Image Generation on STL-10 (test)
10.61Inception ScoreSSGAN-LA
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SSGAN-LABase Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.61 | 14.58 | — | |
| SSGAN-MSBase Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.4 | 16.46 | — | |
| SSGANBase Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.36 | 16.84 | — | |
| DAGAN-MDBase Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.31 | 16.16 | — | |
| DAGAN+Base Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.27 | 16.42 | — | |
| GANBase Model=unconditional BigGAN, Label Usage=unsupervised, Evaluation Sampling=50k images2021.06 | 10.25 | 20.15 | — | |
| SDST-Strong-SETBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 9.39 | — | — | |
| SDST-Balance-SETBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 9.31 | — | — | |
| Improving MMD GANCategory=GAN-based2020.11 | 9.3 | 37.6 | — | |
| ADAPTrelaxBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 9.29 | — | — | |
| SDST-Balance-RigLBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 9.28 | — | — | |
| SDST-Balance-SETBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 9.26 | — | — | |
| ADAPTrelaxBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 9.26 | — | — | |
| SDST-Strong-SETBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 9.21 | — | — | |
| AutoGANCategory=GAN-based2020.11 | 9.2 | 31 | — | |
| SDST-Strong-RigLBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 9.17 | — | — | |
| SDST-Strong-RigLBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 9.16 | — | — | |
| SDST-Balance-RigLBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 9.12 | — | — | |
| SN-GANCategory=GAN-based2020.11 | 9.1 | 40.1 | — | |
| SDST-Strong-RigLBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 9.1 | — | — | |
| ADAPTrelaxBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 9.08 | — | — | |
| SDST-Balance-RigLBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 9.07 | — | — | |
| ADAPTrelaxBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 9.06 | — | — | |
| SDST-Balance-RigLBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 8.98 | — | — | |
| SDST-Balance-SETBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 8.92 | — | — | |
| ProbGANCategory=GAN-based2020.11 | 8.9 | 46.7 | — | |
| Static-StrongBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 8.7 | — | — | |
| Static-BalanceBackbone=SNGAN, Generator density=50%, Maximal discriminator density (d_max)=100%2023.02 | 8.69 | — | — | |
| SDST-Strong-SETBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 8.53 | — | — | |
| Dense BaselineBackbone=SNGAN, Generator density=100%, Maximal discriminator density (d_max)=100%2023.02 | 8.48 | — | — | |
| Static-BalanceBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 8.44 | — | — | |
| SDST-Balance-SETBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 8.43 | — | — | |
| DC-VAECategory=VAE-based, Mode=Reconstruction2020.11 | 8.4 | 43.6 | — | |
| Static-StrongBackbone=SNGAN, Generator density=30%, Maximal discriminator density (d_max)=100%2023.02 | 8.35 | — | — | |
| Static-StrongBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 8.22 | — | — | |
| Static-BalanceBackbone=SNGAN, Generator density=20%, Maximal discriminator density (d_max)=100%2023.02 | 8.19 | — | — | |
| SDST-Strong-RigLBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 8.15 | — | — | |
| DC-VAECategory=VAE-based, Mode=Sampling2020.11 | 8.1 | 41.9 | — | |
| Static-BalanceBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 7.94 | — | — | |
