Semantic Segmentation on Potsdam (test)
92.12mIoUBFM (UperNet)
Evaluation results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| BFM (UperNet)General pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=1.00M (MillionAID), Parameters=2042M (ViT-G)2026.05 | 92.12 | — | 92.58 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LSKNet-SFramework=UNetFormer2024.03 | 87.2 | 93.1 | 92 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CMIDGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=1.00M (MillionAID), Parameters=24M (ResNet-50)2026.05 | 87.04 | 92.81 | 92.74 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UNetFormer2024.03 | 86.8 | 92.8 | 91.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LSKNet-TFramework=UNetFormer2024.03 | 86.7 | 92.9 | 91.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hybrid U-netGeneral pre-training=1.3M (ImageNet-1k), Parameters=11M (ResNet18)2026.05 | 86.68 | 92.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ABCNet2024.03 | 86.5 | 92.7 | 91.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BANet2024.03 | 86.3 | 92.5 | 91 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SwinUperNet2024.03 | 85.8 | 92.2 | 90.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CMIDGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=1.00M (MillionAID), Parameters=31M (Swin-B)2026.05 | 85.17 | 91.83 | 91.77 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ShelfNet2024.03 | 84.4 | 91.3 | 89.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FANet2024.03 | 84.2 | 91.3 | 89.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MARESU-Net2024.03 | 83.9 | 90.5 | 89 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SwiftNet2024.03 | 83.8 | 91 | 89.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EaNet2024.03 | 83.4 | 90.6 | 88.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STARSTrain Modality=Optical + DSM, Test Modality=Optical2026.01 | 81.76 | 89.28 | — | 86.17 | 91.69 | 76.96 | 78.57 | 75.42 | — | — | — | — | — | — | — | — | |
| BiSeNet2024.03 | 81.7 | 89.8 | 88.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MetaRSTrain Modality=Optical + DSM, Test Modality=Optical2026.01 | 80.83 | 88.63 | — | 86.64 | 91.67 | 75.08 | 76.22 | 74.53 | — | — | — | — | — | — | — | — | |
| SemiEarthRatio=10%2026.01 | 80.78 | — | — | 83.24 | 90.59 | 75.44 | 75.01 | 79.64 | — | — | — | — | — | — | — | — | |
| Segmenter2024.03 | 80.7 | 89.2 | 88.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DANet2024.03 | 80.3 | 88.9 | 89.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Baseline (Concat)Train Modality=Optical + DSM, Test Modality=Optical + DSM2026.01 | 80.09 | 88.79 | — | 85.29 | 89.74 | 75.31 | 76.64 | 73.48 | — | — | — | — | — | — | — | — | |
| DisOptNetTrain Modality=Optical + DSM, Test Modality=Optical2026.01 | 79.72 | 87.16 | — | 84.22 | 88.12 | 75.34 | 76.21 | 74.23 | — | — | — | — | — | — | — | — | |
| MMANetTrain Modality=Optical + DSM, Test Modality=Optical2026.01 | 79.72 | 87.88 | — | 84.76 | 90.69 | 72.36 | 74.79 | 74.07 | — | — | — | — | — | — | — | — | |
| Fine-tuningTrain Modality=Optical + DSM, Test Modality=Optical2026.01 | 79.64 | 87.08 | — | 84.31 | 91.51 | 73.32 | 75.05 | 74.03 | — | — | — | — | — | — | — | — | |
