Chinese Sign Language Alphabet Recognition Based on Random Forest Algorithm
Simin Yuan, Yuan Wang, Xin Wang, Hanjie Deng, Shurui Sun, Hui Wang, Pingao Huang, Guanglin Li
Abstract
Sign language is the language deaf-mute people use to communicate with each other. While people with normal hearing generally can not understand it. Sign language recognition allows hard of hearing people to communicate with general society. In this study, we utilized surface Electromyography (sEMG) to recognize Chinese Sign Language alphabet which is an important part of Chinese Sign Language and recognizing them accurately is critical. For this purpose we attached 8 sEMG sensors on the right forearm of the subjects and collected sEMG signal when they were performing all the 30 alphabet letters. Random forest algorithm was used to classify the data after filtering and feature extraction process. Experimental results showed that random forest algorithm achieved an average recognition rate of 95.48% which was higher than Artificial Neural Networks (ANN) and Support Vector Machine (SVM) and had a more stable performance.