Research map / Image generation / Editing and style transfer
MAUGen: A Unified Diffusion Approach for Multi-Identity Facial Expression and AU Label Generation
Xiangdong Li, Ye Lou, Ao Gao, Wei Zhang, Siyang Song
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 identity features without decoupling (ablation) | MAUGen · DISFA ablation | F1 (%): 92.6 -> 88.5 | Hurt |
| add Identity Decoupling Modules (modulation mask + residual filtering) (ablation) | MAUGen · DISFA ablation | F1 (%): 88.5 -> 93.6 | Improved |
| add Identity Decoupling Modules (ablation) | MAUGen · DISFA ablation | FID: 6.0795 -> 5.9120 | Improved |
| add AU-guided mutual conditioning (ablation) | MAUGen · DISFA ablation | CLIP score: 0.2769 -> 0.2817 | Improved |
| add language-based query initialization (ablation) | MAUGen · DISFA ablation | F1 (%): to 95.7 | Improved |
| add LGFE (ablation) | MAUGen · DISFA ablation | F1 (%): 95.7 -> 96.2 | Improved |
| add pseudo-label image-label alignment loss with cross-detector agreement (ablation) | MAUGen · DISFA ablation | F1 (%): 96.2 -> 96.7 | Improved |
| train AU detectors on real plus synthetic MIFA data | FMAE (ViT-B) · DISFA test (real) | F1 (%): 63.5 -> 70.9 | Improved |
| train AU detectors on real plus synthetic MIFA data | GraphAU (Swin-B) · DISFA test (real) | F1 (%): 62.4 -> 64.1 | Improved |
| train on MIFA instead of DISFA | GraphAU (Res50) · AffectNet | F1 (%): 53.21 -> 56.01 | Improved |
| train on MIFA instead of DISFA | FMAE (Vit-H) · AffectNet | F1 (%): 70.58 -> 69.49 | Hurt |
| MAUGen vs baselines on image-text alignment (CLIP) | MAUGen · DISFA | CLIP score: 0.295 vs StyleCLIP 0.303 | Mixed |
Closest papers by meaning
- From Pixels to Semantics: Unified Facial Action Representation Learning for Micro-Expression Analysis (ICLR 2026 · similarity 0.88)
- AUEditNet: Dual-Branch Facial Action Unit Intensity Manipulation with Implicit Disentanglement (CVPR 2024 · similarity 0.88)
- Dual Semi-Supervised Learning for Facial Action Unit Recognition (AAAI 2019 · similarity 0.87)
- EmotioNet: An Accurate, Real-Time Algorithm for the Automatic Annotation of a Million Facial Expressions in the Wild (CVPR 2016 · similarity 0.87)
- $\text{ID}^3$: Identity-Preserving-yet-Diversified Diffusion Models for Synthetic Face Recognition (NeurIPS 2024 · similarity 0.86)
- When StyleGAN Meets Stable Diffusion: a W+ Adapter for Personalized Image Generation (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.
