Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Facial Action Unit Recognition From Partially Labeled Data
Shan Wu, Shangfei Wang, Bowen Pan, Qiang Ji
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 AU label distribution (RBM) constraint to the deep AU classifier (ablation) | DNN+SVM AU recognizer · BP4D (complete labels) | average F1 score: 0.446 -> 0.491 | Improved |
| add AU label distribution (RBM) constraint to the deep AU classifier (ablation) | DNN+SVM AU recognizer · DISFA (complete labels) | average F1 score: 0.532 -> 0.584 | Improved |
| add AU label distribution constraint (ablation) | DNN+SVM AU recognizer · DISFA AU17 | F1 score: 0.201 -> 0.447 | Improved |
| add AU label distribution constraint (ablation) | DNN+SVM AU recognizer · BP4D AU7 | F1 score: 0.666 -> 0.620 | Hurt |
| add AU label distribution constraint (ablation) | DNN+SVM AU recognizer · BP4D AU2 | F1 score: 0.187 -> 0.166 | Hurt |
| add AU label distribution constraint (ablation) | DNN+SVM AU recognizer · DISFA AU1 | F1 score: 0.608 -> 0.574 | Hurt |
| add AU label distribution constraint under random missing AU labels (10-50%) (ablation) | DNN+SVM AU recognizer · DISFA and BP4D (partial labels) | F1 score | Improved |
| learn deep features instead of raw pixels (ablation) | AU recognizer · BP4D and DISFA | subject-bias in t-SNE embedding | Improved |
| use the deep model without label constraint instead of linear SVM on handcrafted features | AU recognizer · BP4D | average F1 score: 0.353 -> 0.446 | Improved |
| proposed method vs JPML (handcrafted SIFT, dataset-statistics AU relations) | AU recognizer · BP4D | average F1 score: 0.459 -> 0.491 | Improved |
| proposed method vs DRML region-layer network | AU recognizer · BP4D | average F1 score: 0.483 -> 0.491 | Mixed |
Closest papers by meaning
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.92)
- Weakly Supervised Facial Action Unit Recognition Through Adversarial Training (CVPR 2018 · similarity 0.92)
- Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition (NeurIPS 2019 · similarity 0.92)
- Knowledge-Driven Self-Supervised Representation Learning for Facial Action Unit Recognition (CVPR 2022 · similarity 0.91)
- Deep Region and Multi-Label Learning for Facial Action Unit Detection (CVPR 2016 · similarity 0.91)
- Deep Structure Inference Network for Facial Action Unit Recognition (ECCV 2018 · similarity 0.91)
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.
