Gesture Recognition with Residual Attention Using Commodity Millimeter Wave Radar

Gesture recognition technology based on millimeter wave radar can identify and classify user gestures in non-contact scenarios. To address the complexity of data processing with multi-feature input in neural networks and the poor recognition performance with single-feature input, this paper proposes a gesture recognition algorithm based on ResNet-Long Short Term Memory with attention mechanism (RLA). In the signal processing stage of RLA, a range-Doppler map is generated by extracting range and velocity features from the original mmWave radar signal. In the network architecture, RLA combines the features of a residual network with channel and spatial attention modules, ensuring that no useful information is overlooked. We introduce the residual-attention mechanism to enhance the network’s focus on gesture features and avoid the impact of irrelevant features on recognition accuracy. Furthermore, the long short-term memory network is employed to handle temporal features, ensuring high recognition accuracy even when only single-feature inputs are available. A series of experimental results show that the proposed RLA algorithm in this paper has superior recognition performance.