Research map / Label-efficient and robust learning / Self-supervised and contrastive
Learning Gait Representation From Massive Unlabelled Walking Videos: A Benchmark
Chao Fan, Saihui Hou, Jilong Wang, 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 spatial augmentation (ablation) | GaitSSB · OU-MVLP (unsupervised) | rank-1 accuracy (%): 26.7 -> 37.2 | Improved |
| add spatial augmentation (ablation) | GaitSSB · Gait3D (unsupervised) | rank-1 accuracy (%): 18.5 -> 24.7 | Improved |
| add spatial augmentation (ablation) | GaitSSB · GREW (unsupervised) | rank-1 accuracy (%): 10.9 -> 16.6 | Improved |
| add intra-sequence and sampling augmentation on top of spatial augmentation (ablation) | GaitSSB · OU-MVLP (unsupervised) | rank-1 accuracy (%): 0.1 -> 37.2 | Improved |
| add intra-sequence and sampling augmentation on top of spatial augmentation (ablation) | GaitSSB · CASIA-B* NM (unsupervised) | rank-1 accuracy (%): 6.9 -> 75.0 | Improved |
| add sampling augmentation (ablation) | GaitSSB · CASIA-B* NM (unsupervised) | rank-1 accuracy (%): 67.5 -> 75.0 | Improved |
| add sampling augmentation (ablation) | GaitSSB · OU-MVLP (unsupervised) | rank-1 accuracy (%): 33.5 -> 37.2 | Improved |
| add sampling augmentation (ablation) | GaitSSB · GREW (unsupervised) | rank-1 accuracy (%): 17.0 -> 16.6 | Hurt |
| add sampling augmentation (ablation) | GaitSSB · Gait3D (unsupervised) | rank-1 accuracy (%): 24.2 -> 24.7 | Improved |
| add negative samples (InfoNCE instead of cosine loss) (ablation) | GaitSSB · CASIA-B* NM (unsupervised) | rank-1 accuracy (%): 43.4 -> 75.0 | Improved |
| add negative samples (InfoNCE instead of cosine loss) (ablation) | GaitSSB · OU-MVLP (unsupervised) | rank-1 accuracy (%): 14.6 -> 37.2 | Improved |
| add negative samples (InfoNCE instead of cosine loss) (ablation) | GaitSSB · GREW (unsupervised) | rank-1 accuracy (%): 8.8 -> 16.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GaitSet — GaitSet: Regarding Gait as a Set for Cross-View Gait Recognition (AAAI 2019) · better on 3 of 3 shared results
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
- BiggerGait: Unlocking Gait Recognition with Layer-wise Representations from Large Vision Models (NeurIPS 2025 · similarity 0.91)
- On Denoising Walking Videos for Gait Recognition (CVPR 2025 · similarity 0.90)
- Learning Visual Prompt for Gait Recognition (CVPR 2024 · similarity 0.90)
- Vocabulary-Guided Gait Recognition (NeurIPS 2025 · similarity 0.90)
- RA-GAR: A Richly Annotated Benchmark for Gait Attribute Recognition (AAAI 2025 · similarity 0.90)
- Cross-Covariate Gait Recognition: A Benchmark (AAAI 2024 · similarity 0.90)
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.
