Style Transfer on MS-COCO (content) + WikiArt (style) (test)
0.3884LPIPSSaMam
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SaMamArchitecture=Mamba-based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=71.1, Time (s)=0.034, Params (M)=18.52025.03 | 0.3884 | 17.946 | 26.305 | 0.2703 | |
| AesPAArchitecture=CNN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=334.3, Time (s)=0.348, Params (M)=24.22025.03 | 0.405 | 20.236 | 29.837 | 0.3291 | |
| S2WATArchitecture=Transformer based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=582.6, Time (s)=0.278, Params (M)=64.962025.03 | 0.4256 | 23.43 | 34.827 | 0.3736 | |
| ATKArchitecture=CNN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=291.4, Time (s)=0.049, Params (M)=11.182025.03 | 0.432 | 24.788 | 36.928 | 0.3683 | |
| AttenSTcategory=diffusion-based2026.05 | 0.467 | 18.559 | 28.693 | — | |
| StyTr²Resolution=512 x 512, Time (s)=52.35, Memory Usage (MiB)=21602024.09 | 0.48 | 18.77 | 29.31 | 0.32 | |
| StyleIDArchitecture=Diffusion based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=6094.6, Time (s)=47.746, Params (M)=1066.242025.03 | 0.4803 | 24.488 | 37.73 | 0.3132 | |
| StyleFormerArchitecture=Transformer based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=172.1, Time (s)=0.207, Params (M)=19.92025.03 | 0.4898 | 28.798 | 44.392 | 0.3976 | |
| AesPA-NetResolution=512 x 512, Time (s)=165.99, Memory Usage (MiB)=41842024.09 | 0.49 | 22.85 | 35.45 | 0.33 | |
| StyleIDResolution=512 x 512, Time (s)=2744.38, Memory Usage (MiB)=199302024.09 | 0.49 | 18.29 | 28.65 | 0.29 | |
| CAPVSTArchitecture=Reversible-NN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=179.9, Time (s)=0.048, Params (M)=4.092025.03 | 0.4955 | 26.33 | 40.871 | 0.2912 | |
| StyTr2Architecture=Transformer based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=1283.5, Time (s)=0.385, Params (M)=35.392025.03 | 0.4992 | 25.204 | 39.285 | 0.4226 | |
| StyleIDcategory=diffusion-based2026.05 | 0.505 | 18.131 | 28.801 | 0.228 | |
| AdaAttNResolution=512 x 512, Time (s)=23.89, Memory Usage (MiB)=55542024.09 | 0.51 | 19.46 | 30.81 | 0.32 | |
| AesPA-Netcategory=conventional2026.05 | 0.514 | 19.76 | 31.42 | 0.246 | |
| EFDMArchitecture=CNN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=63.3, Time (s)=0.027, Params (M)=7.012025.03 | 0.5252 | 34.834 | 54.655 | 0.3391 | |
| Mamba-STResolution=512 x 512, Time (s)=24.7, Memory Usage (MiB)=14142024.09 | 0.53 | 16.75 | 27.11 | 0.33 | |
| VCTArchitecture=Diffusion based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=77.1, Time (s)=680.972, Params (M)=1066.242025.03 | 0.5429 | 37.485 | 59.379 | 0.4671 | |
| AdaAttNcategory=conventional2026.05 | 0.544 | 18.658 | 30.35 | 0.286 | |
| StyTR2category=conventional2026.05 | 0.545 | 18.89 | 30.72 | 0.301 | |
| CoCoDiffGuidance Type=Reference-guided2026.02 | 0.549 | 18.432 | 30.1 | 0.609 | |
| AesPAGuidance Type=Reference-guided2026.02 | 0.556 | 19.645 | 32.124 | 0.632 | |
| ArtFlowcategory=conventional2026.05 | 0.556 | 21.252 | 34.63 | 0.292 | |
| AdaINResolution=512 x 512, Time (s)=12.26, Memory Usage (MiB)=8242024.09 | 0.56 | 16.8 | 27.81 | 0.35 | |
| SCSAGuidance Type=Reference-guided2026.02 | 0.562 | 20.835 | 34.106 | 0.612 | |
| Scheduled style injectionGamma schedule=√·-γ, Timestep schedule=TS↓, ControlNet=0.25→0.1252026.05 | 0.564 | 16.285 | 27.036 | 0.295 | |
| StyleIDGuidance Type=Reference-guided2026.02 | 0.565 | 21.01 | 34.446 | 0.619 | |
| ArtFlowArchitecture=Reversible-NN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=517, Time (s)=0.152, Params (M)=6.462025.03 | 0.5671 | 29.199 | 47.325 | 0.3125 | |
| ZStarArchitecture=Diffusion based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=6639, Time (s)=42.439, Params (M)=1066.242025.03 | 0.5674 | 31.24 | 50.534 | 0.4277 | |
| Scheduled style injectionGamma schedule=cos-γ, Timestep schedule=TS↓2026.05 | 0.575 | 16.124 | 26.976 | 0.297 | |
| STTRArchitecture=Transformer based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=110.4, Time (s)=0.181, Params (M)=45.642025.03 | 0.5755 | 30.302 | 49.316 | 0.3517 | |
| UCASTArchitecture=CNN based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, MACS (G)=142.4, Time (s)=0.055, Params (M)=10.522025.03 | 0.5786 | 30.527 | 49.768 | 0.3566 | |
| Instant+Guidance Type=Reference-guided2026.02 | 0.584 | 20.982 | 34.82 | 0.71 | |
| StyTR2Guidance Type=Reference-guided2026.02 | 0.587 | 18.886 | 31.559 | 0.687 | |
| AdaINcategory=conventional2026.05 | 0.608 | 18.242 | 30.933 | 0.315 | |
| AdaINGuidance Type=Reference-guided2026.02 | 0.612 | 18.672 | 31.711 | 0.642 | |
| CASTcategory=conventional2026.05 | 0.621 | 20.395 | 34.685 | 0.292 | |
| MASTcategory=conventional2026.05 | 0.629 | 18.199 | 31.282 | 0.304 | |
| EFDMcategory=conventional2026.05 | 0.643 | 20.062 | 34.605 | 0.335 | |
| FreeStyleGuidance Type=Prompt-based2026.02 | 0.689 | 23.654 | 41.64 | 0.66 | |
| DiffuseITArchitecture=Diffusion based, Output Resolution=512 x 512, Hardware=NVIDIA RTX 3090 GPU, Time (s)=705.214, Params (M)=5592025.03 | 0.6954 | 37.172 | 64.702 | 0.7294 | |
| InSTGuidance Type=Reference-guided2026.02 | 0.785 | 21.541 | 40.235 | 0.881 | |
| InSTcategory=diffusion-based2026.05 | 0.8 | 21.571 | 40.633 | 0.676 | |
| SMSGuidance Type=Prompt-based2026.02 | 0.821 | 31.266 | 58.756 | 0.704 | |
| DiffStylecategory=diffusion-based2026.05 | 0.893 | 20.903 | 41.464 | 0.282 |