Research map / Label-efficient and robust learning / Recognition and long-tail
Robust Anchor Embedding for Unsupervised Video Person Re-Identification in the Wild
Mang Ye, Xiangyuan Lan, Pong C. Yuen
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 top-k counts label prediction (ablation) | RACE · MARS label estimation | precision: 47.84 -> 66.22 | Improved |
| add top-k counts label prediction (ablation) | RACE · MARS label estimation | recall: 47.84 -> 40.87 | Hurt |
| add top-k counts label prediction (ablation) | RACE · MARS label estimation | F-score: 47.84 -> 50.54 | Improved |
| add regularized affine hull (RAH) (ablation) | RACE · MARS label estimation | recall: 37.20 -> 40.87 | Improved |
| add regularized affine hull (RAH) (ablation) | RACE · MARS label estimation | F-score: 48.14 -> 50.54 | Improved |
| add smoothing term (ablation) | RACE · MARS label estimation | F-score: 49.65 -> 50.54 | Improved |
| add smoothing term (ablation) | RACE · MARS label estimation | precision: 59.22 -> 66.22 | Improved |
| use RACE label estimation instead of DGM | RACE · MARS label estimation | label estimation time: about 2 hours -> 183s | Improved |
| train with RACE-estimated labels | IDE ResNet-50 · MARS | rank-1: 33.2 -> 41.0 | Improved |
| train with RACE-estimated labels | IDE ResNet-50 · PRID-2011 (wild) | rank-1: 45.3 -> 50.6 | Improved |
| train with RACE-estimated labels | IDE ResNet-50 · iLIDS-VID (wild) | rank-1: 13.6 -> 19.3 | Improved |
| add a second round of label estimation | RACE · MARS | rank-1: 41.0 -> 43.2 | Improved |
Closest papers by meaning
- Unsupervised Person Re-Identification by Soft Multilabel Learning (CVPR 2019 · similarity 0.87)
- Dynamic Label Graph Matching for Unsupervised Video Re-Identification (ICCV 2017 · similarity 0.86)
- A Pose-Sensitive Embedding for Person Re-Identification With Expanded Cross Neighborhood Re-Ranking (CVPR 2018 · similarity 0.86)
- Exploit the Unknown Gradually: One-Shot Video-Based Person Re-Identification by Stepwise Learning (CVPR 2018 · similarity 0.85)
- Unsupervised Domain Adaptation for Person Re-identification via Heterogeneous Graph Alignment (AAAI 2021 · similarity 0.85)
- Intra-Inter Domain Similarity for Unsupervised Person Re-Identification (TPAMI 2022 · similarity 0.85)
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.
