Research map / Label-efficient and robust learning / Recognition and long-tail
SBSGAN: Suppression of Inter-Domain Background Shift for Person Re-Identification
Yan Huang, Qiang Wu, JingSong Xu, Yi Zhong
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 |
|---|---|---|---|
| replace hard masks (JPPNet) with soft-mask images | IDE · Market-1501 -> DukeMTMC-reID | rank-1: 38.6 -> 43.3 | Improved |
| replace style-transferred images (PTGAN) with soft-mask images | IDE · Market-1501 -> DukeMTMC-reID | rank-1: 42.9 -> 43.3 | Improved |
| use soft-mask images instead of original images | IDE · Market-1501 -> DukeMTMC-reID | mAP: 17.7 -> 22.3 | Improved |
| add style consistency loss L_sc to SBSGAN (ablation) | SBSGAN · Market-1501 -> DukeMTMC-reID | rank-1: 41.7 -> 43.3 | Improved |
| add style consistency loss L_sc to SBSGAN (ablation) | SBSGAN · Market-1501 -> DukeMTMC-reID | mAP: 21.2 -> 22.3 | Improved |
| train SBSGAN on two domains only instead of three (ablation) | SBSGAN · Market-1501 -> DukeMTMC-reID | rank-1: 43.3 -> 44.2 | Improved |
| add SEBlock to baseline 2-stream network (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | rank-1: 50.2 -> 50.5 | Improved |
| add ISDC modules to baseline+SEBlock (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | rank-1: 50.5 -> 53.5 | Improved |
| add ISDC modules to baseline+SEBlock (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | mAP: 28.9 -> 30.8 | Improved |
| add SEBlock to every ISDC module (ISDC-SE) (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | rank-1: 53.5 -> 51.5 | Hurt |
| feed both style-transferred and soft-mask images instead of two style-transferred (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | rank-1: 49.6 -> 53.5 | Improved |
| feed both style-transferred and soft-mask images instead of two soft-mask (ablation) | DA-2S · Market-1501 -> DukeMTMC-reID | mAP: 27.0 -> 30.8 | Improved |
Closest papers by meaning
- Person Transfer GAN to Bridge Domain Gap for Person Re-Identification (CVPR 2018 · similarity 0.89)
- Instance-Guided Context Rendering for Cross-Domain Person Re-Identification (ICCV 2019 · similarity 0.88)
- Image-Image Domain Adaptation With Preserved Self-Similarity and Domain-Dissimilarity for Person Re-Identification (CVPR 2018 · similarity 0.88)
- Recover and Identify: A Generative Dual Model for Cross-Resolution Person Re-Identification (ICCV 2019 · similarity 0.88)
- Learning to Adapt Across Dual Discrepancy for Cross-Domain Person Re-Identification (TPAMI 2022 · similarity 0.87)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 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.
