Radar-Based Joint Gesture and Identity Recognition via Multi-Feature Two-Stream Neural Network

This study presents a novel approach for joint gesture and identity recognition utilizing Frequency-Modulated Continuous Wave (FMCW) radar. The proposed approach leveraging a custom-designed two-stream neural network (GI-Radar) to fuse time-varying range and angle information. In the network, we add the Convolutional Block Attention Module (CBAM) to enhance the feature extraction capability of network in both spatial and channel dimensions. Experimental results show that the accuracy of gesture recognition and identity recognition of this method can reach up to 90% and 93% respectively, which is superior to traditional methods and proves that it is a promising wireless solution.