Dichotomous Image Segmentation on DIS-TE3 500 (test)
93.6FmaxPDFNet
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PDFNetBackbone=Swin-B, Params=94M2025.03 | 93.6 | 90.1 | 95.7 | 92.8 | 0.027 | |
| DiffDISBackbone=SD-Turbo, Params=865M+2025.03 | 93.4 | 91.6 | 96.4 | 91.9 | 0.025 | |
| MVANetBackbone=Swin-B, Params=93M2025.03 | 92.9 | 89 | 95.4 | 92 | 0.031 | |
| MVANet*Backbone=Swin-B, Params=93M, Input=RGB-depth concatenation2025.03 | 92.9 | 88.9 | 94.9 | 91.9 | 0.029 | |
| CPNetBackbone=Swin-B & Swin-B, Params=216M2025.03 | 92.2 | 88.7 | 95 | 91.6 | 0.028 | |
| BiRefNetBackbone=Swin-L, Params=215M2025.03 | 92 | 89.3 | 95.5 | 91.9 | 0.028 | |
| InSPyReNetBackbone=Swin-B, Params=88M2025.03 | 91.9 | 87.1 | 94 | 91.8 | 0.034 | |
| DACODBackbone=ResNet-50 & Swin-B, Params=189M2025.03 | 90.4 | 85.5 | 93.8 | 90.1 | 0.037 | |
| GenPerceptBackbone=SD U-Net, Params=865M+2025.03 | 89.8 | 87.9 | 95.4 | 89.6 | 0.032 | |
| MAGNetBackbone=SMT-T & MbNet-V2, Params=16M2025.03 | 89.3 | 85 | 93.5 | 89.3 | 0.039 | |
| FP-DIS2025.03 | 86.8 | 81.1 | 92.2 | 87.1 | 0.049 | |
| UDUNBackbone=ResNet-50, Params=25M2025.03 | 86.5 | 80.9 | 91.7 | 86.5 | 0.05 | |
| RISNetBackbone=PVT-V2-B2, Params=25M2025.03 | 85.3 | 81 | 91.3 | 86.9 | 0.046 | |
| IS-NetBackbone=U2-Net, Params=44M2025.03 | 83 | 75.8 | 88.3 | 83.6 | 0.064 |