Research map / Label-efficient and robust learning / Recognition and long-tail
Benchmarks for Corruption Invariant Person Re-identification
Minghui Chen, Zhiqiang Wang, Feng Zheng
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 |
|---|---|---|---|
| use CIL baseline instead of AGW | ResNet-50 ReID · Market-1501-C corrupted | mAP: 12.13 -> 28.03 | Improved |
| use CIL instead of SBS on clean data | ResNet-50 ReID · Market-1501 clean | mAP: 88.33 -> 84.04 | Hurt |
| use inference features before BNNeck (ablation) | ResNet-50 · Market-1501-C corrupted | mAP: 15.26 -> 18.16 | Improved |
| use inference features before BNNeck (ablation) | ResNet-50 · Market-1501 clean | mAP: 85.52 -> 79.58 | Hurt |
| add soft random erasing (ablation) | ResNet-50 (pre-BNNeck) · Market-1501-C corrupted | mAP: 18.16 -> 25.98 | Improved |
| add self patch mixing (ablation) | ResNet-50 · Market-1501-C corrupted | mAP: 25.98 -> 27.59 | Improved |
| add consistent ID loss (ablation) | ResNet-50 · Market-1501-C corrupted | mAP: 27.59 -> 28.03 | Improved |
| soft random erasing instead of random erasing with AugMix (ablation) | AugMix ResNet-50 · Market-1501-C corrupted | mAP: 24.32 -> 26.66 | Improved |
| self patch mixing instead of random patch mixing with AugMix (ablation) | AugMix ResNet-50 · Market-1501-C corrupted | mAP: 22.05 -> 22.72 | Improved |
| add AugMix to standard augmentation | ResNet-50 · Market-1501-C corrupted | mAP: 14.31 -> 22.47 | Improved |
| add random affine transformation | ResNet-50 · Market-1501-C corrupted | mAP: 14.31 -> 8.01 | Hurt |
| use transformer (Trans-ViT-B) instead of ResNet-50 | ReID backbone · Market-1501-C corrupted | mAP: 8.50 -> 27.10 | 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.
- AugMix — AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty (ICLR 2020) · better on 1 of 1 shared result
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
- Towards Robustness of Person Search against Corruptions (ICCV 2025 · similarity 0.89)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.88)
- Robust Person Re-Identification by Modelling Feature Uncertainty (ICCV 2019 · similarity 0.88)
- Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking (CVPR 2020 · similarity 0.87)
- Adversarially Occluded Samples for Person Re-Identification (CVPR 2018 · similarity 0.87)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · 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.
