sEMG based hand gesture recognition for rehabilitation prosthetics based on machine learning methods

The human hand plays a crucial role in daily activities, and individuals experiencing dysfunction due to strokes or accidents often require therapy to enhance their condition. The usage of surface electromyographic signals to control has become a important research area, being rehabilitation one of it’s applications. The present work propose a CNN architecture that is able to predict six hand gestures with three channels. We enrolled 8 subjects that generated 14,400 samples in our dataset, each one with a hand gesture label. This CNN could work as a control system of a rehabilitation device, for which a database was prepared that accounted for possible hand movements to serve this purpose. The proposed solution shows an accuracy of 90.56% with an inference time of 1.515 ms.