Research map / Video / Video segmentation and tracking
MetaGait: Learning to Learn an Omni Sample Adaptive Representation for Gait Recognition
Huanzhang Dou, Pengyi Zhang, Wei Su, Yunlong Yu, Xi Li
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 Meta Triple Attention (ablation) | Baseline · CASIA-B (64x44, LT) | mean rank-1: 89.0 -> 92.3 | Improved |
| add Meta Temporal Pooling (ablation) | Baseline · CASIA-B (64x44, LT) | mean rank-1: 89.0 -> 91.5 | Improved |
| add both MTA and MTP (ablation) | Baseline · CASIA-B (64x44, LT) | mean rank-1: 89.0 -> 93.4 | Improved |
| add MTP on top of MTA, CL condition (ablation) | Baseline + MTA · CASIA-B (64x44, LT) | CL rank-1: 85.4 -> 86.9 | Improved |
| use local receptive fields {1,3,5} instead of {1} (ablation) | MTA · CASIA-B (64x44, LT) | mean rank-1: 92.2 -> 93.4 | Improved |
| enlarge local receptive fields from {1,3,5} to {1,3,5,7} (ablation) | MTA · CASIA-B (64x44, LT) | mean rank-1: 93.4 -> 92.4 | Hurt |
| enlarge local receptive fields to {1,3,5,7,9} (ablation) | MTA · CASIA-B (64x44, LT) | mean rank-1: 93.4 -> 91.9 | Hurt |
| parameterize calibration network with meta-knowledge instead of static (ablation) | MTA · CASIA-B (64x44, LT) | mean rank-1: 92.9 -> 93.4 | Improved |
| add soft aggregation gate with meta calibration (ablation) | MTA · CASIA-B (64x44, LT) | mean rank-1: 93.1 -> 93.4 | Improved |
| add attention on spatial, channel and temporal dimensions together instead of spatial only (ablation) | MTA (static calibration) · CASIA-B (64x44, LT) | mean rank-1: 92.0 -> 92.5 | Improved |
| combine Max, Mean and GeM pooling with static weights instead of single pooling (ablation) | MTP · CASIA-B (64x44, LT) | mean rank-1: 92.4 -> 92.9 | Improved |
| use meta weighting network instead of static weights in temporal pooling (ablation) | MTP · CASIA-B (64x44, LT) | mean rank-1: 92.9 -> 93.4 | Improved |
Later papers that compared with this method
2 later accepted papers report MetaGait as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Hierarchical Spatio-Temporal Representation Learning for Gait Recognition (ICCV 2023 · better on 3 of 4 shared results)
- DyGait: Exploiting Dynamic Representations for High-performance Gait Recognition (ICCV 2023 · 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
- OpenGait: A Comprehensive Benchmark Study for Gait Recognition Toward Better Practicality (TPAMI 2025 · similarity 0.88)
- Learning Gait Representation From Massive Unlabelled Walking Videos: A Benchmark (TPAMI 2023 · similarity 0.88)
- BigGait: Learning Gait Representation You Want by Large Vision Models (CVPR 2024 · similarity 0.87)
- BiggerGait: Unlocking Gait Recognition with Layer-wise Representations from Large Vision Models (NeurIPS 2025 · similarity 0.86)
- Learning Visual Prompt for Gait Recognition (CVPR 2024 · similarity 0.86)
- On Denoising Walking Videos for Gait Recognition (CVPR 2025 · similarity 0.86)
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.
