GCN-Enhanced Multidomain Fusion Network for Through-Wall Human Activity Recognition

This letter considers the problem of human activity recognition (HAR) behind the walls using ultrawideband (UWB) radar. The graph convolutional network (GCN)-enhanced multidomain fusion network (GMFN) is proposed to improve the recognition performance by utilizing the complementarity of the multidomain features. Specifically, first, a multibranch convolutional neural network (CNN) is proposed to extract the multidomain features from the range, time-frequency (TF), and range-Doppler (RD) domain. Then the multidomain features are constructed as a graph, and the GCN is employed to fuse the multidomain features on the graph. Finally, HAR is implemented in the form of graph classification. The experimental results on the real data show that the proposed GMFN achieves better performance than the state-of-the-art multidomain fusion HAR methods.

Automated summary

This study proposes a method for human activity recognition using ultrawideband radar, and enhances the fusion of multidomain features using graph convolutional networks, achieving better recognition performance. By extracting features from different domains and constructing a graph, human activities can be classified and recognized.