters are assigned to experiment with the sign language recognition program 5 times per symbol and specified that the webcam must be approximately 50 centimeters away from the experimenter. As a result from the experiment, it was found that English sign language recognition was more accurate than Thai sign language. This is because some signs in Thai sign language require two levels of gestures. This causes errors in sign language recognition. This research has a sign language recognition result of 95.39%.

Keywords

• translate sign language
• mediapipe
• hearing imparied
• thai sign language
• english sign language