Human activity recognition (HAR) is an important research area with a wide range of application scenarios, such as smart homes, healthcare, abnormal behavior detection, etc. Wearable sensors, computer vision, radar, and other technologies are commonly used to detect human activity. However, they are severely limited by issues such as cost, lighting, context, and privacy. Therefore, this paper explores a high-performance method of using channel state information (CSI) to identify human activities, which is a deep learning-based spatial module-temporal convolutional network (SM-TCNNET) model. The model consists of a spatial feature extraction module and a temporal convolutional network (TCN) that can extract the spatiotemporal features in CSI signals well. In this paper, extensive experiments are conducted on the self-picked dataset and the public dataset (StanWiFi), and the results show that the accuracy reaches 99.93% and 99.80%, respectively. Compared with the existing methods, the recognition accuracy of the SM-TCNNET model proposed in this paper is improved by 1.8%.

The wideband microwave imaging system is a passive focal-plane imaging system which is used for large-scale, wideband electromagnetic interference source (EMS) imaging. The system is mainly composed of a parabolic reflecting surface and a multi-channel ultra-wideband signal acquisition system. However, due to the influence of manufacturing processes and the varied response characteristics of the sensors to different frequency radiation, the stripe noise exists in the obtained electromagnetic (EM) images, which severely affects the accuracy of localization. To solve this problem, an innovative wavelet deep unfolding network from the perspective of the transform domain is presented in this paper. The network fully considers the inherent characteristics of stripe noise and the complementary information between the coefficients of different wavelet sub-bands to accurately estimate stripe noise while minimizing computational cost. An iterative deep unfolding structure is employed to remove stripe noise by exploiting the correlation between adjacent row signals. It iteratively refines the noise estimation, using the output of each network iteration as input for the subsequent one. A bidirectional gated recurrent unit with a spatial attention mechanism is introduced to enhance the long-time correlation, thus separating the scene details from the stripe noise more thoroughly and restoring the details accurately. Furthermore, a novel stripe noise mathematical model and a wideband dataset are developed. These innovations enable the proposed algorithm to effectively handle dynamically varying noise in wideband. The extensive experiments on simulated and real data demonstrate that our proposed method outperforms several classical de-striping methods on both quantitative and qualitative assessments.