Skin Lesion Segmentation on ISIC 2018 (test)
92.3Dice ScoreDuAT
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DuATResolution=352x3522022.12 | 92.3 | — | — | — | — | — | 86.7 | 0.029 | — | — | — | — | — | — | |
| Siamese-DiffusionArchitecture=SegFormer, Training Set=Real + Synthetic, Mask Prior=Random2025.05 | 91.93 | — | — | — | — | — | — | — | 87.12 | — | — | — | — | — | |
| Siamese-DiffusionArchitecture=SegFormer, Training Set=Real + Synthetic, Mask Prior=Real2025.05 | 91.8 | — | — | — | — | — | — | — | 86.67 | — | — | — | — | — | |
| ControlNetArchitecture=SegFormer, Training Set=Real + Synthetic2025.05 | 91.52 | — | — | — | — | — | — | — | 86.19 | — | — | — | — | — | |
| Real DatasetArchitecture=SegFormer, Training Set=Real Only2025.05 | 91.35 | — | — | — | — | — | — | — | 86.1 | — | — | — | — | — | |
| Polyp-PVTResolution=352x3522022.12 | 91.3 | — | — | — | — | — | 85.2 | 0.032 | — | — | — | — | — | — | |
| PolypPVT2023.10 | 91.3 | — | — | — | — | — | — | — | 85.2 | — | — | — | — | — | |
| Laplacian-Formerbridge=included2023.08 | 91.28 | — | 92.9 | 97.15 | 96.26 | — | — | — | — | — | — | — | — | — | |
| UGDD-Net (Ours)Backbone=Dual-Domain, Architecture Type=Dual-Path, Mechanism=Spatial-Frequency (SF) + Uncertainty-Aware (U)2026.05 | 91.26 | — | — | — | — | — | 84.69 | — | — | — | — | 9.86 | 3.53 | 10.03 | |
| Copy-PasteArchitecture=SegFormer, Training Set=Real + Synthetic2025.05 | 91.25 | — | — | — | — | — | — | — | 86.09 | — | — | — | — | — | |
| T2I-AdapterArchitecture=SegFormer, Training Set=Real + Synthetic2025.05 | 91.19 | — | — | — | — | — | — | — | 85.73 | — | — | — | — | — | |
| PVT-CASCADE2023.10 | 91.1 | — | — | — | — | — | — | — | 84.9 | — | — | — | — | — | |
| Laplacian-Formerbridge=excluded2023.08 | 91 | — | 92.89 | 96.55 | 96.11 | — | — | — | — | — | — | — | — | — | |
| EAM-NetPara=4.6M, Flops=16.85G2025.12 | 90.71 | — | — | — | 96.45 | 91.77 | 84.18 | — | — | — | — | — | — | — | |
| TMU-Net2023.08 | 90.59 | — | 90.38 | 97.46 | 96.03 | — | — | — | — | — | — | — | — | — | |
| UCTransNetResolution=352x3522022.12 | 90.5 | — | — | — | — | — | 83 | 0.035 | — | — | — | — | — | — | |
| UCTransNet2023.10 | 90.5 | — | — | — | — | — | — | — | 83 | — | — | — | — | — | |
| MADGNetTrain Dataset=ISIC20182024.05 | 90.2 | — | — | 82 | — | — | — | 3.6 | 83.7 | 89.2 | 94.1 | — | — | — | |
| PVT-CASCADEPara=35.27M, Flops=8.20G2025.12 | 90.12 | — | — | — | 95.62 | 87.65 | 82.83 | — | — | — | — | — | — | — | |
| Attention-MambaParam (M)=14.05, FLOPs (G)=8.94, Training Size=256 × 2562024.02 | 90.12 | — | 90.75 | 97.16 | 96.04 | 92.29 | 83.51 | — | — | — | — | — | — | — | |