| DABNet2024.03 | 79.6 | 88.3 | 86.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| baseline-RGBTrain Modality=Optical, Test Modality=Optical2026.01 | 79.42 | 87.27 | — | 83.91 | 89.13 | 73.63 | 75.02 | 75.41 | — | — | — | — | — | — | — | — | |
| Full-tuning# P (M)=86.82026.02 | 79.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdaRoute# P (M)=5.22026.02 | 79.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SemiEarthRatio=5%2026.01 | 79.01 | — | — | 79.87 | 88.51 | 74.45 | 74.06 | 78.14 | — | — | — | — | — | — | — | — | |
| CSPT (UperNet)General pre-training=1.3M (ImageNet-1k), Remote sensing pre-training=1.00M (MillionAID), Parameters=86M (ViT-B)2026.05 | 78.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mona# P (M)=5.22026.02 | 78.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LOGCAN++Backbone=ResNet-502024.06 | 78.58 | 86.62 | — | 87.51 | 93.76 | 77.2 | 79.78 | 93.13 | — | — | — | — | — | — | 40.1 | 85.34 | |
| RepAdapter# P (M)=5.82026.02 | 78.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LoRand# P (M)=5.92026.02 | 78.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AdaptFormer# P (M)=5.42026.02 | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FarSegBackbone=ResNet-502024.06 | 77.98 | 86.29 | — | 86.93 | 93.56 | 76.72 | 79 | 91.37 | — | — | — | — | — | — | 40.32 | 84.9 | |
| SNELL# P (M)=5.82026.02 | 77.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PSPNetBackbone=ResNet-502024.06 | 77.87 | 86.08 | — | 86.96 | 93.59 | 76.44 | 79.06 | 92.62 | — | — | — | — | — | — | 38.55 | 84.68 | |
| RSSFormerBackbone=RSS-B2024.06 | 77.85 | 86.01 | — | 86.92 | 93.12 | 77.19 | 79.55 | 91.43 | — | — | — | — | — | — | 38.88 | 84.55 | |
| DDPBackbone=Swin-T2024.06 | 77.77 | 86.07 | — | 86.88 | 93.5 | 75.63 | 78.79 | 92.59 | — | — | — | — | — | — | 39.24 | 84.78 | |
| UnetFormerBackbone=ResNet-502024.06 | 77.48 | 85.81 | — | 86.5 | 93.33 | 75.8 | 78.77 | 92.43 | — | — | — | — | — | — | 38.06 | 84.48 | |
| LoRA# P (M)=5.42026.02 | 77.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DeepLabV3+Backbone=ResNet-502024.06 | 77.33 | 85.87 | — | 85.99 | 92.44 | 75.49 | 77.84 | 92.18 | — | — | — | — | — | — | 40.06 | 84.99 | |
| Semantic FPNBackbone=ResNet-502024.06 | 77.32 | 85.4 | — | 86.95 | 93.63 | 76.71 | 79.85 | 92.63 | — | — | — | — | — | — | 34.15 | 83.94 | |
| OCRNetBackbone=HRNet-322024.06 | 77.29 | 85.91 | — | 86.01 | 91.66 | 75.91 | 78.49 | 91 | — | — | — | — | — | — | 40.66 | 85 | |
| VPT# P (M)=0.12026.02 | 77.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SemiEarthRatio=1%2026.01 | 77.02 | — | — | 79.01 | 86.8 | 71.22 | 71.96 | 76.11 | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-502024.06 | 76.68 | 85.16 | — | 85.92 | 93.13 | 74.62 | 78.63 | 91.91 | — | — | — | — | — | — | 35.9 | 83.92 | |
| CBC-SLPModality Availability=Full, Training Setting=Random modality dropout2026.04 | 76.6 | 81.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DIS2Modality=Full Modality2026.01 | 76.42 | 86.27 | — | — | — | — | — | — | 88.56 | 94.99 | 80.21 | 80.6 | 86.98 | — | — | — | |
