LidarGait: Benchmarking 3D Gait Recognition with Point Clouds

1The University of Hong Kong, 2Southern University of Science and Technology, 3The Hong Kong Polytechnic University

SUSTech1K and LidarGait endow 3D information for gait recognition in the real world.

Download

Step1: Download

Download when you are applying dataset via link: (GoogleDrive, BaiduYun code: 4zbf, OpenXLab (Recommend) ).

Step2: Agreement

Signing the Agreement and sending it to email (shencf2019@mail.sustech.edu.cn) with the subject “[SUSTech1K Dataset Application]”. Then follow the instructions to play the dataset.


Abstract

Video-based gait recognition has achieved impressive results in constrained scenarios. However, visual cameras neglect human 3D structure information, which limits the feasibility of gait recognition in the 3D wild world.

Instead of extracting gait features from images, this work explores precise 3D gait features from point clouds and proposes a simple yet efficient 3D gait recognition framework, termed LidarGait. Our proposed approach projects sparse point clouds into depth maps to learn the representations with 3D geometry information, which outperforms existing point-wise and camera-based methods by a significant margin. Due to the lack of point cloud datasets, we build the first large-scale LiDAR-based gait recognition dataset, SUSTech1K, collected by a LiDAR sensor and an RGB camera. The dataset contains 25,239 sequences from 1,050 subjects and covers many variations, including visibility, views, occlusions, clothing, carrying, and scenes.

Extensive experiments show that (1) 3D structure information serves as a significant feature for gait recognition. (2) LidarGait outperforms existing point-based and silhouette-based methods by a significant margin, while it also offers stable cross-view results. (3) The LiDAR sensor is superior to the RGB camera for gait recognition in the outdoor environment.

Video

The SUSTech1K Benchmark

Diverse Attributes

The SUSTech1K dataset preserves the variances found in ex- isting datasets, such as Normal, Bag, Clothes Changing, Views and Object Carrying, while also considering other common but challenging variances encountered outdoors, including Occlusion, Illumination, Uniform, and Umbrella.

Multiple Modalities

The SUSTech1K dataset is a synchronized multimodal dataset, with timestamped frames for each modality of frames.

Examplas

Interpolate start reference image.

Statistics about SUSTech1K dataset

Interpolate start reference image.

Our proposed baseline: LidarGait

Interpolate start reference image.
-->

BibTeX

@InProceedings{Shen_2023_CVPR,
      author    = {Shen, Chuanfu and Fan, Chao and Wu, Wei and Wang, Rui and Huang, George Q. and Yu, Shiqi},
      title     = {LidarGait: Benchmarking 3D Gait Recognition With Point Clouds},
      booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      month     = {June},
      year      = {2023},
      pages     = {1054-1063}
  }