Finger Recognition Using a Wearable Device while Typing

Abstract

In the information society, the ability to use a computer is important. To use a computer, users commonly need a keyboard as an input device. If users place their fingers on the home keys and stroke each key using the correct finger, they will lead to improve their typing skills. In this study, we develop a stroked finger recognition method for keyboard typing using Myo, a wearable device that can simply measure the surface electromyography (EMG) signal of the user's arm. Our method detects the user's stroked finger through machine learning that uses the measured EMG. We introduced window functions during feature extraction in order to suppress the influence of the keystroke speed. Our method was capable of recognizing six categories (five fingers and a neutral state) with an accuracy of about 80% when our method was evaluated by a 10-fold cross validation for each subject's data.