Research map / Label-efficient and robust learning / Recognition and long-tail
OpenGait: A Comprehensive Benchmark Study for Gait Recognition Toward Better Practicality
Chao Fan, Saihui Hou, Junhao Liang, Chuanfu Shen, Jingzhe Ma, Dongyang Jin, Yongzhen Huang, Shiqi Yu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the MGP branch (ablation) | GaitSet · Gait3D | Rank-1: 45.8 -> 44.3 | Hurt |
| add multi-scale HPP (ablation) | GaitSet · Gait3D | Rank-1: 45.8 -> 43.7 | Hurt |
| replace regular convolution with Focal Convolution (ablation) | GaitPart · Gait3D | Rank-1: 36.2 -> 29.2 | Hurt |
| add the local branch (ablation) | GaitGL · Gait3D | Rank-1: 32.2 -> 31.4 | Hurt |
| add the SMPL branch with SLN silhouette backbone (ablation) | SMPLGait · Gait3D | Rank-1: 42.9 -> 46.3 | Improved |
| add the SMPL branch with strong ResNet9 backbone (ablation) | SMPLGait · Gait3D | Rank-1: 56.5 -> 55.2 | No clear effect |
| deepen network from 10 to 22 layers (ablation) | DeepGaitV2 · Gait3D | Rank-1: 66.8 -> 72.8 | Improved |
| deepen network from 10 to 22 layers (ablation) | DeepGaitV2 · CASIA-B | Rank-1: 89.6 -> 75.3 | Hurt |
| deepen network from 10 to 30 layers (ablation) | DeepGaitV2 · OU-MVLP | Rank-1: 92.0 -> 91.2 | Hurt |
| widen channels from 64 to 128 (ablation) | DeepGaitV2 · Gait3D | Rank-1: 74.4 -> 75.0 | No clear effect |
| widen channels from 32 to 64 (ablation) | DeepGaitV2 · Gait3D | Rank-1: 67.9 -> 74.4 | Improved |
| use pseudo-3D blocks instead of full 3D (ablation) | DeepGaitV2 · Gait3D and GREW | Rank-1: +1.6% on Gait3D, -1.7% on GREW | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- QAGait — QAGait: Revisit Gait Recognition from a Quality Perspective (AAAI 2024) · better on 1 of 1 shared result
- GaitSet — GaitSet: Regarding Gait as a Set for Cross-View Gait Recognition (AAAI 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OpenGait: Revisiting Gait Recognition Towards Better Practicality (CVPR 2023 · similarity 0.97)
- MMGait: Towards Multi-Modal Gait Recognition (CVPR 2026 · similarity 0.93)
- BigGait: Learning Gait Representation You Want by Large Vision Models (CVPR 2024 · similarity 0.91)
- SkeletonGait: Gait Recognition Using Skeleton Maps (AAAI 2024 · similarity 0.90)
- BiggerGait: Unlocking Gait Recognition with Layer-wise Representations from Large Vision Models (NeurIPS 2025 · similarity 0.90)
- RA-GAR: A Richly Annotated Benchmark for Gait Attribute Recognition (AAAI 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
