Image Generation on LSUN-Cat 1k samples (train)
2.82SwAV FIDVision-aided Adversarial training
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Vision-aided Adversarial trainingAugmentation Strategy=DiffAugment, Fine-tuning=true2021.12 | 2.82 | — | |
| Vision-aided Adversarial trainingAugmentation Strategy=ADA, Fine-tuning=true2021.12 | 3.27 | — | |
| ADAAugmentation Strategy=ADA2021.12 | 10.33 | — | |
| DiffAugmentAugmentation Strategy=DiffAugment2021.12 | 10.74 | — | |
| StyleGAN2Augmentation Strategy=none2021.12 | 22.71 | — | |
| ADA2020.06 | — | 43.25 | |
| StyleGAN22020.06 | — | 182.85 | |
| StyleGAN2 + DiffAugmentAugmentation=DiffAugment (Color + Translation + Cutout)2020.06 | — | 42.26 |