Gait Recognition in Large-scale Free Environment via Single LiDAR

Xiao Han, Yiming Ren, Peishan Cong, Yujing Sun, Jingya Wang, Lan Xu, Yuexin Ma

Abstract: Human gait recognition is crucial in multimedia, enabling identification through walking patterns without direct interaction, enhancing the integration across various media forms in real-world applications like smart homes, healthcare and non-intrusive security. LiDAR's ability to capture depth makes it pivotal for robotic perception and holds promise for real-world gait recognition. In this paper, based on a single LiDAR, we present the Hierarchical Multi-representation Feature Interaction Network (HMRNet) for robust gait recognition. Prevailing LiDAR-based gait datasets primarily derive from controlled settings with predefined trajectory, remaining a gap with real-world scenarios. To facilitate LiDAR-based gait recognition research, we introduce FreeGait, a comprehensive gait dataset from large-scale, unconstrained settings, enriched with multi-modal and varied 2D/3D data. Notably, our approach achieves state-of-the-art performance on prior dataset (SUSTech1K) and on FreeGait. Code and dataset will be released upon publication of this paper.

Primary Subject Area: [Content] Media Interpretation

Secondary Subject Area: [Experience] Multimedia Applications

Relevance To Conference: The International Workshop on Human-Centric Multimedia Analysis (HCMA) has been a part of the ACM MM conference since 2020. In line with the increasing focus on human-centric studies at ACM MM, our paper explores human gait recognition, a core topic of the HCMA workshop. By accurately identifying individuals through their gait characteristics, our research paves the way for more intuitive, responsive, and engaging multimedia applications in areas such as non-intrusive security, smart homes, and healthcare. Utilizing a single LiDAR, we have developed a novel gait recognition method that effectively captures dense body structure features from LiDAR range views along with gait-related geometric and motion information from raw LiDAR point clouds. This method is particularly effective in real-world scenarios, free from constraints related to lighting and viewpoint. Additionally, we introduce an extensive, in-the-wild, free-trajectory gait recognition dataset that includes diverse real-world scenarios, data modalities, and representations, thus providing a rich resource for the gait research community. The growing acceptance of point cloud processing research at ACM MM—27 papers in 2023, 15 in 2022, and 14 in 2021—highlights the community's increasing recognition of its significance in the multimedia field. We believe our research significantly enriches the ongoing discourse in human-centric multimedia analysis, pushing the boundaries of how we understand and interact with human dynamics through multimedia technologies.

Submission Number: 2843