| TransFuseResolution=352x3522022.12 | 90.1 | — | — | — | — | — | 84 | 0.035 | — | — | — | — | — | — | |
| TransFuse2023.10 | 90.1 | — | — | — | — | — | — | — | 84 | — | — | — | — | — | |
| Samba+2026.02 | 90.05 | — | — | 97.44 | 95.25 | — | 81.91 | — | — | — | — | — | — | — | |
| Parallel MERITPara=147.9M, Flops=33.43G2025.12 | 90.05 | — | — | — | 95.66 | 89.99 | 83.44 | — | — | — | — | — | — | — | |
| MulitTransParam (M)=39.38, FLOPs (G)=23.88, Training Size=256 × 2562024.02 | 90.03 | — | 90.95 | 96.9 | 96.07 | 92.09 | 83.43 | — | — | — | — | — | — | — | |
| Swin_UMambaParam (M)=59.89, FLOPs (G)=43.94, Training Size=256 × 2562024.02 | 89.95 | — | 91.86 | 96.83 | 96.01 | 91.02 | 83.33 | — | — | — | — | — | — | — | |
| H2FormerParam (M)=33.68, FLOPs (G)=32.25, Training Size=256 × 2562024.02 | 89.91 | — | 91.77 | 96.77 | 96.07 | 91.07 | 83.3 | — | — | — | — | — | — | — | |
| UDELBackbone=CNN+Mamba, Architecture Type=Dual-Path, Mechanism=Uncertainty-Aware (U)2026.05 | 89.84 | — | — | — | — | — | 83.12 | — | — | — | — | 11.32 | 4.78 | 12.49 | |
| HiFormerParam (M)=37.46, FLOPs (G)=17.51, Training Size=224 × 2242024.02 | 89.81 | — | 91.35 | 96.49 | 96.11 | 91.19 | 83.15 | — | — | — | — | — | — | — | |
| CDMT-UNetBackbone=CNN+Trans, Architecture Type=Single-Path, Mechanism=Spatial-Frequency (SF)2026.05 | 89.8 | — | — | — | — | — | 82.83 | — | — | — | — | 11.83 | 4.96 | 16.71 | |
| SFma-UNetBackbone=Mamba, Architecture Type=Dual-Path, Mechanism=Spatial-Frequency (SF)2026.05 | 89.75 | — | — | — | — | — | 82.86 | — | — | — | — | 11.4 | 4.84 | 16.16 | |
| VM-UNetV2year=20242026.02 | 89.73 | — | — | 97.13 | 95.06 | — | 81.37 | — | — | — | — | — | — | — | |
| CFFormerBackbone=CNN+Trans, Architecture Type=Dual-Path2026.05 | 89.72 | — | — | — | — | — | 82.71 | — | — | — | — | 13.84 | 5.48 | 16.57 | |
| VM-UNetyear=20242026.02 | 89.71 | — | — | 96.13 | 94.91 | — | 81.35 | — | — | — | — | — | — | — | |
| SwinUNetParam (M)=27.17, FLOPs (G)=5.95, Training Size=224 × 2242024.02 | 89.68 | — | 91.43 | 96.76 | 95.99 | 91.1 | 83.04 | — | — | — | — | — | — | — | |
| SkinMambaBackbone=CNN+Mamba, Architecture Type=Single-Path, Mechanism=Spatial-Frequency (SF)2026.05 | 89.64 | — | — | — | — | — | 82.66 | — | — | — | — | 11.46 | 4.87 | 15.86 | |
| CPF-NetPara=30.65M, Flops=8.83G2025.12 | 89.63 | — | — | — | 96.02 | 90.71 | 82.92 | — | — | — | — | — | — | — | |
| VM_UNetParam (M)=27.43, FLOPs (G)=4.11, Training Size=256 × 2562024.02 | 89.62 | — | 91.42 | 96.85 | 95.96 | 90.96 | 83.01 | — | — | — | — | — | — | — | |
