Real-time Gesture Recognition Based on Improved Artificial Neural Network and sEMG Signals

Abstract

In this paper, we propose a real-time gesture recognition model based on a feedforward artificial neural network for surface electromyography (sEMG) signals, which contains four processes: pre-processing, feature extraction, classification, and post-processing. In the feature extraction and selection stage, sliding windows are used to segment the data, and time-domain features and convolution results are used as input features. To avoid over-fitting of training, a dropout layer is added to the neural network model. In the post-processing part, two statistical methods are used: majority voting and continuous voting result verification. Finally, a test trial was performed using sEMG signals collected from the human body. The results show that the algorithm can achieve 92.7% recognition accuracy and has good real-time performance.