Text-to-Image Synthesis on MS-COCO (val)
6.09FIDOriginal Images
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Original Images2019.10 | 6.09 | 34.88 | — | 68.58 | 0.795 | 74.97 | 80.84 | — | |
| ERNIE-ViLGTraining mode=Supervised (fine-tuned)2021.12 | 7.9 | — | — | — | — | — | — | — | |
| XMC-GANTraining mode=Supervised (fine-tuned)2021.12 | 9.3 | — | — | — | — | — | — | — | |
| NÜWATraining mode=Supervised (fine-tuned)2021.12 | 12.9 | — | — | — | — | — | — | — | |
| ERNIE-ViLGTraining mode=Zero-shot2021.12 | 14.7 | — | — | — | — | — | — | — | |
| DF-GANTraining mode=Supervised (fine-tuned)2021.12 | 21.4 | — | — | — | — | — | — | — | |
| OP-GANVersion=Best Model, Bounding box source=Generated2019.10 | 24.7 | 27.88 | — | 89.01 | 0.819 | 35.85 | 50.47 | — | |
| OPv2 + MOObjects=102019.10 | 24.94 | 27.59 | — | 89.14 | 0.805 | 33.46 | 47.93 | — | |
| OPv2 + MOObjects=32019.10 | 24.96 | 27.65 | — | 89.13 | 0.807 | — | — | — | |
| OPv2 + MOObjects=12019.10 | 25.06 | 27.36 | — | 88.33 | 0.789 | — | — | — | |
| OPv2 + BBL + MOObjects=102019.10 | 25.8 | 27.9 | — | 89 | 0.814 | 34.51 | 48.9 | — | |
| OPv2 + BBL + MOObjects=32019.10 | 25.91 | 28.04 | — | 88.9 | 0.81 | — | — | — | |
| OPv2Objects=12019.10 | 26.18 | 27.68 | — | 87.37 | 0.798 | — | — | — | |
| OPv2 + BBL + MOObjects=12019.10 | 26.19 | 27.61 | — | 87.85 | 0.791 | — | — | — | |
| OPv2Objects=32019.10 | 26.45 | 27.78 | — | 87.74 | 0.805 | — | — | — | |
| OPv2Objects=102019.10 | 26.52 | 27.66 | — | 87.73 | 0.806 | 33.82 | 48.39 | — | |
| OPv2 + BBLObjects=12019.10 | 26.59 | 26.34 | — | 86.42 | 0.783 | — | — | — | |
| OPv2 + BBLObjects=32019.10 | 26.74 | 26.52 | — | 87.08 | 0.793 | — | — | — | |
| OPv2 + BBLObjects=102019.10 | 26.83 | 26.48 | — | 86.8 | 0.794 | 33.19 | 48.24 | — | |
| CogViewTraining mode=Zero-shot2021.12 | 27.1 | — | — | — | — | — | — | — | |
| DM-GANFine-tuned=true2019.10 | 27.34 | 32.32 | — | 91.87 | 0.823 | 33.44 | 48.03 | — | |
| DALL-ETraining mode=Zero-shot2021.12 | 27.5 | — | — | — | — | — | — | — | |
| Huang et al.Training mode=Supervised (fine-tuned)2021.12 | 29.9 | — | — | — | — | — | — | — | |
| OPv2Objects=02019.10 | 30.01 | 26.8 | — | 83.87 | 0.76 | 26.04 | 37.56 | — | |
| DM-GANTraining mode=Supervised (fine-tuned)2021.12 | 32.6 | — | — | — | — | — | — | — | |
| OPv2 + BBLObjects=02019.10 | 33.03 | 24.6 | — | 81.27 | 0.735 | 24 | 34.01 | — | |
| AttnGANFine-tuned=true2019.10 | 33.1 | 23.61 | — | 83.8 | 0.695 | 25.88 | 39.01 | — | |
| AttnGAN + OPFine-tuned=true2019.10 | 33.35 | 24.76 | — | 82.44 | 0.689 | 25.46 | 40.48 | — | |
| OPv2 + MOObjects=02019.10 | 35.36 | 24.32 | — | 79.75 | 0.695 | 21.15 | 30.24 | — | |
| AttnGANTraining mode=Supervised (fine-tuned)2021.12 | 35.5 | — | — | — | — | — | — | — | |
| Obj-GANFine-tuned=true2019.10 | 36.52 | 24.09 | — | 87.84 | 0.783 | 27.14 | 41.24 | — | |
| X-LXMERTTraining mode=Supervised (fine-tuned)2021.12 | 37.4 | — | — | — | — | — | — | — | |
| OPv2 + BBL + MOObjects=02019.10 | 45.79 | 21.84 | — | 72.71 | 0.626 | 16.55 | 22.76 | — | |
| StackGAN-v12017.10 | 74.05 | 8.45 | 1.45 | — | — | — | — | — | |
| StackGAN-v22017.10 | 81.59 | 8.3 | 1.55 | — | — | — | — | — | |
| [34]2019.10 | — | 23.74 | — | 86.44 | — | — | — | — | |
| ControlGAN2019.10 | — | 24.06 | — | 82.43 | — | — | — | — | |
| GAN-INT-CLSResolution=64x642016.12 | — | 7.88 | 1.89 | — | — | — | — | — | |
| HfGAN2019.10 | — | 27.53 | — | — | — | — | — | — | |
| MirrorGAN2019.10 | — | 26.47 | — | 74.52 | — | — | — | — | |
| SD-GAN2019.10 | — | 35.69 | — | — | — | — | — | — | |
| StackGANResolution=256x2562016.12 | — | 8.45 | 1.11 | — | — | — | — | — |