Research map / Label-efficient and robust learning / Recognition and long-tail
Uncertainty Modeling of Contextual-Connections Between Tracklets for Unconstrained Video-Based Face Recognition
Jingxiao Zheng, Ruichi Yu, Jun-Cheng Chen, Boyu Lu, Carlos D. Castillo, Rama Chellappa
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 positive gates (ablation) | UGG on CSM (avg) · CSM ACROSS | R@1 (%): 53.15% -> 66.02% | Improved |
| add positive gates (ablation) | UGG on IJB-S S2SG · IJB-S S2SG | top-1 accuracy (%): 64.86% -> 71.21% | Improved |
| add adaptive update of positive gates (ablation) | UGG on CSM (avg) · CSM ACROSS | R@1 (%): 66.02% -> 66.73% | Improved |
| add adaptive update of positive gates (ablation) | UGG on CSM (max) · CSM ACROSS | R@1 (%): 64.63% -> 65.72% | Improved |
| add adaptive update of positive gates (ablation) | UGG on IJB-S S2SG · IJB-S S2SG | top-1 accuracy (%): 71.21% -> 72.32% | Improved |
| add cannot-link control to positive gates (ablation) | UGG on IJB-S S2SG · IJB-S S2SG | top-1 accuracy (%): 71.21% -> 71.26% | No clear effect |
| add negative gates (ablation) | UGG with PGcl on IJB-S S2SG · IJB-S S2SG | top-1 accuracy (%): 71.26% -> 73.24% | Improved |
| add adaptive gates, positive, negative and cannot-link gates together (ablation) | UGG on IJB-S S2B · IJB-S S2B | top-1 accuracy (%): 73.78% -> 74.79% | Improved |
| train with fixed positive gates under 25% labels (ablation) | UGG semi-supervised · CSM ACROSS | mAP (%): 58.34% -> 58.94% | Improved |
| use UGG at test time after semi-supervised training (ablation) | UGG semi-supervised · CSM ACROSS | mAP (%): 58.34% -> 65.82% | Improved |
| add adaptive gates during semi-supervised training plus UGG test (ablation) | UGG semi-supervised · CSM ACROSS | R@1 (%): 70.74% -> 71.31% | Improved |
| replace PPCC fixed graph with UGG adaptive gates | UGG-U vs PPCC · CSM ACROSS | R@1 (%): 63.26% -> 66.73% | Improved |
Closest papers by meaning
- Uncertain Graph Neural Networks for Facial Action Unit Detection (AAAI 2021 · similarity 0.87)
- Face Association for Videos Using Conditional Random Fields and Max-Margin Markov Networks (TPAMI 2016 · similarity 0.86)
- UGG-ReID: Uncertainty-Guided Graph Model for Multi-Modal Object Re-Identification (NeurIPS 2025 · similarity 0.84)
- Breaking Spurious Correlations: Uncertainty-Driven Causal Transformers for AU Detection (CVPR 2026 · similarity 0.84)
- Data Uncertainty Learning in Face Recognition (CVPR 2020 · similarity 0.84)
- CUPS: Improving Human Pose-Shape Estimators with Conformalized Deep Uncertainty (ICML 2025 · similarity 0.82)
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.
