Research map / Video / Video segmentation and tracking
Person Search in Videos with One Portrait Through Visual and Temporal Links
Qingqiu Huang, Wentao Liu, Dahua Lin
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 |
|---|---|---|---|
| replace linear diffusion label propagation with competitive consensus propagation | label propagation with visual and temporal links · CSM ACROSS | mAP (%): 0.37 -> 59.58 | Improved |
| replace linear diffusion label propagation with competitive consensus propagation | label propagation with visual and temporal links · CSM IN | mAP (%): 8.19 -> 62.37 | Improved |
| add temporal links (tracklet identity invariance) to visual-only propagation (ablation) | PPCC · CSM ACROSS | mAP (%): 59.58 -> 62.27 | Improved |
| add temporal links to visual-only propagation (ablation) | PPCC · CSM IN | mAP (%): 62.37 -> 63.49 | Improved |
| add temporal links to visual-only propagation (ablation) | PPCC · CSM IN | R@1 (%): 84.31 -> 83.44 | Hurt |
| add progressive step freezing to conventional all-node updates (ablation) | PPCC-vt · CSM ACROSS | mAP (%): 57.42 -> 62.27 | Improved |
| add progressive step freezing to conventional all-node updates (ablation) | PPCC-vt · CSM IN | mAP (%): 60.54 -> 63.49 | Improved |
| use step freezing schedule instead of threshold freezing (ablation) | PPCC-vt · CSM ACROSS | mAP (%): 61.20 -> 62.27 | Improved |
| combine IDE feature with face feature (weights 0.8 and 0.2) (ablation) | visual matching · CSM ACROSS | mAP (%): 42.16 -> 40.43 | Hurt |
| combine IDE feature with face feature (ablation) | visual matching · CSM IN | mAP (%): 53.33 -> 53.71 | No clear effect |
| use smaller softmax temperature in competitive consensus (ablation) | PPCC · CSM | mAP | Improved |
| use top-k average (k>1) instead of max over neighbors (ablation) | competitive consensus · CSM | mAP | Hurt |
Closest papers by meaning
- Stepwise Metric Promotion for Unsupervised Video Person Re-Identification (ICCV 2017 · similarity 0.88)
- Tracklet Self-Supervised Learning for Unsupervised Person Re-Identification (AAAI 2020 · similarity 0.88)
- Exploit the Unknown Gradually: One-Shot Video-Based Person Re-Identification by Stepwise Learning (CVPR 2018 · similarity 0.87)
- Unsupervised Tracklet Person Re-Identification (TPAMI 2019 · similarity 0.87)
- Learning Instance-Level Spatial-Temporal Patterns for Person Re-Identification (ICCV 2021 · similarity 0.87)
- Cheb-GR: Rethinking K-nearest Neighbor Search in Re-ranking for Person Re-identification (CVPR 2025 · 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.
