Machine Learning-Based Hand Gesture Recognition via EMG Data

Abstract

Electromyography (EMG) data gives information about the electrical activity related to muscles. EMG data obtained from arm through sensors helps to understand hand gestures. For this work, hand gesture data were taken from UCI2019 EMG dataset obtained from MYO thalmic armband were classied with six dierent machine learning algorithms. Articial Neural Network (ANN), Support Vector Machine (SVM), k-Nearest Neighbor (k-NN), Naive Bayes (NB), Decision Tree (DT) and Random Forest (RF) methods were preferred for comparison based on several performance metrics which are accuracy, precision, sensitivity, specicity, classication error, kappa, root mean squared error (RMSE) and correlation. The data belongs to seven hand gestures. 700 samples from 7 classes (100 samples per group) were used in the experiments. The splitting ratio in the classication was 0.8-0.2, i.e. 80% of the samples were used in training and 20% of data were used in testing phase of the classier. NB was found to be the best among other methods because of high accuracy (96.43%) and sensitivity (96.43%) and the lowest RMSE (0.189). Considering the results of the performance parameters, it can be said that this study recognizes and classies seven hand gestures successfully in comparison with the literature.