Research map / Label-efficient and robust learning / Domain adaptation and generalization
Bridging the Source-to-Target Gap for Cross-Domain Person Re-identification with Intermediate Domains
Yongxing Dai, Yifan Sun, Jun Liu, Zekun Tong, Ling‐Yu Duan
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 the diversity loss L_div to IDM++ (ablation) | Baseline2 + IDM++ · Market-1501->DukeMTMC-reID | Rank-1: 1.1% lower than the full method without it | Improved |
| add the bridge loss L_bridge^phi to IDM++ (ablation) | Baseline2 + IDM++ · Market-1501->DukeMTMC-reID | mAP/Rank-1: 3.2%/2.5% lower without it | Improved |
| add the bridge loss L_bridge^f to IDM++ (ablation) | Baseline2 + IDM++ · Market-1501->DukeMTMC-reID | mAP/Rank-1: 1.0%/1.6% lower without it | Improved |
| add MGM (domain mapping with consistency loss) on top of IDM (ablation) | Baseline2 + IDM · Market-1501->DukeMTMC-reID | mAP/Rank-1: 70.5 -> 73.2 mAP; 2.7%/1.9% higher | Improved |
| add IDM++ to the strong baseline | Baseline2 · Market-1501->DukeMTMC-reID | mAP/Rank-1: 7.4%/5.4% higher | Improved |
| add IDM++ to the naive baseline | Baseline1 · Market-1501->DukeMTMC-reID | mAP/Rank-1: 8.2% and 5.0% higher | Improved |
| add MGM to IDM (ablation) | IDM · Market-1501->MSMT17 | mAP/Rank-1: up to 6.7% mAP and 8.6% Rank-1 | Improved |
| use IDM++ instead of prior best CCL | IDM++ vs CCL · Market-1501->MSMT17 | mAP: 4.4% mAP | Improved |
| use IDM++ instead of CCL | IDM++ vs CCL · DukeMTMC-reID->MSMT17 | mAP: 36.3 -> 40.5 | Improved |
| add XBM memory-bank hard negative mining (Strong vs Naive baseline) (ablation) | Baseline1 · Market-1501->DukeMTMC-reID | mAP: 63.4 -> 65.8 | Improved |
| add IDM++ over IDM with synthetic source Unreal | IDM · Unreal->MSMT17 | mAP: 38.3 -> 41.5 | Improved |
Closest papers by meaning
- IDM: An Intermediate Domain Module for Domain Adaptive Person Re-ID (ICCV 2021 · similarity 0.94)
- Unsupervised Multi-Source Domain Adaptation for Person Re-Identification (CVPR 2021 · similarity 0.90)
- Adaptive Cross-Domain Learning for Generalizable Person Re-identification (ECCV 2022 · similarity 0.90)
- Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification (ECCV 2020 · similarity 0.90)
- Learning to Adapt Across Dual Discrepancy for Cross-Domain Person Re-Identification (TPAMI 2022 · similarity 0.89)
- Smoothing Adversarial Domain Attack and P-Memory Reconsolidation for Cross-Domain Person Re-Identification (CVPR 2020 · 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.