| ERFNet2024.03 | 76.2 | 85.8 | 84.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Cross-scale MAEGeneral pre-training=14.0M (ImageNet-21k), Remote sensing pre-training=1.00M (fMoW), Parameters=307M (ViT-L)2026.05 | 76.17 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CBC-SLPModality Availability=DSM missing, Training Setting=Random modality dropout2026.04 | 76.1 | 80.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CBC-SLPModality Availability=RGB missing, Training Setting=Random modality dropout2026.04 | 75.9 | 80.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MUCARatio=10%2026.01 | 75.65 | — | — | 79.92 | 88.02 | 70.58 | 64.53 | 75.2 | — | — | — | — | — | — | — | — | |
| DC-SwinBackbone=Swin-B2024.06 | 75.12 | 84.18 | — | 84.88 | 90.47 | 74.95 | 76.42 | 89.16 | — | — | — | — | — | — | 34.83 | 82.8 | |
| CBC-SLPModality Availability=IRRG only, Training Setting=Random modality dropout2026.04 | 75 | 79.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AllSparkRatio=10%2026.01 | 74.76 | — | — | 78.31 | 86.29 | 69.83 | 64.17 | 75.23 | — | — | — | — | — | — | — | — | |
| MUCARatio=5%2026.01 | 74.62 | — | — | 79.56 | 88.45 | 69.53 | 61.39 | 74.18 | — | — | — | — | — | — | — | — | |
| DWLRatio=10%2026.01 | 74.57 | — | — | 77.14 | 76.37 | 88.42 | 66.54 | 64.37 | — | — | — | — | — | — | — | — | |
| FixMatchRatio=10%2026.01 | 74.27 | — | — | 76.14 | 77.97 | 76.17 | 70.09 | 70.97 | — | — | — | — | — | — | — | — | |
| Mean teacherRatio=10%2026.01 | 74.21 | — | — | 76.51 | 84.76 | 69.28 | 68.83 | 71.66 | — | — | — | — | — | — | — | — | |
| LSSTRatio=10%2026.01 | 74.2 | — | — | 74.89 | 70.92 | 86.06 | 68.91 | 70.22 | — | — | — | — | — | — | — | — | |
| CPSRatio=10%2026.01 | 74.09 | — | — | 75.89 | 77.8 | 87.15 | 61.12 | 68.48 | — | — | — | — | — | — | — | — | |
| UniMatchRatio=10%2026.01 | 73.8 | — | — | 76.46 | 77.34 | 87.75 | 70.79 | 56.65 | — | — | — | — | — | — | — | — | |
| GEMMNetModality=Full Modality2026.01 | 73.42 | 84.28 | — | — | — | — | — | — | 87.21 | 94.06 | 78.43 | 78.98 | 82.7 | — | — | — | |
| EfficientViTBackbone=EfficientViT-L22024.06 | 73.38 | 82.77 | — | 83.58 | 89.38 | 73.91 | 73.77 | 88.74 | — | — | — | — | — | — | 30.92 | 81.25 | |
| DWLRatio=5%2026.01 | 73.1 | — | — | 75.68 | 74.81 | 85.64 | 66.38 | 62.99 | — | — | — | — | — | — | — | — | |
| AllSparkRatio=5%2026.01 | 72.88 | — | — | 77.15 | 85.57 | 67.62 | 60.61 | 73.48 | — | — | — | — | — | — | — | — | |
| Foundation ModelRemote sensing pre-training=0.28M, Parameters=31M (Swin-T)2026.05 | 72.87 | 87.48 | 87.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FixMatchRatio=5%2026.01 | 72.75 | — | — | 75.3 | 78.12 | 74.87 | 68.89 | 66.58 | — | — | — | — | — | — | — | — | |
| CPSRatio=5%2026.01 | 72.74 | — | — | 75.39 | 76.53 | 84.34 | 57.98 | 69.45 | — | — | — | — | — | — | — | — | |
| LSSTRatio=5%2026.01 | 72.5 | — | — | 73.86 | 69.26 | 84.55 | 67.33 | 67.49 | — | — | — | — | — | — | — | — | |
| Mean teacherRatio=5%2026.01 | 72.4 | — | — | 74.6 | 82.15 | 65.92 | 67.11 | 72.21 | — | — | — | — | — | — | — | — | |
| M3LModality Availability=Full, Training Setting=Random modality dropout2026.04 | 72.2 | 77.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UniMatchRatio=5%2026.01 | 72.14 | — | — | 75.07 | 78.24 | 73.59 | 67.17 | 66.64 | — | — | — | — | — | — | — | — | |
