Research map / Label-efficient and robust learning / Recognition and long-tail
An Open-World, Diverse, Cross-Spatial-Temporal Benchmark for Dynamic Wild Person Re-Identification
Lei Zhang, Xiaowei Fu, Fuxiang Huang, Yi Yang, Xinbo Gao
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 nighttime data to OWD (ablation) | OWD as source data · Direct transfer to MSMT17 | Rank-1: 20.5 -> 32.5 | Improved |
| train LDE with OWD added to source data | LDE · DG ReID small-scale targets average | mAP: 75.2 -> 77.0 | Improved |
| train LDE with OWD added to source data | LDE · Market1501 target (large-scale) | mAP: 52.8 -> 58.5 | Improved |
| train LDE with OWD added to source data | LDE · Dukemtmc target (large-scale) | mAP: 51.5 -> 56.9 | Improved |
| pre-train on OWD instead of ImageNet | ReID model · Market1501 | mAP: 71.1 -> 88.4 | Improved |
| pre-train on OWD instead of LaST_Cloth | ReID model · Market1501 | mAP: 81.7 -> 88.4 | Improved |
| add OWD instead of MSMT17 to target training data | ReID model · Market1501 combined training | mAP: 77.4 -> 84.5 | Improved |
| use OWD instead of MSMT17 as source for cloth-changing transfer | ReID model · DeepChange | Rank-1: 18.3 -> 24.9 | Improved |
| use OWD instead of ClonedPerson as source with QAConv 2.0 | QAConv 2.0 · Market1501 transfer | mAP: 59.9 -> 68.3 | Improved |
| move evaluation from close-scene to open-scene to day-night protocol | SBS · OWD | mAP: 70.1 -> 59.6 -> 45.5 | Hurt |
| use LDE instead of MetaBIN | LDE · OWD multi-source DG | mAP: 38.5 -> 43.5 | Improved |
| use LDE instead of OSNet-AIN | LDE · Market1501 target (large-scale) | mAP: 45.8 -> 52.8 | Improved |
Closest papers by meaning
- When Person Re-Identification Meets Event Camera: A Benchmark Dataset and an Attribute-Guided Re-Identification Framework (AAAI 2026 · similarity 0.90)
- Person Re-Identification in the Wild (CVPR 2017 · similarity 0.89)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.88)
- From Laboratory to Real World: A New Benchmark Towards Privacy-Preserved Visible-Infrared Person Re-Identification (CVPR 2025 · similarity 0.87)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · similarity 0.86)
- Exploring the Camera Bias of Person Re-identification (ICLR 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.
