Automatic Leg Gesture Recognition Based on Portable Electromyography Readers

In this paper, recognition of leg gestures is performed using Linear Discriminant Analysis in order to propose a real application for prosthetic leg considering transfemoral amputee. As results, the confusion matrix shows the performance of the algorithm, where the Class #1 and #3 were the best classes classified (sensitivity is 100%), and Class #2 was the worst classified (sensitivity is 67%). In addition, the probability that the classifier ranks a randomly chosen positive instance higher than a randomly chosen negative for Class #2 and #4 is the same, AUC =0.94, and AUC =1 for Class #1 and #3. Although the hardware and algorithm used have adequate performance, the optimization and improve the real testing conditions are important requirements for real human applications.