Fine-Grained Aesthetic Assessment (Series-level) on FGAesthetics Natural
75.3s-AccFGAesQ (w DiffToken)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FGAesQ (w DiffToken)Params=86.3M, DiffToken=Enabled2026.03 | 75.3 | 0.729 | |
| FGAesQ (w/o DiffToken)Params=86.3M, DiffToken=Disabled2026.03 | 72 | 0.664 | |
| Q-Align (ICML 2024)Params=8.20B2026.03 | 67.6 | 0.496 | |
| Charm (Fine-tuned)Params=85.7M, Evaluation Protocol=Fine-tuned2026.03 | 67 | 0.474 | |
| UNIAA (arXiv 2024)Params=7.06B2026.03 | 65.6 | 0.486 | |
| VILA (Fine-tuned)Params=303M, Evaluation Protocol=Fine-tuned2026.03 | 64.3 | 0.43 | |
| Charm (CVPR 2025)Params=85.7M2026.03 | 60.2 | 0.404 | |
| RealQA (arXiv 2025)Params=8.29B2026.03 | 59.3 | 0.382 | |
| VILA (CVPR 2023)Params=303M2026.03 | 54.6 | 0.296 | |
| MUSIQ (Fine-tuned)Params=78.6M, Evaluation Protocol=Fine-tuned2026.03 | 54.5 | 0.356 | |
| TANet (IJCAI 2022)Params=13.9M2026.03 | 53.9 | 0.281 | |
| NIMA (Fine-tuned)Params=54.3M, Evaluation Protocol=Fine-tuned2026.03 | 53.6 | 0.312 | |
| MLSP (CVPR 2019)Params=114 M2026.03 | 53.1 | 0.255 | |
| NIMA (TIP 2018)Params=54.3M2026.03 | 49.3 | 0.225 | |
| MUSIQ (ICCV 2021)Params=78.6M2026.03 | 48.7 | 0.233 |