Research map / Label-efficient and robust learning / Domain adaptation and generalization
Delving into Probabilistic Uncertainty for Unsupervised Domain Adaptive Person Re-identification
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 |
|---|---|---|---|
| add probabilistic uncertainty-guided progressive label refinery (ablation) | MMT baseline · Duke→Market | mAP: 74.5 -> 81.0 | Helped |
| use probabilistic distribution uncertainty rather than l₂ centroid distance (ablation) | MMT baseline on Duke→Market · Duke→Market | mAP: 77.8 -> 81.0 | Helped |
| use an external centroid classifier rather than the internal classifier (ablation) | MMT baseline on Duke→Market · Duke→Market | mAP: 80.2 -> 81.0 | Helped |
| use probabilistic uncertainty rather than prediction consistency (ablation) | MMT baseline on Duke→Market · Duke→Market | mAP: 79.2 -> 81.0 | Helped |
| filter uncertain pseudo-label samples rather than reweighting them (ablation) | MMT baseline on Duke→Market · Duke→Market | mAP: 77.7 -> 81.0 | Helped |
| set ε to 0.95 instead of 0.99 (ablation) | P²LR on Duke→Market · Duke→Market | mAP: 81.0 -> 74.8 | Hurt |
| set ε to 0.97 instead of 0.99 (ablation) | P²LR on Duke→Market · Duke→Market | mAP: 81.0 -> 77.1 | Hurt |
| set ε to 1.00 instead of 0.99 (ablation) | P²LR on Duke→Market · Duke→Market | mAP: 81.0 -> 80.8 | Hurt |
| add P²LR to the MMT baseline | UDA ReID across the four reported transfer settings · Duke→Market; Market→Duke; Market→MSMT; Duke→MSMT | mAP: 9.8%, 5.7%, 6.1%, and 6.6% | Helped |
Closest papers by meaning
- Exploiting Sample Uncertainty for Domain Adaptive Person Re-Identification (AAAI 2021 · similarity 0.95)
- Reliability Exploration with Self-Ensemble Learning for Domain Adaptive Person Re-identification (AAAI 2022 · similarity 0.93)
- Group-aware Label Transfer for Domain Adaptive Person Re-identification (CVPR 2021 · similarity 0.92)
- Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification (ICLR 2020 · similarity 0.91)
- Unsupervised Domain Adaptive Person Search via Dual Self-Calibration (AAAI 2025 · similarity 0.87)
- Exploring the Camera Bias of Person Re-identification (ICLR 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.