Research map / Label-efficient and robust learning / Semi-supervised learning
Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition
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 |
|---|---|---|---|
| Use multi-label co-regularization with multi-view feature learning and AU relationship modeling instead of the labeled-only baseline | ResNet-34 AU recognition · EmotioNet | Avg. F1 (%): 62.3 -> 68.1 | Helped |
| Use multi-label co-regularization with multi-view feature learning and AU relationship modeling instead of the labeled-only baseline | ResNet-34 AU recognition · BP4D | Avg. F1 (%): 56.5 -> 59.8 | Helped |
| Use the proposed method rather than traditional co-training | Semi-supervised AU recognition · EmotioNet | Avg. F1 (%): 65.5 -> 68.1 | Helped |
| Use the proposed method rather than mean-teacher | Semi-supervised AU recognition · BP4D | Avg. F1 (%): 57.8 -> 59.8 | Helped |
| Add multi-label co-regularization loss Lcr (ablation) | Labeled-only ResNet-34 AU recognizer · EmotioNet | Avg. F1 (%): 62.3 -> 67.1 | Helped |
| Add multi-view loss Lmv to the Lcr model (ablation) | Multi-label co-regularized AU recognizer · EmotioNet | Avg. F1 (%): 67.1 -> 67.5 | Helped |
| Add GCN for AU relationship modeling (ablation) | Multi-label co-regularized AU recognizer · EmotioNet | Avg. F1 (%): 67.5 -> 68.1 | Helped |
| Add GCN for AU relationship modeling (ablation) | EmotioNet-trained AU recognizer · BP4D cross-database test | Avg. F1 (%): 36.0 -> 41.4 | Helped |
| Add GCN for AU relationship modeling (ablation) | EmotioNet-trained AU recognizer · UNBC-McMaster cross-database test | Avg. F1 (%): 55.6 -> 68.8 | Helped |
| Ensemble the two view predictions by averaging probabilities without Lmv (ablation) | Multi-label co-regularized AU recognizer · EmotioNet | Avg. F1 (%): 67.2 | No clear effect |
| Ensemble the two view predictions by averaging probabilities when Lmv is used (ablation) | Multi-label co-regularized AU recognizer · EmotioNet | Avg. F1 (%): 67.5 -> 68.1 | Helped |
Closest papers by meaning
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.91)
- Weakly Supervised Facial Action Unit Recognition Through Adversarial Training (CVPR 2018 · similarity 0.89)
- Action Unit Detection With Region Adaptation, Multi-Labeling Learning and Optimal Temporal Fusing (CVPR 2017 · similarity 0.89)
- Deep Region and Multi-Label Learning for Facial Action Unit Detection (CVPR 2016 · similarity 0.88)
- Knowledge-Driven Self-Supervised Representation Learning for Facial Action Unit Recognition (CVPR 2022 · similarity 0.88)
- Distribution Matching for Multi-Task Learning of Classification Tasks: A Large-Scale Study on Faces & Beyond (AAAI 2024 · 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.