G3R: Generating Rich and Fine-Grained Mmwave Radar Data From 2D Videos for Generalized Gesture Recognition

• Published in: IEEE transactions on mobile computing 2024-11, p.1-18
• Main Authors: Deng, Kaikai, Zhao, Dong, Zheng, Wenxin, Ling, Yue, Yin, Kangwen, Ma, Huadong
• Format: Magazinearticle
• Language: English
• Subjects: 2D videos; Accuracy; cross domain translation; Data models; generalized sensing; Gesture recognition; Radar; Radar cross-sections; Radar signal processing; Skeleton; synthetic radar data; Testing; Training; Videos
• ISSN: 1536-1233
• DOI: 10.1109/TMC.2024.3502668

Millimeter wave radar is gaining traction recently as a promising modality for enabling pervasive and privacy-preserving gesture recognition. However, the lack of rich and fine-grained radar datasets hinders progress in developing generalized deep learning models for gesture recognition across various user postures (e.g., standing, sitting), positions, and scenes. To remedy this, we resort to designing a software pipeline that exploits wealthy 2D videos to generate realistic radar data, but it needs to address the challenge of simulating diversified and fine-grained reflection properties of user gestures. To this end, we design G 3 R with three key components: (i) a gesture reflection point generator expands the arm's skeleton points to form human reflection points; (ii) a signal simulation model simulates the multipath reflection and attenuation of radar signals to output the human intensity map; (iii) an encoder-decoder model combines a sampling module and a fitting module to address the differences in number and distribution of points between generated and real-world radar data for generating realistic radar data. We implement and evaluate G 3 R using 2D videos from public data sources and self-collected real-world radar data, demonstrating its superiority over other state-of-the-art approaches for gesture recognition.