Semantic Image Synthesis on ADE20K (val)
22.3FIDPLACE
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PLACEArchitecture=Diffusion-based2024.03 | 22.3 | 0.507 | — | — | — | — | — | — | — | — | — | — | |
| DP-SIMSBackbone=ConvNeXt-L2023.12 | 22.7 | 0.543 | — | — | — | — | — | — | — | — | — | 67.8 | |
| FreestyleNetArchitecture=Diffusion-based2024.03 | 25 | 0.419 | — | — | — | — | — | — | — | — | — | — | |
| ECGANArchitecture=GAN-based2024.03 | 25.8 | 0.506 | — | — | — | — | — | — | — | — | — | — | |
| DP-GAN2023.12 | 26.1 | 0.441 | — | — | — | — | — | — | — | — | — | 52.7 | |
| SDM2023.12 | 27.5 | 0.44 | — | — | — | — | — | — | — | — | — | 51.9 | |
| SDMArchitecture=Diffusion-based2024.03 | 27.5 | 0.392 | — | — | — | — | — | — | — | — | — | — | |
| PITIArchitecture=Diffusion-based2024.03 | 27.9 | 0.294 | — | — | — | — | — | — | — | — | — | — | |
| OASISNumber of parameters=94M, VGG Perceptual Loss=false2020.12 | 28.3 | 0.488 | — | — | — | — | — | — | — | — | — | — | |
| OASIS2023.12 | 28.3 | 0.488 | — | — | — | — | — | — | — | — | — | 53.5 | |
| OASISArchitecture=GAN-based2024.03 | 28.3 | 0.483 | — | — | — | — | — | — | — | — | — | — | |
| SC-GANArchitecture=GAN-based2024.03 | 29.3 | 0.452 | — | — | — | — | — | — | — | — | — | — | |
| RESAILArchitecture=GAN-based2024.03 | 30.2 | 0.493 | — | — | — | — | — | — | — | — | — | — | |
| ControlNetArchitecture=Diffusion-based2024.03 | 31.2 | 0.369 | — | — | — | — | — | — | — | — | — | — | |
| LGGANNumber of parameters=n/a, VGG Perceptual Loss=true2020.12 | 31.6 | 0.416 | — | — | — | — | — | — | — | — | — | — | |
| LGGANArchitecture=GAN-based2024.03 | 31.6 | 0.416 | — | — | — | — | — | — | — | — | — | — | |
| CC-FPSENumber of parameters=131M, VGG Perceptual Loss=true2020.12 | 31.7 | 0.437 | — | — | — | — | — | — | — | — | — | — | |
| CC-FPSEArchitecture=GAN-based2024.03 | 31.7 | 0.437 | — | — | — | — | — | — | — | — | — | — | |
| CC-FPSE-AUGAugmentation=Label map warping + Edge loss2020.11 | 32.6 | 0.44 | 0.458 | 0.431 | — | — | — | — | — | — | — | — | |
| CC-FPSEedge_loss=true, warping_augmentation=true2020.11 | 32.6 | 0.44 | 0.458 | 0.431 | 83 | 88.2 | 73.5 | 51.3 | 54.5 | 49.6 | — | — | |
| SAFMArchitecture=GAN-based2024.03 | 32.8 | 0.501 | — | — | — | — | — | — | — | — | — | — | |
| SPADE+Number of parameters=102M, VGG Perceptual Loss=true2020.12 | 32.9 | 0.425 | — | — | — | — | — | — | — | — | — | — | |
| CC-FPSEAugmentation=None2020.11 | 33.2 | 0.426 | 0.445 | 0.416 | — | — | — | — | — | — | — | — | |
| CC-FPSEedge_loss=false, warping_augmentation=false2020.11 | 33.2 | 0.426 | 0.445 | 0.416 | 82.2 | 87.6 | 72 | 49.8 | 52.7 | 48.3 | — | — | |
| SPADENumber of parameters=102M, VGG Perceptual Loss=true2020.12 | 33.9 | 0.385 | — | — | — | — | — | — | — | — | — | — | |
| SPADE2023.12 | 33.9 | 0.385 | — | — | — | — | — | — | — | — | — | — | |
| SPADEArchitecture=GAN-based2024.03 | 33.9 | 0.385 | — | — | — | — | — | — | — | — | — | — | |
