Causal Localization Network for Radar Human Localization With Micro-Doppler Signature

Abstract

The Micro-Doppler (MD) signature includes unique characteristics from different-sized body parts such as arms, legs, and torso. Existing radar identification systems have attempted to classify human identification using these characteristics in MD signatures, achieving remarkable classification performance. However, we argue that radar identification systems should also be extended to perform more fine-grained tasks for greater identification flexibility. In this paper, we introduce the radar human localization (RHL) task, which involves temporally localizing human identifications within untrimmed MD signatures. To facilitate RHL, we have constructed a micro-Doppler dataset named IDRad-TBA. Additionally, we propose the Causal Localization Network (CLNet) as the baseline system for RHL, built on the IDRad-TBA dataset. CLNet utilizes a novel temporal causal prediction approach for MD signature localization. Experimental results demonstrate CLNet’s effectiveness in executing the RHL task. Our project is available at: https://github.com/dbstjswo505/CLNet.

Keywords

• Deep learning
• temporal human identification
• micro-Doppler radar
• information retrieval