Research map / Detection and segmentation / CNN features
End-to-End Thorough Body Perception for Person Search
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 |
|---|---|---|---|
| replace heavy head (conv4_4-conv5_3 + GAP) with light head of four conv layers and softmax+circular queue loss (ablation) | OIM baseline · CUHK-SYSU | Rank-1 (%): 78.7 -> 83.4 | Helped |
| light head instead of heavy head (ablation) | OIM baseline · PRW | Rank-1 (%): 42.7 -> 55.9 | Helped |
| light head instead of heavy head (ablation) | OIM baseline · CUHK-SYSU | Runtime (s): 0.30 -> 0.15 | Helped |
| add FPN to backbone (ablation) | light base model · PRW | Rank-1 (%): 56.3 -> 59.6 | Helped |
| add FPN to backbone (ablation) | light base model · CUHK-SYSU | Rank-1 (%): 84.1 -> 85.0 | Helped |
| add mask branch and FFEM foreground feature enhancement (ablation) | FPN base model · CUHK-SYSU | Rank-1 (%): 85.0 -> 86.8 | Helped |
| add mask branch and FFEM foreground feature enhancement (ablation) | FPN base model · PRW | Rank-1 (%): 59.6 -> 64.0 | Helped |
| add keypoint-guided part-aligned feature learning algorithm (ablation) | BPNet with FFEM · CUHK-SYSU | Rank-1 (%): 86.8 -> 88.7 | Helped |
| add keypoint-guided part-aligned feature learning algorithm (ablation) | BPNet with FFEM · PRW | Rank-1 (%): 64.0 -> 66.7 | Helped |
| add multi-scale training augmentation (ablation) | BPNet · PRW | Rank-1 (%): 66.7 -> 68.9 | Helped |
| add head-shoulder region as third part (P=3) (ablation) | part-aligned algorithm · CUHK-SYSU and PRW | Rank-1 / mAP | Hurt |
| only increase final feature dimension without part algorithm (ablation) | base model · CUHK-SYSU and PRW | Rank-1 / mAP | Hurt |
Closest papers by meaning
- Diverse Knowledge Distillation for End-to-End Person Search (AAAI 2021 · similarity 0.89)
- Sequential End-to-end Network for Efficient Person Search (AAAI 2021 · similarity 0.87)
- Divide and Conquer: Hybrid Pre-training for Person Search (AAAI 2024 · similarity 0.87)
- Joint Detection and Identification Feature Learning for Person Search (CVPR 2017 · similarity 0.87)
- Hierarchical Online Instance Matching for Person Search (AAAI 2020 · similarity 0.86)
- Prototype-Guided Saliency Feature Learning for Person Search (CVPR 2021 · similarity 0.86)
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.