Research map / Label-efficient and robust learning / Domain adaptation and generalization
Unsupervised Domain Adaptation for Person Re-identification via Heterogeneous Graph Alignment
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 fine-grained graph alignment to coarse-grained alignment (ablation) | HGA · Duke→Market and Market→Duke | mAP: 31.5 -> 70.3; 29.2 -> 67.1 | Helped |
| Use adversarial coarse alignment rather than Same, Random, or CAMEL projection initialization (ablation) | HGA · Duke→Market and Market→Duke | mAP: 25.2 -> 31.5; 20.8 -> 29.2 | Helped |
| Add node alignment loss (ablation) | HGA after coarse alignment · Duke→Market and Market→Duke | mAP: 31.5 -> 55.4; 29.2 -> 49.8 | Helped |
| Add edge alignment loss (ablation) | HGA after coarse alignment · Duke→Market and Market→Duke | mAP: 31.5 -> 59.1; 29.2 -> 54.7 | Helped |
| Combine node and edge alignment losses (ablation) | HGA after coarse alignment · Duke→Market and Market→Duke | mAP: 55.4 -> 63.5; 59.1 -> 63.5; 49.8 -> 61.4; 54.7 -> 61.4 | Helped |
| Add projection invariant loss (ablation) | HGA with node and edge alignment · Duke→Market and Market→Duke | mAP: 63.5 -> 69.5; 61.4 -> 66.2 | Helped |
| Add projection constraint loss (ablation) | HGA with coarse and fine alignment · Duke→Market and Market→Duke | mAP: 69.5 -> 70.3; 66.2 -> 67.1 | Helped |
| Use the proposed adversarial coarse alignment instead of CAMEL initialization, with fine alignment (ablation) | HGA · Duke→Market and Market→Duke | mAP: 42.3 -> 70.3; 39.3 -> 67.1 | Helped |
| Use heterogeneous graph alignment for domain-adapted re-identification | HGA · Market-1501→MSMT17 | mAP: 22.1 -> 25.5 | Helped |
| Use heterogeneous graph alignment for domain-adapted re-identification | HGA · DukeMTMC-reID→MSMT17 | Rank-1: 48.8 -> 58.6 | Helped |
| Use heterogeneous graph alignment for domain-adapted re-identification | HGA · DukeMTMC-reID→Market-1501 | mAP: 71.7 vs 70.3 | Mixed |
Closest papers by meaning
- Unsupervised Multi-Source Domain Adaptation for Person Re-Identification (CVPR 2021 · similarity 0.91)
- Dual Distribution Alignment Network for Generalizable Person Re-Identification (AAAI 2021 · similarity 0.90)
- Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification (ICLR 2020 · similarity 0.90)
- Image-Image Domain Adaptation With Preserved Self-Similarity and Domain-Dissimilarity for Person Re-Identification (CVPR 2018 · similarity 0.89)
- Distilled Person Re-Identification: Towards a More Scalable System (CVPR 2019 · similarity 0.89)
- Generalizable Person Re-identification via Balancing Alignment and Uniformity (NeurIPS 2024 · similarity 0.89)
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.