Research map / Detection and segmentation / CNN features
TCTS: A Task-Consistent Two-Stage Framework for Person Search
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 task-consistent IDGQ detection and DRA re-identification | TCTS person search · CUHK-SYSU | mAP: 93.0 -> 93.9 | Helped |
| Use task-consistent IDGQ detection and DRA re-identification | TCTS person search · PRW | rank-1 accuracy: 70.2 -> 87.5 | Helped |
| Replace Faster R-CNN with the query-guided IDGQ detector (ablation) | Person search on CUHK-SYSU · CUHK-SYSU | query recall: 96.9 -> 98.2 | Helped |
| Replace Faster R-CNN with the query-guided IDGQ detector (ablation) | Person search on CUHK-SYSU · CUHK-SYSU | mAP: 91.4 -> 93.9 | Helped |
| Replace IDGQ loss with OIM loss (ablation) | IDGQ detector · CUHK-SYSU | mAP: 93.9 -> 92.0 | Hurt |
| Train the re-ID model on detected boxes only (ablation) | DRA re-ID model · CUHK-SYSU | mAP: 93.1 -> 82.9 | Hurt |
| Use mixed hand-drawn and detected boxes without example reweighting (ablation) | DRA re-ID model · CUHK-SYSU | mAP: 93.1 -> 86.9 | Hurt |
| Add DRA-ER example reweighting to mixed-box training (ablation) | DRA re-ID model · CUHK-SYSU | mAP: 86.9 -> 93.9 | Helped |
| Replace the hardness factor with focal-loss weighting (ablation) | DRA-ER re-ID training · CUHK-SYSU | search performance | Hurt |
Closest papers by meaning
- Diverse Knowledge Distillation for End-to-End Person Search (AAAI 2021 · similarity 0.88)
- Divide and Conquer: Hybrid Pre-training for Person Search (AAAI 2024 · similarity 0.87)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · similarity 0.84)
- Human-centered Interactive Learning via MLLMs for Text-to-Image Person Re-identification (CVPR 2025 · similarity 0.84)
- Doubly Contrastive Learning for Source-Free Domain Adaptive Person Search (AAAI 2025 · similarity 0.84)
- Mixture of Submodules for Domain Adaptive Person Search (CVPR 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.