Research map / Label-efficient and robust learning / Domain adaptation and generalization
Global Distance-distributions Separation for Unsupervised Person Re-identification
Xin Jin, Cuiling Lan, Wenjun Zeng, Zhibo Chen
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 GDS loss (ablation) | clustering Baseline · Market1501→DukeMTMC-reID | mAP: 48.4 -> 52.9 | Improved |
| add GDS loss (ablation) | clustering Baseline · DukeMTMC-reID→Market1501 | mAP: 52.1 -> 57.1 | Improved |
| add distribution-based hard mining to GDS (ablation) | clustering Baseline · Market1501→DukeMTMC-reID | mAP: 52.9 -> 55.1 | Improved |
| add distribution-based hard mining to GDS (ablation) | clustering Baseline · DukeMTMC-reID→Market1501 | mAP: 57.1 -> 61.2 | Improved |
| add GDS-H loss | SPGAN · Market1501→DukeMTMC-reID | mAP: 22.3 -> 27.5 | Improved |
| add GDS-H loss | SPGAN · DukeMTMC-reID→Market1501 | mAP: 22.8 -> 29.4 | Improved |
| add GDS-H loss | B-SNR · Market1501→DukeMTMC-reID | mAP: 54.3 -> 59.7 | Improved |
| add GDS-H loss | B-SNR · DukeMTMC-reID→Market1501 | mAP: 66.1 -> 72.5 | Improved |
| use momentum coefficient 0.99 instead of 0 (global instead of batch-wise distributions) (ablation) | Baseline+GDS-H · Market1501→DukeMTMC-reID | mAP: 51.4 -> 55.1 | Improved |
| use momentum coefficient 0.99 instead of 0 (global instead of batch-wise distributions) (ablation) | Baseline+GDS-H · DukeMTMC-reID→Market1501 | mAP: 56.2 -> 61.2 | Improved |
| replace Histogram loss with GDS-H loss (ablation) | clustering Baseline · Market1501→DukeMTMC-reID | mAP: 50.2 -> 55.1 | Improved |
| add GDS-H loss with K-means clustering | Baseline · Market1501→DukeMTMC-reID (K-means) | mAP: 40.2 -> 49.0 | Improved |
Closest papers by meaning
- Person30K: A Dual-Meta Generalization Network for Person Re-Identification (CVPR 2021 · similarity 0.89)
- Relation Network for Person Re-Identification (AAAI 2020 · similarity 0.89)
- Style Normalization and Restitution for Generalizable Person Re-Identification (CVPR 2020 · similarity 0.89)
- Joint Visual and Temporal Consistency for Unsupervised Domain Adaptive Person Re-Identification (ECCV 2020 · similarity 0.88)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.88)
- Person De-reidentification: A Variation-guided Identity Shift Modeling (CVPR 2025 · similarity 0.88)
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.
