Research map / Label-efficient and robust learning / Recognition and long-tail
GaitSet: Regarding Gait as a Set for Cross-View Gait Recognition
Hanqing Chao, Yiwei He, Junping Zhang, Jianfeng Feng
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 |
|---|---|---|---|
| Represent gait as a set of silhouettes instead of a GEI (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 80.4 -> 91.3 (NM); 68.1 -> 82.3 (BG); 40.8 -> 67.1 (CL) | Improved |
| Use independent rather than shared HPM weights (ablation) | GaitSet with max pooling · CASIA-B LT | rank-1 accuracy: 91.3 -> 93.2 (NM); 82.3 -> 84.7 (BG); 67.1 -> (70.2) (CL) | Improved |
| Replace max pooling with mean pooling (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 91.3 -> 90.0 (NM); 82.3 -> 79.5 (BG); 67.1 -> 57.1 (CL) | Hurt |
| Replace max pooling with median pooling (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 91.3 -> 89.5 (NM); 82.3 -> 78.1 (BG); 67.1 -> 53.5 (CL) | Hurt |
| Replace max pooling with joint sum pooling (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 91.3 -> 92.4 (NM); 82.3 -> 82.8 (BG); 67.1 -> 63.4 (CL) | Hurt |
| Replace max pooling with joint 1_1C pooling (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 91.3 -> 93.3 (NM); 82.3 -> (85.7) (BG); 67.1 -> 66.3 (CL) | Mixed |
| Replace max pooling with attention pooling (ablation) | GaitSet · CASIA-B LT | rank-1 accuracy: 91.3 -> (93.7) (NM); 82.3 -> 84.2 (BG); 67.1 -> 69.4 (CL) | Mixed |
| Remove MGP (ablation) | GaitSet with max pooling and independent HPM weights · CASIA-B LT | rank-1 accuracy: 95.0 -> 93.2 (NM); 87.2 -> 84.7 (BG); 70.4 -> (70.2) (CL) | Hurt |
| Use silhouettes from two views rather than one, with 10 silhouettes total | GaitSet probe input · CASIA-B LT multi-view | rank-1 accuracy: 87.7 -> 87.9 (18°/162°); 87.7 -> 90.6 (36°/144°); 87.7 -> 92.7 (54°/126°); 87.7 -> 93.7 (72°/108°); 87.7 -> 93.7 (90°) | Improved |
| Combine silhouettes from BG and CL walking conditions | GaitSet probe input · CASIA-B LT multiple walking conditions | rank-1 accuracy: 77.1 -> 84.6 | Improved |
| Replace some NM silhouettes with silhouettes from BG or CL | GaitSet probe input · CASIA-B LT multiple walking conditions | rank-1 accuracy: 89.8 -> 87.9; 89.8 -> 85.8 | Hurt |
| Increase the number of silhouettes per set | GaitSet · CASIA-B LT limited silhouettes | rank-1 accuracy | Improved |
Later papers that compared with this method
5 later accepted papers report GaitSet as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Walking Further: Semantic-Aware Multimodal Gait Recognition Under Long-Range Conditions (AAAI 2026 · better on 2 of 2 shared results)
- Gait Recognition in the Wild: A Large-Scale Benchmark and NAS-Based Baseline (TPAMI 2025 · better on 2 of 2 shared results)
- OpenGait: A Comprehensive Benchmark Study for Gait Recognition Toward Better Practicality (TPAMI 2025 · better on 1 of 1 shared result)
- Learning Gait Representation From Massive Unlabelled Walking Videos: A Benchmark (TPAMI 2023 · better on 3 of 3 shared results)
- Fine-grained Unsupervised Domain Adaptation for Gait Recognition (ICCV 2023 · better on 2 of 2 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
- GaitSet: Cross-view Gait Recognition through Utilizing Gait as a Deep Set (TPAMI 2021 · similarity 0.97)
- CASIA-E: A Large Comprehensive Dataset for Gait Recognition (TPAMI 2022 · similarity 0.92)
- Learning Gait Representation From Massive Unlabelled Walking Videos: A Benchmark (TPAMI 2023 · similarity 0.90)
- DyGait: Exploiting Dynamic Representations for High-performance Gait Recognition (ICCV 2023 · similarity 0.90)
- Cross-Covariate Gait Recognition: A Benchmark (AAAI 2024 · similarity 0.87)
- Exploring More from Multiple Gait Modalities for Human Identification (AAAI 2025 · similarity 0.87)
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.
