Development of Thai Sign Language Interpretation with MediaPipe
Jakkrapan Sudthipadh, Suree Pumrin · 2022
Thai Sign language is a sign language used in Thailand to communicate within the group of people with hearing losses or between ordinary people and people with hearing losses. For a given handshape, it is detected by a camera/webcam. 21 landmarks of a detected hand are inferred as a 3-tuple by MediaPipe. Some of angles between these landmarks are computed using Python programming. The angle set is compared with the training set and classified as a word by K-Nearest Neighbors algorithm. The dataset of Thai sign language used in this research is limited to the vocabularies that can be interpreted by one movement of handshape only. The accuracy of the interpretation in this research is about 84%.