Research map / Label-efficient and robust learning / Recognition and long-tail
Disentangled Representations for Short-Term and Long-Term Person Re-Identification
Chanho Eom, Won-Kyung Lee, Geon Lee, Bumsub Ham
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 identity-shuffling, part-level shuffling and GAN losses (full IS-GAN_DC) to the baseline (ablation) | IS-GAN_DC · Celeb-reID | rank-1: 48.59 -> 53.43 | Improved |
| add identity-shuffling, part-level shuffling and GAN losses (full IS-GAN_DC) to the baseline (ablation) | IS-GAN_DC · CUHK03-labeled | rank-1: 72.43 -> 77.43 | Improved |
| add identity-shuffling, part-level shuffling and GAN losses (full IS-GAN_KL) to the baseline (ablation) | IS-GAN_KL · Market-1501 | mAP: 85.10 -> 87.72 | Improved |
| add decorrelation loss with moving mean and std (ablation) | IS-GAN_DC · Market-1501 | mAP: 87.8 -> 88.3 | Improved |
| add decorrelation loss with moving mean and std (ablation) | IS-GAN_DC · CUHK03-labeled | rank-1: 74.3 -> 77.4 | Improved |
| use moving mean/std instead of batch-wise mean/std in decorrelation loss (ablation) | IS-GAN_DC · CUHK03-labeled | rank-1: 63.0 -> 77.4 | Improved |
| add decorrelation term on top of KL loss (ablation) | IS-GAN_KL · Market-1501 | mAP: 87.7 -> 87.8 | No clear effect |
| use decorrelation loss (DC) instead of KL loss (ablation) | IS-GAN · CUHK03-labeled | rank-1: 74.93 -> 77.43 | Improved |
| use decorrelation loss (DC) instead of KL loss (ablation) | IS-GAN · Celeb-reID | rank-1: 53.60 -> 53.43 | Hurt |
| add IS-GAN_KL modules on top of PCB (ablation) | PCB · Market-1501 | rank-1: 91.5 -> 93.1 | Improved |
| add part-level shuffling loss (ablation) | PCB + IS-GAN_DC · Market-1501 | rank-1: 93.4 -> 93.9 | Improved |
| add part-level shuffling loss (part-3 branch) (ablation) | IS-GAN_DC · Market-1501 | rank-1: 94.4 -> 95.7 | Improved |
Closest papers by meaning
- Learning Disentangled Representation for Robust Person Re-identification (NeurIPS 2019 · similarity 0.98)
- FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identification (NeurIPS 2018 · similarity 0.92)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.91)
- DIFFER: Disentangling Identity Features via Semantic Cues for Clothes-Changing Person Re-ID (CVPR 2025 · similarity 0.91)
- 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.