| mmformerModality=Full Modality2026.01 | 71.79 | 83.11 | — | — | — | — | — | — | 86.18 | 93.44 | 75.84 | 77.02 | 83.07 | — | — | — | |
| Dformerv2-SModality Availability=Full, Training Setting=Random modality dropout2026.04 | 71.6 | 76 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MUCARatio=1%2026.01 | 71.33 | — | — | 76.64 | 84.56 | 66.98 | 56.96 | 71.52 | — | — | — | — | — | — | — | — | |
| CBC-SLPModality Availability=IRRG missing, Training Setting=Random modality dropout2026.04 | 71.3 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Dformerv2-SModality Availability=RGB missing, Training Setting=Random modality dropout2026.04 | 71.2 | 75.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Dformerv2-SModality Availability=DSM missing, Training Setting=Random modality dropout2026.04 | 71.2 | 75.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| AllSparkRatio=1%2026.01 | 70.87 | — | — | 75.31 | 83.7 | 65.92 | 59.64 | 69.77 | — | — | — | — | — | — | — | — | |
| M3LModality Availability=DSM missing, Training Setting=Random modality dropout2026.04 | 70.8 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimSiamTransferModality=Full Modality2026.01 | 70.77 | 82.37 | — | — | — | — | — | — | 86.16 | 93.49 | 76.4 | 76.72 | 79.09 | — | — | — | |
| Dformerv2-SModality Availability=IRRG only, Training Setting=Random modality dropout2026.04 | 70.7 | 75.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| M3LModality Availability=RGB missing, Training Setting=Random modality dropout2026.04 | 70.5 | 75.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UniMatchRatio=1%2026.01 | 70.44 | — | — | 72.64 | 76.52 | 70.99 | 65.44 | 66.62 | — | — | — | — | — | — | — | — | |
| FixMatchRatio=1%2026.01 | 70.38 | — | — | 72.81 | 76.95 | 71.59 | 64.71 | 65.85 | — | — | — | — | — | — | — | — | |
| CCTRatio=10%2026.01 | 70.33 | — | — | 73.06 | 73.09 | 83.94 | 61.12 | 60.45 | — | — | — | — | — | — | — | — | |
| DIS2Modality=Missing NDSM - Only RGIR2026.01 | 70.05 | 82.12 | — | — | — | — | — | — | 84.2 | 86.27 | 77.13 | 77.42 | 85.56 | — | — | — | |
| CCTRatio=5%2026.01 | 70.02 | — | — | 74.42 | 72.9 | 80.25 | 64.23 | 58.32 | — | — | — | — | — | — | — | — | |
| MMANetModality Availability=Full, Training Setting=Random modality dropout2026.04 | 69.9 | 74.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| M3LModality Availability=IRRG missing, Training Setting=Random modality dropout2026.04 | 69.9 | 75 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimSiamModality=Full Modality2026.01 | 69.78 | 81.68 | — | — | — | — | — | — | 86.43 | 91.77 | 75.81 | 72.61 | 81.79 | — | — | — | |
| ShaSpecModality=Full Modality2026.01 | 69.4 | 81.34 | — | — | — | — | — | — | 86.01 | 90.79 | 73.84 | 74.76 | 81.31 | — | — | — | |
| DWLRatio=1%2026.01 | 69.39 | — | — | 72.22 | 72.34 | 77.08 | 62.74 | 62.57 | — | — | — | — | — | — | — | — | |
| MMANetModality Availability=RGB missing, Training Setting=Random modality dropout2026.04 | 69.2 | 74.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CBC-SLPModality Availability=RGB only, Training Setting=Random modality dropout2026.04 | 69.1 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — |