Research map / Video / Action recognition
GaitSnippet: Gait Recognition Beyond Unordered Sets and Ordered Sequences
Saihui Hou, Chenye Wang, Wenpeng Lang, Zhengxiang Lan, Yongzhen Huang
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 |
|---|---|---|---|
| Snippet Sampling instead of set sampling (ablation) | DeepGaitV2-2D · Gait3D | rank-1: 68.2 -> 69.5 | Improved |
| sequence-based continuous sampling instead of set sampling (ablation) | DeepGaitV2-2D · Gait3D | rank-1: 68.2 -> 66.0 | Hurt |
| add Gathering (intra-snippet pooling and snippet-level supervision) (ablation) | GaitSnippet · Gait3D | rank-1: 73.3 -> 77.5 | Improved |
| add smoothing 1x1 conv layer after Gathering (ablation) | GaitSnippet · Gait3D | rank-1: 74.8 -> 77.5 | Improved |
| add residual merge of snippet context with frame features (ablation) | GaitSnippet · Gait3D | rank-1: 72.5 -> 77.5 | Improved |
| add snippet-level supervision (alpha 0 to 0.75) (ablation) | GaitSnippet · Gait3D | rank-1: 75.8 -> 77.5 | Improved |
| share weights between sequence-level and snippet-level branches (ablation) | GaitSnippet · Gait3D | rank-1: 77.5 -> 75.5 | Hurt |
| segment length L=16 versus 8 or 32 (ablation) | GaitSnippet · Gait3D | rank-1: 76.4 (L=8), 74.7 (L=32) -> 77.5 | Improved |
| N=8 frames per snippet versus 4 or 16 (ablation) | GaitSnippet · Gait3D | rank-1: 74.3 (N=4), 75.2 (N=16) -> 77.5 | Improved |
| ensemble of multiple segment partitions at evaluation (L1 in {8,16}) (ablation) | GaitSnippet · Gait3D | mAP: 69.4 -> 69.8 | No clear effect |
| evaluate with snippet-level instead of sequence-level features (ablation) | GaitSnippet · Gait3D | rank-1: 77.5 -> 65.2 | Hurt |
| add snippet sampling and Snippet Block to GaitSet (ablation) | GaitSet · Gait3D | rank-1: 36.7 -> 48.2 | Improved |
Closest papers by meaning
- Learning Visual Prompt for Gait Recognition (CVPR 2024 · similarity 0.89)
- Vocabulary-Guided Gait Recognition (NeurIPS 2025 · similarity 0.89)
- Bridging Gait Recognition and Large Language Models Sequence Modeling (CVPR 2025 · similarity 0.89)
- On Denoising Walking Videos for Gait Recognition (CVPR 2025 · similarity 0.88)
- BiggerGait: Unlocking Gait Recognition with Layer-wise Representations from Large Vision Models (NeurIPS 2025 · similarity 0.88)
- Exploring More from Multiple Gait Modalities for Human Identification (AAAI 2025 · similarity 0.88)
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.
