Research map / Label-efficient and robust learning / Recognition and long-tail
Knowledge Augmented Deep Neural Networks for Joint Facial Expression and 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 joint training with AU knowledge (expression-augmented regularization) (ablation) | image-based FER (VGG-19) · BP4D | accuracy: 61.68 -> 83.82 | Helped |
| add joint training with AU knowledge (ablation) | image-based FER (VGG-19) · CK+ | accuracy: 94.29 -> 97.59 | Helped |
| add joint training with AU knowledge (ablation) | image-based FER (VGG-19) · MMI | accuracy: 67.35 -> 84.90 | Helped |
| add joint training with AU knowledge (ablation) | image-based FER (VGG-19) · EmotioNet | accuracy: 80.85 -> 95.55 | Helped |
| add expression-augmented regularization (model integration) (ablation) | AUD-BN · CK+ | average F1: .69 -> .74 | Helped |
| add expression-augmented regularization (model integration) (ablation) | AUD-BN · MMI | average F1: .47 -> .58 | Helped |
| add expression-augmented regularization (model integration) (ablation) | AUD-BN · BP4D | average F1: .56 -> .57 | No clear effect |
| add generic BN supervision with 80% GT AU labels (ablation) | AU detector (AUD-GT) · CK+ | average F1: .78 -> .79 | Helped |
| add generic BN supervision with 100% GT AU labels (ablation) | AU detector (AUD-GT) · CK+ | average F1: .78 -> .83 | Helped |
| add generic BN supervision with 60% GT AU labels (ablation) | AU detector (AUD-GT) · BP4D | average F1: .62 -> .62 | Helped |
Closest papers by meaning
- Semantic Relationships Guided Representation Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.91)
- Classifier Learning With Prior Probabilities for Facial Action Unit Recognition (CVPR 2018 · similarity 0.91)
- Deep Structured Learning for Facial Action Unit Intensity Estimation (CVPR 2017 · similarity 0.90)
- Weakly Supervised Facial Action Unit Recognition Through Adversarial Training (CVPR 2018 · similarity 0.90)
- Facial Action Unit Detection With Transformers (CVPR 2021 · similarity 0.90)
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.89)
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.