Table 2 Comparison with state-of-the-art methods and model complexity.
From: StyDiff: a refined style transfer method based on diffusion models
Methods | SSIM \(\uparrow\) | PD \(\downarrow\) | GM \(\uparrow\) | LPIPS \(\downarrow\) | Inference time \(\downarrow\) | Parameters (M)\(\downarrow\) |
|---|---|---|---|---|---|---|
InST36 | 0.5512 | 21.23 | 13.22 | 0.35 | 13.57 | 10.5 |
DiffuseIT37 | 0.6802 | 19.98 | 14.10 | 0.30 | 1.73 | 8.3 |
StyleFlow38 | 0.7311 | 18.77 | 15.67 | 0.28 | 1.13 | 17.4 |
AesPA-Net39 | 0.7476 | 19.10 | 15.99 | 0.22 | 1.43 | 13.8 |
Cartoon-flow40 | 0.7522 | 17.49 | 16.08 | 0.19 | 1.38 | 11.2 |
StyTr241 | 0.7878 | 16.33 | 17.47 | 0.20 | 0.94 | 16.0 |
Mamba-ST42 | 0.7601 | 17.38 | 16.02 | 0.18 | 1.20 | 10.7 |
CAST43 | 0.7447 | 18.10 | 15.22 | 0.11 | 1.73 | 7.2 |
AdaAttN44 | 0.7756 | 16.78 | 17.22 | 0.10 | 0.89 | 14.9 |
ArtFlow45 | 0.7612 | 16.92 | 17.30 | 0.14 | 1.19 | 12.5 |
Stable-diffusion28 | 0.7889 | 17.78 | 18.22 | 0.08 | 4.22 | 59.8 |
DiT46 | 0.7901 | 17.88 | 18.99 | 0.07 | 5.79 | 95.2 |
EFDM47 | 0.7857 | 17.83 | 17.00 | 0.15 | 2.55 | 33.6 |
StyleID48 | 0.7692 | 19.93 | 14.22 | 0.22 | 2.93 | 9.1 |
Ours | 0.7986 | 15.67 | 18.90 | 0.05 | 2.93 | 24.3 |