| CE-NetPara=29.0M, Flops=9.76G2025.12 | 89.59 | — | — | — | 95.97 | 90.67 | 82.82 | — | — | — | — | — | — | — | |
| G-CASCADEPara=141.4M, Flops=30.42G2025.12 | 89.57 | — | — | — | 95.74 | 89.73 | 82.98 | — | — | — | — | — | — | — | |
| DPMF-NetBackbone=CNN+Trans, Architecture Type=Dual-Path, Mechanism=Spatial-Frequency (SF)2026.05 | 89.56 | — | — | — | — | — | 82.47 | — | — | — | — | 11.73 | 5.01 | 15.33 | |
| VM_UNet_V2Param (M)=22.77, FLOPs (G)=4.40, Training Size=256 × 2562024.02 | 89.51 | — | 90.47 | 97.2 | 95.9 | 91.44 | 82.7 | — | — | — | — | — | — | — | |
| Swin-Unet2023.08 | 89.46 | — | 90.56 | 97.98 | 96.05 | — | — | — | — | — | — | — | — | — | |
| EGE-UNetyear=20232026.02 | 89.46 | — | — | 97.01 | 94.92 | — | 80.94 | — | — | — | — | — | — | — | |
| EMCADPara=30.0M, Flops=6.36G2025.12 | 89.46 | — | — | — | 95.55 | 89.25 | 82.81 | — | — | — | — | — | — | — | |
| GCA-ResUNet2025.12 | 89.41 | — | 91.39 | 96.75 | 94.67 | — | — | — | 81.05 | — | — | — | — | — | |
| DenseASPPPara=33.7M, Flops=107.74G2025.12 | 89.35 | — | — | — | 95.89 | 91.38 | 82.52 | — | — | — | — | — | — | — | |
| UNetV2year=20252026.02 | 89.32 | — | — | 96.94 | 94.86 | — | 80.71 | — | — | — | — | — | — | — | |
| F2CAU-NetBackbone=CNN, Architecture Type=Dual-Path, Mechanism=Uncertainty-Aware (U)2026.05 | 89.32 | — | — | — | — | — | 82.34 | — | — | — | — | 12.48 | 5.15 | 13.17 | |
| TransFuseyear=20212026.02 | 89.27 | — | — | 95.74 | 94.66 | — | 80.63 | — | — | — | — | — | — | — | |
| TransFuse2025.12 | 89.27 | — | 91.28 | 95.74 | 94.66 | — | — | — | 80.63 | — | — | — | — | — | |
| M2SNetTrain Dataset=ISIC20182024.05 | 89.2 | — | — | 81.8 | — | — | — | 3.7 | 83.4 | 88.9 | 93.8 | — | — | — | |
| DPGNetBackbone=CNN+Trans, Architecture Type=Single-Path, Mechanism=Uncertainty-Aware (U)2026.05 | 89.11 | — | — | — | — | — | 81.98 | — | — | — | — | 13.45 | 5.41 | 13.52 | |
| CENetTrain Dataset=ISIC20182024.05 | 89.1 | — | — | 81.3 | — | — | — | 4.3 | 82.1 | 88.1 | 93 | — | — | — | |
| EffFormer2023.08 | 89.09 | — | 90.34 | 97.01 | 95.79 | — | — | — | — | — | — | — | — | — | |
| MALUNet2025.12 | 89.04 | — | 89.74 | 96.19 | 94.62 | — | — | — | 80.25 | — | — | — | — | — | |
| FAT-Net2023.08 | 89.03 | — | 91 | 96.99 | 95.78 | — | — | — | — | — | — | — | — | — | |
| DCSAUNetTrain Dataset=ISIC20182024.05 | 89 | — | — | 81.4 | — | — | — | 4.4 | 82 | 87.8 | 92.9 | — | — | — | |
| MISSFormerParam (M)=42.46, FLOPs (G)=7.28, Training Size=224 × 2242024.02 | 89 | — | 90.65 | 96.35 | 95.54 | 91.15 | 82.29 | — | — | — | — | — | — | — | |