| Static-StrongBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 7.7 | — | — | |
| SDST-Strong-SETBackbone=SNGAN, Generator density=10%, Maximal discriminator density (d_max)=100%2023.02 | 7.65 | — | — | |
| OCFGAN-GP + DG flow (KL)DG flow variant=KL2020.12 | 7.46 | — | — | |
| OCFGAN-GP + DG flow (log D)DG flow variant=log D2020.12 | 7.33 | — | — | |
| MicroBatchGAN2021.01 | 7.23 | — | — | |
| OCFGAN-GP + DG flow (JS)DG flow variant=JS2020.12 | 7.1 | — | — | |
| OCFGAN-GPDG flow variant=None (Base Model)2020.12 | 7.09 | — | — | |
| DuelGAN2021.01 | 6.22 | — | — | |
| MMDGAN + DG flow (KL)DG flow variant=KL2020.12 | 6.16 | — | — | |
| D2GAN2021.01 | 6.15 | — | — | |
| MMDGAN + DG flow (JS)DG flow variant=JS2020.12 | 6.12 | — | — | |
| MMDGAN + DG flow (log D)DG flow variant=log D2020.12 | 6.12 | — | — | |
| MMDGANDG flow variant=None (Base Model)2020.12 | 6.07 | — | — | |
| DCGAN2021.01 | 5.87 | — | — | |
| WGAN2021.01 | 3.97 | — | — | |
| VAE + DG flow (KL)DG flow variant=KL2020.12 | 3.72 | — | — | |
| VAE + DG flow (log D)DG flow variant=log D2020.12 | 3.65 | — | — | |
| VAE + DG flow (JS)DG flow variant=JS2020.12 | 3.27 | — | — | |
| VAEDG flow variant=None (Base Model)2020.12 | 3.25 | — | — | |
| GAN2021.01 | 2.17 | — | — | |
| DDGAN2023.11 | — | 21.79 | 40 | |
| DG flowRefinement=KL, Backbone=WGAN-GP2020.12 | — | 39.07 | — | |
| DG flowRefinement=JS, Backbone=WGAN-GP2020.12 | — | 50.83 | — | |
| DG flowRefinement=log D, Backbone=WGAN-GP2020.12 | — | 39.71 | — | |
| DG flowRefinement=KL, Backbone=SN-DCGAN, Loss=hinge (hi)2020.12 | — | 34.95 | — | |
| DG flowRefinement=JS, Backbone=SN-DCGAN, Loss=hinge (hi)2020.12 | — | 36.37 | — | |
| DG flowRefinement=log D, Backbone=SN-DCGAN, Loss=hinge (hi)2020.12 | — | 36.56 | — | |
| DG flowRefinement=KL, Backbone=SN-DCGAN, Loss=non-saturating (ns)2020.12 | — | 34.6 | — | |
| DG flowRefinement=JS, Backbone=SN-DCGAN, Loss=non-saturating (ns)2020.12 | — | 35.37 | — | |
| DG flowRefinement=log D, Backbone=SN-DCGAN, Loss=non-saturating (ns)2020.12 | — | 37.07 | — | |
| Diffusion StyleGAN22023.11 | — | 11.53 | — | |
| Dist-GANBackbone=SN-GAN, Conditionality=unconditional2019.11 | — | 36.19 | — | |
| Dist-GANBackbone=ResNet, Conditionality=unconditional2019.11 | — | 28.5 | — | |
| Dist-GAN + SSBackbone=SN-GAN, Conditionality=unconditional2019.11 | — | 29.79 | — | |
| Dist-GAN + SSBackbone=ResNet, Conditionality=unconditional2019.11 | — | 27.98 | — | |
| DOTBackbone=WGAN-GP2020.12 | — | 44.45 | — | |
| DOTBackbone=SN-DCGAN, Loss=hinge (hi)2020.12 | — | 34.85 | — | |
| DOTBackbone=SN-DCGAN, Loss=non-saturating (ns)2020.12 | — | 34.84 | — | |
| GN-GANBackbone=SN-GAN, Conditionality=unconditional2019.11 | — | 30.8 | — | |
| Ours(Dist-GAN + MS)Backbone=SN-GAN, Conditionality=unconditional2019.11 | — | 27.95 | — | |
| Ours(Dist-GAN + MS)Backbone=ResNet, Conditionality=unconditional2019.11 | — | 27.1 | — | |
| RDUOT2023.11 | — | 11.5 | 49 | |
| SN-DCGANRefinement=None (Base Model), Loss=hinge (hi)2020.12 | — | 40.54 | — | |
| SN-DCGANRefinement=None (Base Model), Loss=non-saturating (ns)2020.12 | — | 41.86 | — | |
| SN-GANBackbone=SN-GAN, Conditionality=unconditional2019.11 | — | 43.2 | — | |
| SN-GANBackbone=ResNet, Conditionality=unconditional2019.11 | — | 40.1 | — | |
| SNGAN2023.11 | — | 40.1 | — | |
| StyleFormer2023.11 | — | 15.17 | — | |
| StyleGAN2-ADA2023.11 | — | 13.72 | 36 | |
| StyleGAN2-Aug2023.11 | — | 12.97 | 39 | |
| TransGAN2023.11 | — | 18.28 | — | |
| WaveDiff2023.11 | — | 12.93 | 41 | |
| WGAN-GPBackbone=SN-GAN, Conditionality=unconditional2019.11 | — | 55.1 | — | |
| WGAN-GPRefinement=None (Base Model)2020.12 | — | 51.5 | — |