Turkish Sign Language Recognition by Using Wearable MYO Armband

Abstract

In this study, it is aimed to automatically recognize Turkish Sign Language based on EMG signals gathered by a wearable device namely MYO armband. EMG, acceleration, and gyroscope data were recorded and then they were subjected to preprocessing and feature extraction. The data set formed as a result of these processes was classified by using Support Vector Machine, k-Nearest Neighbor, and Random Forest algorithms. In addition, the effect of EMG and inertial measurement unit (IMU) sensors on overall performance were evaluated. Additionally, the effects of the window size and sliding interval variables used in the sliding window method were also evaluated. As a result, the highest performance was obtained by using the RF algorithm with a window size of 100 and a scroll size of 20, with 96% when EMG and IMU sensors used together. It can be concluded that the MYO armband can be used successfully in recognition of Turkish Sign Language.