Research map / 3D vision / 3D generation
Dissecting Person Re-Identification From the Viewpoint of Viewpoint
Check 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 |
|---|---|---|---|
| Use a due-left rather than due-front query | PCB on PersonX45 · PersonX45 | rank-1 accuracy: 93.8% vs. 91.7% | Helped |
| Train with left+right rather than front+back orientations | PCB · PersonX | mAP: 0.8%-1.1% lower | Helped |
| Train with left+right rather than front+back orientations | PCB · Market-1203 | mAP: 2.2% higher | Helped |
| Train with left or right rather than front or back orientation | PCB · PersonX single-orientation training | mAP: as large as 6% | Helped |
| Remove continuous viewpoints rather than the same amount of randomly selected viewpoints or images (ablation) | PersonX re-ID training · PersonX12 and PersonX46 | mAP | Hurt |
| Remove more continuous viewpoints (ablation) | PCB trained on PersonX46 · PersonX46 | mAP | Hurt |
| Train with only two orientations instead of a balanced viewpoint set (ablation) | PCB · PersonX and Market-1203 | mAP | Hurt |
| Train with only one orientation instead of a balanced viewpoint set (ablation) | PCB · PersonX | mAP | Hurt |
| Use a due-left or due-right query rather than a due-front or due-back query | PCB · PersonX45 | rank-1 accuracy: 93.8% and 91.7% | Helped |
| Use a true match with a viewpoint similar to the query | PCB retrieval · PersonX45 | rank-1 accuracy | Helped |
| Remove 9 true matches with viewpoints similar to the query instead of randomly removing 9 (ablation) | PCB retrieval · PersonX45 | mAP: about 1.0% | Hurt |
| Remove 3 or 5 true matches with viewpoints similar to the query instead of randomly removing the same number (ablation) | PCB retrieval · Market-1203 | mAP: about 3.0% | Hurt |
Closest papers by meaning
- Alice Benchmarks: Connecting Real World Re-Identification with the Synthetic (ICLR 2024 · similarity 0.86)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.86)
- Exploring the Camera Bias of Person Re-identification (ICLR 2025 · similarity 0.85)
- SEAS: ShapE-Aligned Supervision for Person Re-Identification (CVPR 2024 · similarity 0.84)
- Fusing Personal and Environmental Cues for Identification and Segmentation of First-Person Camera Wearers in Third-Person Views (CVPR 2024 · similarity 0.84)
- FlexiReID: Adaptive Mixture of Expert for Multi-Modal Person Re-Identification (ICML 2025 · similarity 0.84)
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.