Research map / 3D vision / Human pose estimation
Reliability-Aware 3D Geometric Injection for Universal Person Re-identification
Bohan Su, Jiashuo Wang, Fangyi Liu, Mang Ye
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 reliability gate to 3D fusion (vs naive 3D concat) (ablation) | UniGeo · PRCC | Rank-1: 55.0 -> 59.0 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · PRCC | Rank-1: 56.0 -> 59.0 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · Occluded-Duke | Rank-1: 73.9 -> 74.8 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · SYSU-MM01 | Rank-1: 63.1 -> 64.3 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · Celeb-ReID | Rank-1: 60.0 -> 60.4 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · Market-1501 | Rank-1: 96.5 -> 96.6 | No clear effect |
| add reliability-gated 3D injection | Pure 2D ReID baseline · CUHK03 | Rank-1: 96.8 -> 96.6 | Hurt |
| add reliability-gated 3D injection | Pure 2D ReID baseline · MSMT17 | mAP: 71.3 -> 71.6 | Improved |
| add reliability-gated 3D injection | Pure 2D ReID baseline · AG-ReID.v2 A->C | Rank-1: 92.3 -> 92.6 | Improved |
| add naive 3D concatenation (ablation) | Pure 2D ReID baseline · PRCC | Rank-1: 56.0 -> 55.0 | Hurt |
| add kinematic decoupling of SMPL (full topology) vs black-box SMPL vector (ablation) | UniGeo · PRCC | Rank-1: 51.3 -> 59.0 | Improved |
| use full topology instead of global parameters only (ablation) | UniGeo · PRCC | Rank-1: 57.2 -> 59.0 | Improved |
Closest papers by meaning
- Learning Clothing and Pose Invariant 3D Shape Representation for Long-Term Person Re-Identification (ICCV 2023 · similarity 0.88)
- Learning 3D Shape Feature for Texture-Insensitive Person Re-Identification (CVPR 2021 · similarity 0.88)
- From Poses to Identity: Training-Free Person Re-Identification via Feature Centralization (CVPR 2025 · similarity 0.87)
- Camera-Incremental Person Re-Identification Via Geometric Representation Evolution (TPAMI 2026 · similarity 0.87)
- Distilling CLIP with Dual Guidance for Learning Discriminative Human Body Shape Representation (CVPR 2024 · similarity 0.86)
- 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.
