3D-LSTM Wireless Sensing Gesture Recognition —A Collaborative Bachelor and Master Project-Based Learning Case

Abstract

When learning core theoretical courses in bachelor and professional master majors in electronic information engineering, such as communication engineering, electronic information engineering, artificial intelligence, etc., many students face difficulties in understanding abstract signals and deep learning knowledge taught using rigorous mathematical derivations. They are often at a loss applying the theoretical knowledge. We propose a collaborative bachelor and master student project-based learning (PBL) approach and present the Wi-Fi gesture recognition application case. The project not only provides undergraduates an intuitive understanding of signals, signal processing methods, and deep learning methods in real-life applications, but also improves their technical application skills and teamwork skills. The master student proposes a 3D-LSTM method in deep learning algorithm which improves the gesture recognition accuracy and significantly reduces the model complexity. The project not only improves the master student’s research ability, but also improves his teaching and teamwork skills.

This work is supported by 2020 Zhejiang Provincial-level Top Undergraduate Courses (Zhejiang Provincial Department of Education General Office Notice (2021) No. 195), Classroom Teaching Reform Project at Zhejiang University of Science and Technology (No. 2018-ky3), Top Undergraduate Courses Development Project (Nos. 2020-k11, 2022-k4, 2020-k10).