Research map / Label-efficient and robust learning / Recognition and long-tail
Classifier Learning With Prior Probabilities for 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 |
|---|---|---|---|
| add expression-independent and expression-dependent joint AU relationships to single-AU probability constraints | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.679 -> 0.719; MMI: 0.419 -> 0.508; BP4D: 0.515 -> 0.564; ENet: 0.311 -> 0.336 | Helped |
| add expression-independent AU relationships (ablation) | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.686 -> 0.719; MMI: 0.414 -> 0.508; BP4D: 0.493 -> 0.564; ENet: 0.312 -> 0.336 | Helped |
| add expression-dependent AU relationships (ablation) | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.700 -> 0.719; MMI: 0.452 -> 0.508; BP4D: 0.550 -> 0.564; ENet: 0.322 -> 0.336 | Helped |
| combine FACS and Study [5] single-AU probability sources (ablation) | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.680 / 0.708 -> 0.719; MMI: 0.464 / 0.488 -> 0.508; BP4D: 0.344 / 0.554 -> 0.564; ENet: 0.278 / 0.334 -> 0.336 | Helped |
| use landmark coordinates instead of LBP features | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.719 -> 0.732; MMI: 0.508 -> 0.481; BP4D: 0.564 -> 0.563; ENet: 0.336 -> 0.337 | Mixed |
| use logistic loss instead of hinge loss | LP-SM · CK+, MMI, BP4D, ENet | F1 score: CK+: 0.719 -> 0.690; MMI: 0.508 -> 0.481; BP4D: 0.564 -> 0.552; ENet: 0.336 -> 0.351 | Mixed |
| add AU annotations for half of the training samples alongside prior probabilities | LP-SM · CK+ | F1 score: 0.719 -> 0.754 | Helped |
| use generic prior constraints rather than database-specific AU annotations for training | LP-SM · CK+, MMI, BP4D | F1 score: CK+: 0.719 vs 0.783; MMI: 0.508 vs 0.519; BP4D: 0.564 vs 0.667 | Mixed |
Closest papers by meaning
- Knowledge Augmented Deep Neural Networks for Joint Facial Expression and Action Unit Recognition (NeurIPS 2020 · similarity 0.91)
- Weakly Supervised Facial Action Unit Recognition Through Adversarial Training (CVPR 2018 · similarity 0.91)
- Joint Representation and Estimator Learning for Facial Action Unit Intensity Estimation (CVPR 2019 · similarity 0.89)
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.89)
- Facial-R1: Aligning Reasoning and Recognition for Facial Emotion Analysis (AAAI 2026 · similarity 0.86)
- AUEditNet: Dual-Branch Facial Action Unit Intensity Manipulation with Implicit Disentanglement (CVPR 2024 · 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.