期刊名称：International Journal of Advances in Engineering and Management
电子版ISSN：2395-5252
出版年度：2022
卷号：4
期号：4
页码：755-759
DOI：10.35629/5252-0404519536
语种：English
出版社：IJAEM JOURNAL
摘要：sEMG is a widely used method of human-computer interaction that has been used in a number of scenarios. To enable sEMG classification, numerous methods of machine learning based approaches have also been developed. However, despite its popularity in the computer vision sector, In sEMG deciphering, the deep neural network has a very restricted use. In this research, we used a novel deep learning framework to classify hand gestures based on sEMG. In particular, we used a convolutional neural network (CNN) to classify sEMG with numerous sessions. This is more problematic because of the different time-varying biodynamics of the participants. As a result, we've looked at several CNN topologies in the hopes of finding an optimum design that can successfully discover hidden characteristics in signals. For surface EMG-based hand gesture recognition, the proposed CNN framework has a better classification accuracy, and the varied topologies have a substantial impact on CNN performance. This research lays a solid foundation for CNN to recognise multiple-session sEMG signal patterns.