| UNetPlusPlusParam (M)=9.16, FLOPs (G)=34.90, Training Size=256 × 2562024.02 | 88.97 | — | 90.5 | 97.12 | 95.57 | 90.76 | 81.88 | — | — | — | — | — | — | — | |
| DEviSBackbone=CNN, Architecture Type=Single-Path, Mechanism=Uncertainty-Aware (U)2026.05 | 88.96 | — | — | — | — | — | 81.92 | — | — | — | — | 13.42 | 5.37 | 12.75 | |
| BDFormerBackbone=Trans, Architecture Type=Dual-Path2026.05 | 88.91 | — | — | — | — | — | 81.83 | — | — | — | — | 12.56 | 5.25 | 17.25 | |
| FRCUNetTrain Dataset=ISIC20182024.05 | 88.9 | — | — | 82 | — | — | — | 3.7 | 83.1 | 89.3 | 93.9 | — | — | — | |
| DualAttenUNetParam (M)=19.57, FLOPs (G)=28.88, Training Size=256 × 2562024.02 | 88.9 | — | 91.27 | 96.74 | 95.53 | 90.32 | 81.98 | — | — | — | — | — | — | — | |
| AHF-U-NetBackbone=CNN, Architecture Type=Single-Path, Mechanism=Uncertainty-Aware (U)2026.05 | 88.84 | — | — | — | — | — | 81.71 | — | — | — | — | 13.72 | 5.47 | 13.89 | |
| U-NetPara=32.9M, Flops=65.53G2025.12 | 88.81 | — | — | — | 95.68 | 91.31 | 81.69 | — | — | — | — | — | — | — | |
| AttenUNetParam (M)=34.88, FLOPs (G)=66.63, Training Size=256 × 2562024.02 | 88.8 | — | 90.58 | 96.77 | 95.45 | 90.49 | 81.77 | — | — | — | — | — | — | — | |
| DSU-NetBackbone=CNN+Trans, Architecture Type=Dual-Path2026.05 | 88.74 | — | — | — | — | — | 81.6 | — | — | — | — | 14.27 | 5.7 | 16.58 | |
| HiFormerTrain Dataset=ISIC20182024.05 | 88.7 | — | — | 80.8 | — | — | — | 4.4 | 81.9 | 87.6 | 92.6 | — | — | — | |
| MPBA-NetBackbone=CNN+Trans, Architecture Type=Single-Path2026.05 | 88.7 | — | — | — | — | — | 81.37 | — | — | — | — | 12.88 | 5.33 | 16.21 | |
| WinGraphUNetBackbone=CNN+Graph, Architecture Type=Single-Path2026.05 | 88.69 | — | — | — | — | — | 81.42 | — | — | — | — | 11.32 | 4.76 | 16.22 | |
| SANet2025.12 | 88.59 | — | 89.46 | 95.97 | 94.39 | — | — | — | 79.52 | — | — | — | — | — | |
| UnetParam (M)=31.04, FLOPs (G)=54.74, Training Size=256 × 2562024.02 | 88.42 | — | 90.85 | 96.85 | 95.32 | 89.67 | 81.22 | — | — | — | — | — | — | — | |
| DMA-NetBackbone=CNN+Mamba, Architecture Type=Dual-Path2026.05 | 88.41 | — | — | — | — | — | 81.02 | — | — | — | — | 13.87 | 5.55 | 15.93 | |
| BCDU-NetPara=28.8M, Flops=360.18G2025.12 | 88.33 | — | — | — | 95.48 | 89.68 | 80.84 | — | — | — | — | — | — | — | |
| UTNetV22025.12 | 88.25 | — | 87.6 | 96.48 | 94.32 | — | — | — | 78.97 | — | — | — | — | — | |
| SF-UNetBackbone=CNN, Architecture Type=Dual-Path, Mechanism=Spatial-Frequency (SF)2026.05 | 88.23 | — | — | — | — | — | 80.94 | — | — | — | — | 12.39 | 5.08 | 14.96 | |
