EMG Based Gesture Recognition Using Machine Learning
Nikitha Anil, S H Sreeletha · 2018
Gesture recognition basically involves the usage of hardware equipments and software development tools where human movements are captured and Human Computer Interaction are improved. Gesture recogntion can be employed in various applications like gaming technology, virtual reality, in the domain of medicine, sign language interpretation, home automation etc. This work basically focuses on EMG based gesture recogntion taken in real time using Myoband. The Myoband is worn on the forearm and electrical impulses are measured for five gestures namely Rest, Fist, Wavein, Waveout and Fingerspread. These signals undergo a signal processing technique called Wavelet decomposition where the signals are fragmented into wavelet coefficients that are localised in time domain as well as in frequency domain. These coefficients make up the dataset and are classified using Support Vector Machines. When a user poses for a particular gesture, the model would recognises it and labels accordingly.