Dichotomous Image Segmentation on DIS TE1 500 (test)
89.1FmaxPDFNet
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PDFNetBackbone=Swin-B, Params=94M2025.03 | 89.1 | 84.8 | 92.8 | 89.9 | 0.03 | |
| DiffDISBackbone=SD-Turbo, Params=865M+2025.03 | 88.3 | 86.2 | 93.3 | 89.1 | 0.03 | |
| MVANet*Backbone=Swin-B, Params=93M, Input=RGB-depth concatenation2025.03 | 88.1 | 83.7 | 92.2 | 89.4 | 0.031 | |
| MVANetBackbone=Swin-B, Params=93M2025.03 | 87.3 | 82.3 | 91.1 | 87.9 | 0.037 | |
| BiRefNetBackbone=Swin-L, Params=215M2025.03 | 86.6 | 81.9 | 91.1 | 88.5 | 0.037 | |
| CPNetBackbone=Swin-B & Swin-B, Params=216M2025.03 | 86.2 | 81.9 | 91 | 88 | 0.037 | |
| DACODBackbone=ResNet-50 & Swin-B, Params=189M2025.03 | 85.2 | 80 | 90.6 | 87 | 0.038 | |
| GenPerceptBackbone=SD U-Net, Params=865M+2025.03 | 85 | 82.7 | 91.9 | 87.8 | 0.036 | |
| InSPyReNetBackbone=Swin-B, Params=88M2025.03 | 84.5 | 78.8 | 89.4 | 87.3 | 0.043 | |
| MAGNetBackbone=SMT-T & MbNet-V2, Params=16M2025.03 | 83.8 | 79 | 89.9 | 86.2 | 0.044 | |
| RISNetBackbone=PVT-V2-B2, Params=25M2025.03 | 82.4 | 77.6 | 89 | 85.9 | 0.044 | |
| FP-DIS2025.03 | 78.4 | 71.3 | 86 | 82.1 | 0.06 | |
| UDUNBackbone=ResNet-50, Params=25M2025.03 | 78.4 | 72 | 86.4 | 81.7 | 0.059 | |
| IS-NetBackbone=U2-Net, Params=44M2025.03 | 74 | 66.2 | 82 | 78.7 | 0.074 |