| MSRFNetTrain Dataset=ISIC20182024.05 | 88.2 | — | — | 80.7 | — | — | — | 4.7 | 81.3 | 86.9 | 92 | — | — | — | |
| Mamba_UNetParam (M)=19.12, FLOPs (G)=4.60, Training Size=256 × 2562024.02 | 88.18 | — | 89.46 | 96.99 | 95.22 | 90.83 | 81.02 | — | — | — | — | — | — | — | |
| VM-UNetBackbone=Mamba, Architecture Type=Single-Path2026.05 | 88.15 | — | — | — | — | — | 80.79 | — | — | — | — | 14.48 | 5.77 | 15.43 | |
| Swin-UnetBackbone=Trans, Architecture Type=Single-Path2026.05 | 88.09 | — | — | — | — | — | 80.73 | — | — | — | — | 16.28 | 6.06 | 17.44 | |
| H-NetBackbone=CNN, Architecture Type=Dual-Path2026.05 | 88.01 | — | — | — | — | — | 80.44 | — | — | — | — | 14.89 | 5.85 | 15.28 | |
| TransUNetResolution=352x3522022.12 | 88 | — | — | — | — | — | 80.9 | 0.036 | — | — | — | — | — | — | |
| TransUNet2023.10 | 88 | — | — | — | — | — | — | — | 80.9 | — | — | — | — | — | |
| Att-UNet2025.12 | 87.91 | — | 87.6 | 96.23 | 94.13 | — | — | — | 78.43 | — | — | — | — | — | |
| UNet++2025.12 | 87.83 | — | 88.65 | 95.75 | 94.02 | — | — | — | 78.31 | — | — | — | — | — | |
| AttUNetTrain Dataset=ISIC20182024.05 | 87.8 | — | — | 80.5 | — | — | — | 4.5 | 80.5 | 86.5 | 92 | — | — | — | |
| U-NetBackbone=CNN, Architecture Type=Single-Path2026.05 | 87.77 | — | — | — | — | — | 80.21 | — | — | — | — | 15.51 | 5.88 | 14.75 | |
| U-Net2025.12 | 87.55 | — | 85.86 | 96.69 | 94.05 | — | — | — | 77.86 | — | — | — | — | — | |
| PraNetResolution=352x3522022.12 | 87.5 | — | — | — | — | — | 78.7 | 0.037 | — | — | — | — | — | — | |
| PraNet2023.10 | 87.5 | — | — | — | — | — | — | — | 78.7 | — | — | — | — | — | |
| UNetTrain Dataset=ISIC20182024.05 | 87.3 | — | — | 80.4 | — | — | — | 4.7 | 80.2 | 87.9 | 91.3 | — | — | — | |
| UNet++Train Dataset=ISIC20182024.05 | 87.3 | — | — | 80.1 | — | — | — | 4.7 | 80.2 | 86.3 | 91.6 | — | — | — | |
| TransUNetTrain Dataset=ISIC20182024.05 | 87.3 | — | — | 80.8 | — | — | — | 4.2 | 81.2 | 88.6 | 91.9 | — | — | — | |
| CaraNetResolution=352x3522022.12 | 87 | — | — | — | — | — | 78.2 | 0.038 | — | — | — | — | — | — | |
| CaraNet2023.10 | 87 | — | — | — | — | — | — | — | 78.2 | — | — | — | — | — | |
| Att-UNet2023.08 | 85.66 | — | 86.74 | 98.63 | 93.76 | — | — | — | — | — | — | — | — | — | |
| U-NetResolution=352x3522022.12 | 85.5 | — | — | — | — | — | 78.5 | 0.045 | — | — | — | — | — | — | |
| UNet2023.10 | 85.5 | — | — | — | — | — | — | — | 78.5 | — | — | — | — | — | |
| U-Net2023.08 | 85.45 | — | 88 | 96.97 | 94.04 | — | — | — | — | — | — | — | — | — |