Lightweight Hand Gesture Recognition Using FMCW RADAR With Multibranch Temporal Convolutional Networks and Channel Attention

Abstract

A novel lightweight hand gesture recognition approach that is based on frequency-modulated continuous-wave (FMCW) radio detection and ranging (RADAR), which aims to minimize computational complexity and memory usage as well as maintain a high recognition performance, is proposed in this article. Most of the existing methods use 2-D or 3-D features that are combined with complex neural network structures, which result in high computational costs. The proposed approach in contrast extracts four components, which include range, Doppler, azimuth, and elevation, as the 1-D time-series features. These features are fed into a neural network that comprises a multibranch temporal convolutional network (TCN), depthwise separable (DS) convolutions, and a channel attention mechanism to enhance the classification performance. The experiments were conducted with nine hand gestures that were collected from nine participants. The proposed system achieved a high accuracy of 99.38% with only 44.6 K parameters and 1.84 M floating point operations per second (FLOPs). Extensive ablation studies and comparative experiments against the existing models demonstrated that the proposed method effectively balances the performance and computational efficiency. This study validates the expressive capability of 1-D features for hand gesture recognition and suggests practical applicability in resource-constrained environments, such as embedded systems.