| CC-FPSEedge_loss=true, warping_augmentation=false2020.11 | 34.2 | 0.431 | 0.447 | 0.422 | 81.7 | 87.5 | 70.8 | 50.4 | 53.4 | 48.8 | — | — | |
| SPADEAugmentation=None2020.11 | 34.4 | 0.396 | 0.417 | 0.386 | — | — | — | — | — | — | — | — | |
| SPADEedge_loss=false, warping_augmentation=false2020.11 | 34.4 | 0.396 | 0.417 | 0.386 | 81 | 85.8 | 72.1 | 46.6 | 49.1 | 45.3 | — | — | |
| Reg-VQgenerative model=auto-regressive, resolution=256x2562023.03 | 34.47 | — | — | — | — | — | — | — | — | — | — | — | |
| SPADE-AUGAugmentation=Label map warping + Edge loss2020.11 | 34.6 | 0.412 | 0.432 | 0.402 | — | — | — | — | — | — | — | — | |
| SPADEedge_loss=true, warping_augmentation=true2020.11 | 34.6 | 0.412 | 0.432 | 0.402 | 81.3 | 86 | 72.5 | 48 | 50.8 | 46.5 | — | — | |
| SPADEedge_loss=true, warping_augmentation=false2020.11 | 34.9 | 0.408 | 0.422 | 0.401 | 81.3 | 86.2 | 72.2 | 47.7 | 50 | 46.5 | — | — | |
| Gumbel-VQgenerative model=auto-regressive, resolution=256x2562023.03 | 37.51 | — | — | — | — | — | — | — | — | — | — | — | |
| VQ-GANgenerative model=auto-regressive, resolution=256x2562023.03 | 38.53 | — | — | — | — | — | — | — | — | — | — | — | |
| Pix2PixHD-AUGAugmentation=Label map warping + Edge loss2020.11 | 41.5 | 0.325 | 0.36 | 0.307 | — | — | — | — | — | — | — | — | |
| Pix2PixHDedge_loss=true, warping_augmentation=true2020.11 | 41.5 | 0.325 | 0.36 | 0.307 | 77.9 | 84.4 | 65.9 | 38.2 | 42.7 | 35.8 | — | — | |
| Pix2PixHDAugmentation=None2020.11 | 59.1 | 0.244 | 0.279 | 0.226 | — | — | — | — | — | — | — | — | |
| Pix2PixHDedge_loss=false, warping_augmentation=false2020.11 | 59.1 | 0.244 | 0.279 | 0.226 | 71.8 | 80.5 | 55.3 | 29.5 | 34.7 | 26.8 | — | — | |
| VQ-VAEgenerative model=auto-regressive, resolution=256x2562023.03 | 60.29 | — | — | — | — | — | — | — | — | — | — | — | |
| SPADE+Number of parameters=102M, VGG Perceptual Loss=false2020.12 | 60.7 | 0.21 | — | — | — | — | — | — | — | — | — | — | |
| Pix2PixHDedge_loss=true, warping_augmentation=false2020.11 | 65.2 | 0.204 | 0.241 | 0.184 | 69.1 | 77.8 | 52.9 | 25 | 30.8 | 22 | — | — | |
| CRNNumber of parameters=84M, VGG Perceptual Loss=true2020.12 | 73.3 | 0.224 | — | — | — | — | — | — | — | — | — | — | |
| Pix2pixHD2023.12 | 73.3 | 0.224 | — | — | — | — | — | — | — | — | — | — | |
| Pix2pixHDNumber of parameters=183M, VGG Perceptual Loss=true2020.12 | 81.8 | 0.203 | — | — | — | — | — | — | — | — | — | — | |
| pix2pixHDArchitecture=GAN-based2024.03 | 81.8 | 0.203 | — | — | — | — | — | — | — | — | — | — | |
| DP-SIMS3D noise=false2023.12 | — | — | — | — | — | — | — | — | — | — | 0.47 | — | |
| OASIS3D noise=true2023.12 | — | — | — | — | — | — | — | — | — | — | 0.35 | — | |
| SPADE+3D noise=true2023.12 | — | — | — | — | — | — | — | — | — | — | 0.16 | — | |
| SPADE+3D noise=false2023.12 | — | — | — | — | — | — | — | — | — | — | 0.5 | — |