Methods for classifying different human activities have been explored widely in recent years. In radar-based human activity classification, most deep learning-based techniques pay great attention to the frame-level structure. These approaches generally adopt short-time Fourier transform, or convolutional layers with a frame-level filter to process the raw radar data, while they seldom consider the overall efficiency. In this paper, we propose a sample-level convolutional neural network named SampleRadarNet. The proposed SampleRadarNet utilizes one-dimensional convolutional layers with small filters to learn more temporal information from the raw data, breaking through the frame-level model's capacity limitation. In addition, the Squeeze-Inception-Mobile module is designed for the classification task, which involves the following three components: Fire module, inception block, and depthwise convolution. The experimental results show that the proposed architecture can achieve better performance compared with the existing related methods in a seven-class radar-based human activity classification problem.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.