Research map / Label-efficient and robust learning / Recognition and long-tail
Person Transfer GAN to Bridge Domain Gap for Person Re-Identification
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 |
|---|---|---|---|
| Transfer CUHK03 training images to both PRID cameras | GoogLeNet trained on CUHK03 for PRID cam1/cam2 · PRID cam1/cam2 | Rank-1 accuracy: 2.0 -> 37.5 | Helped |
| Transfer CUHK03 training images to both PRID cameras | GoogLeNet trained on CUHK03 for PRID cam2/cam1 · PRID cam2/cam1 | Rank-1 accuracy: 1.5 -> 37.5 | Helped |
| Transfer Market training images to both PRID cameras | GoogLeNet trained on Market for PRID cam1/cam2 · PRID cam1/cam2 | Rank-1 accuracy: 5.0 -> 33.5 | Helped |
| Transfer Market training images to both PRID cameras | GoogLeNet trained on Market for PRID cam2/cam1 · PRID cam2/cam1 | Rank-1 accuracy: 11.0 -> 31.0 | Helped |
| Transfer Duke images to MSMT17 | GoogLeNet trained on Duke · MSMT17 | Rank-1 accuracy: 7.1 -> 11.8 | Helped |
| Transfer Market images to MSMT17 | GoogLeNet trained on Market · MSMT17 | Rank-1 accuracy: 3.4 -> 10.2 | Helped |
| Transfer CUHK03 images to MSMT17 | GoogLeNet trained on CUHK03 · MSMT17 | Rank-1 accuracy: 2.8 -> 6.5 | Helped |
| Transfer Duke images before combining with 10% labeled MSMT17 | GoogLeNet trained with Duke and 10% MSMT17 · MSMT17, 10% training set | Rank-1 accuracy: 1.9% | Helped |
| Transfer Market images before combining with 10% labeled MSMT17 | GoogLeNet trained with Market and 10% MSMT17 · MSMT17, 10% training set | Rank-1 accuracy: 5.1% | Helped |
| Transfer CUHK03 images before combining with 10% labeled MSMT17 | GoogLeNet trained with CUHK03 and 10% MSMT17 · MSMT17, 10% training set | Rank-1 accuracy: 2.4% | Helped |
| Train on transfers to both PRID cameras instead of just one camera | GoogLeNet trained on CUHK03 for PRID cam1/cam2 · PRID cam1/cam2 | Rank-1 accuracy: 18.0 -> 37.5 | Helped |
Closest papers by meaning
- Image-Image Domain Adaptation With Preserved Self-Similarity and Domain-Dissimilarity for Person Re-Identification (CVPR 2018 · similarity 0.89)
- Learning Disentangled Representation for Robust Person Re-identification (NeurIPS 2019 · similarity 0.89)
- Pose Transferrable Person Re-Identification (CVPR 2018 · similarity 0.88)
- FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identification (NeurIPS 2018 · similarity 0.88)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.85)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.85)
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.