Programmable Metasurface Based Microwave Gesture Detection and Recognition Using Deep Learning

摘要

Gesture is an important way of communication in people’s daily life, helping people to understand each other in some special cases. Most of recent research is of the optical field, which is limited when in dark or with obstacles, while the microwave can address these issues with its all-day, all-weather, and penetrability properties. Programmable metasurfaces have been widely applied in many fields since it’s programmable, flexible, controllable and multifunctional. In this paper, we realize the gesture detection and recognition using deep learning neural network, Faster Region-based Conventional Neural Network (Faster R-CNN) and Conventional Neural Network (CNN). The detection and recognition here are based on the microwave images with programmable metasurface. The accuracy of gesture detection and recognition are both very high, beyond 90%.