Research map / Label-efficient and robust learning / Recognition and long-tail
Weakly Supervised Facial Action Unit Recognition Through Adversarial Training
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 adversarial training with expression-dependent and expression-independent AU priors instead of HTL | AU recognition from expression-labeled images · CK+ within-database | F1 (Avg. of com): 0.6263 -> 0.7691 | Helped |
| Use adversarial training with expression-dependent and expression-independent AU priors instead of HTL | AU recognition from expression-labeled images · MMI within-database | F1 (Avg. of com): 0.4310 -> 0.5206 | Helped |
| Use adversarial training with expression-dependent and expression-independent AU priors instead of HTL | AU recognition from expression-labeled images · UNBC within-database | F1 (Avg. of com): 0.1367 -> 0.3365 | Helped |
| Use adversarial training with summarized AU priors instead of HTL | AU recognition across databases · CK+ to MMI | F1 (Avg. of com): 0.4200 -> 0.5066 | Helped |
| Use adversarial training with summarized AU priors instead of HTL | AU recognition across databases · CK+ to UNBC | F1 (Avg. of com): 0.1508 -> 0.2948 | Helped |
| Use adversarial training with summarized AU priors instead of HTL | AU recognition across databases · MMI to CK+ | F1 (Avg. of com): 0.6263 -> 0.6418 | Helped |
| Use adversarial training with summarized AU priors instead of HTL | AU recognition across databases · MMI to UNBC | F1 (Avg. of com): 0.1376 -> 0.2482 | Helped |
| Train on pain-expression data and test on basic-expression data instead of using HTL | AU recognition across databases · UNBC to CK+ | F1 (Avg. of com): 0.5543 -> 0.3572 | Hurt |
| Train on pain-expression data and test on basic-expression data instead of using HTL | AU recognition across databases · UNBC to MMI | F1 (Avg. of com): 0.3632 -> 0.2933 | Hurt |
| Train without AU labels using summarized domain knowledge | RAN · CK+ within-database | F1 (Avg.): 0.7707 -> 0.7156 | Mixed |
Closest papers by meaning
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.94)
- Classifier Learning With Prior Probabilities for Facial Action Unit Recognition (CVPR 2018 · similarity 0.91)
- Knowledge Augmented Deep Neural Networks for Joint Facial Expression and Action Unit Recognition (NeurIPS 2020 · similarity 0.90)
- Knowledge-Driven Self-Supervised Representation Learning for Facial Action Unit Recognition (CVPR 2022 · similarity 0.90)
- Joint Representation and Estimator Learning for Facial Action Unit Intensity Estimation (CVPR 2019 · similarity 0.89)
- AUEditNet: Dual-Branch Facial Action Unit Intensity Manipulation with Implicit Disentanglement (CVPR 2024 · similarity 0.87)
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.