Table 1 Experimental results of Cold Diffusion on different datasets
From: Generating oracle bone inscriptions based on the structure aware diffusion model
Dataset | Metric | 32 × 32 | 64 × 64 | 128 × 128 | |
|---|---|---|---|---|---|
Testsingle | Degraded | FID↓ | 195.773 | 68.449 | 28.692 ± 0.3243 |
SSIM↑ | 0.173 | 0.669 | 0.912 ± 0.0001 | ||
RMSE↓ | 0.252 | 0.162 | 0.082 ± 0.0004 | ||
Standard Sampled | FID↓ | 43.095 | 10.302 | 4.757 ± 0.0297 | |
SSIM↑ | 0.650 | 0.929 | 0.982 ± 0.0014 | ||
RMSE↓ | 0.150 | 0.064 | 0.027 ± 0.0017 | ||
OBS Sampled | FID↓ | 14.086 | 5.194 | 4.634 ± 0.0393 | |
SSIM↑ | 0.594 | 0.910 | 0.982 ± 0.0003 | ||
RMSE↓ | 0.188 | 0.077 | 0.027 ± 0.0005 | ||
OBC306 | Degraded | FID↓ | 194.855 | 73.207 | 34.501 ± 0.2255 |
SSIM↑ | 0.166 | 0.666 | 0.905 ± 0.0007 | ||
RMSE↓ | 0.252 | 0.165 | 0.092 ± 0.0003 | ||
Standard Sampled | FID↓ | 53.409 | 11.632 | 7.985 ± 0.0831 | |
SSIM↑ | 0.619 | 0.912 | 0.976 ± 0.0003 | ||
RMSE↓ | 0.157 | 0.071 | 0.034 ± 0.0004 | ||
OBS Sampled | FID↓ | 14.537 | 4.358 | 6.031 ± 0.0868 | |
SSIM↑ | 0.570 | 0.894 | 0.972 ± 0.0003 | ||
RMSE↓ | 0.197 | 0.086 | 0.040 ± 0.0004 | ||
HWOBC | Degraded | FID↓ | 187.575 | 105.084 | 46.500 ± 0.4855 |
SSIM↑ | 0.228 | 0.681 | 0.901 ± 0.0006 | ||
RMSE↓ | 0.418 | 0.272 | 0.154 ± 0.0001 | ||
Standard Sampled | FID↓ | 98.829 | 32.194 | 26.908 ± 0.5083 | |
SSIM↑ | 0.525 | 0.907 | 0.975 ± 0.0004 | ||
RMSE↓ | 0.168 | 0.093 | 0.060 ± 0.0005 | ||
OBS Sampled | FID↓ | 106.022 | 29.841 | 22.726 ± 0.2186 | |
SSIM↑ | 0.507 | 0.898 | 0.976 ± 0.0004 | ||
RMSE↓ | 0.173 | 0.102 | 0.059 ± 0.0008 | ||
