Research map / Label-efficient and robust learning / Semi-supervised learning
Dual Semi-Supervised Learning 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 labels as auxiliary input (ablation) | DSGANne · CK+ AU recognition, missing rate 0.1 | avg F1: .8141 -> .8287 | Helped |
| add expression labels as auxiliary input (ablation) | DSGANne · MMI AU recognition, missing rate 0.1 | avg F1: .5672 -> .5732 | Helped |
| add expression labels as auxiliary input (ablation) | DSGANne · Pain AU recognition, missing rate 0.5 | avg F1: .4939 -> .5195 | Helped |
| add reconstruction losses (dual-task feedback) (ablation) | DSGAN · CK+ AU recognition, missing rate 0.1 | avg F1: .8131 -> .8287 | Helped |
| add reconstruction losses (dual-task feedback) (ablation) | DSGAN · Pain AU recognition, missing rate 0.5 | avg F1: .4355 -> .5195 | Helped |
| add reconstruction losses (dual-task feedback) (ablation) | DSGAN · MMI AU recognition, missing rate 0.5 | avg F1: .5264 -> .5373 | Helped |
| add reconstruction losses (ablation) | DSGAN · CK+ face synthesis, missing rate 0.2 | RMSE (lower better): 1.0103 -> 0.9687 | Helped |
| add expression labels (ablation) | DSGAN · CK+ face synthesis, missing rate 0.2 | RMSE (lower better): 1.0002 -> 0.9687 | Helped |
| dual semi-supervised GAN with face synthesis vs DRBM | DSGAN · CK+ face synthesis, missing rate 0.2 | RMSE (lower better): 1.3866 -> 0.9687 | Helped |
| train on Pain, test on other databases (cross-database) | DSGAN · Cross-database AU recognition | avg F1 | Hurt |
| DSGAN vs best baseline RBM-P | DSGAN · CK+ AU recognition, missing rate 0.5 | avg F1: .7868 -> .7917 | Mixed |
| semi-supervised (0.5 missing) vs fully supervised methods | DSGAN · MMI AU recognition | avg F1: .6712 -> .5373 | Hurt |
Closest papers by meaning
- Weakly Supervised Facial Action Unit Recognition Through Adversarial Training (CVPR 2018 · similarity 0.94)
- Multi-label Co-regularization for Semi-supervised Facial Action Unit Recognition (NeurIPS 2019 · similarity 0.91)
- Classifier Learning With Prior Probabilities for Facial Action Unit Recognition (CVPR 2018 · similarity 0.89)
- Knowledge-Driven Self-Supervised Representation Learning for Facial Action Unit Recognition (CVPR 2022 · similarity 0.89)
- Knowledge Augmented Deep Neural Networks for Joint Facial Expression and Action Unit Recognition (NeurIPS 2020 · similarity 0.89)
- AUEditNet: Dual-Branch Facial Action Unit Intensity Manipulation with Implicit Disentanglement (CVPR 2024 · similarity 0.88)
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.