Image Inpainting On Celeba Hq
Métriques
FID
P-IDS
U-IDS
Résultats
Résultats de performance de divers modèles sur ce benchmark
Nom du modèle | FID | P-IDS | U-IDS | Paper Title | Repository |
|---|---|---|---|---|---|
| MAT | 4.86 | 13.83 | 25.33 | MAT: Mask-Aware Transformer for Large Hole Image Inpainting | - |
| WavePaint | 5.53 | - | - | WavePaint: Resource-efficient Token-mixer for Self-supervised Inpainting | - |
| MADF | 6.83 | 3.41 | 11.26 | Image Inpainting by End-to-End Cascaded Refinement with Mask Awareness | - |
| ICT | 12.84 | 0.13 | 0.58 | High-Fidelity Pluralistic Image Completion with Transformers | - |
| CoModGAN | 5.65 | 11.23 | 22.54 | Large Scale Image Completion via Co-Modulated Generative Adversarial Networks | - |
| LaMa | 8.15 | 2.07 | 7.58 | Resolution-robust Large Mask Inpainting with Fourier Convolutions | - |
| BDPM | 1.17 | 14.14 | 28.4 | Binary Diffusion Probabilistic Model | - |
0 of 7 row(s) selected.