Field Boundary Segmentation on FTW (test)
76Pixel IoUPRUE
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| PRUEBackbone=U-Net+EfficientNet-B72026.03 | 76 | 89 | 83 | 62 | 40 | 47 | 26 | 40 | 67.1 | 306.94 | |
| DECODEBackbone=FracTALResUNet2026.03 | 71 | 83 | 83 | 27 | 17 | 21 | 9 | 17 | 64.8 | 113.47 | |
| FTW-BaselineBackbone=U-Net+EfficientNet-B32026.03 | 70 | 90 | 72 | 40 | 37 | 38 | 22 | 39 | 13.2 | 623.28 | |
| Mask2FormerBackbone=Swin-S, Inference Protocol=fine-tuned, Input channels=8ch2026.03 | 68 | 88 | 75 | 62 | 30 | 39 | 28 | 44 | 68.8 | 26.66 | |
| ClayBackbone=ViT-Large2026.03 | 67 | 91 | 72 | 38 | 36 | 36 | 24 | 41 | 363.8 | 10.98 | |
| GalileoBackbone=ViT-Base2026.03 | 66 | 86 | 72 | 29 | 36 | 32 | 21 | 37 | 119 | — | |
| DINOv3Backbone=ViT-Large2026.03 | 60 | 90 | 64 | 39 | 27 | 31 | 20 | 35 | 412.2 | 46.59 | |
| TerraMindBackbone=ViT-Base2026.03 | 57 | 88 | 62 | 30 | 24 | 26 | 17 | 31 | 189.1 | 123.12 | |
| TerraFMBackbone=ViT-Base2026.03 | 57 | 86 | 62 | 29 | 23 | 25 | 16 | 29 | 218.6 | 110.68 | |
| Prithvi 2.0Backbone=ViT-Large2026.03 | 56 | 88 | 60 | 29 | 23 | 25 | 16 | 30 | 439.5 | 63.38 | |
| CROMABackbone=ViT-Base2026.03 | 52 | 86 | 56 | 26 | 18 | 21 | 13 | 25 | 137 | 133.1 | |
| SoftConBackbone=ViT-Small2026.03 | 52 | 85 | 57 | 24 | 18 | 21 | 12 | 24 | 101.8 | 234.18 | |
| DeCURBackbone=ViT-Small2026.03 | 49 | 85 | 53 | 23 | 16 | 19 | 11 | 22 | 120.2 | 271.67 | |
| DOFA-v1Backbone=ViT-Large2026.03 | 49 | 85 | 53 | 21 | 15 | 17 | 10 | 19 | 446 | 64.85 | |
| SAMBackbone=ViT-Huge, Inference Protocol=fine-tuned, Input channels=8ch2026.03 | 45 | 73 | 54 | 56 | 34 | 37 | 21 | 19 | 642.7 | 0.17 | |
| SatlasBackbone=Swin-Tiny2026.03 | 45 | 79 | 50 | 13 | 11 | 12 | 7 | 14 | 131.7 | 79.85 | |
| Del-Any SBackbone=YOLOv11, Inference Protocol=zero-shot, Input channels=3ch, Window setting=win A2026.03 | 44 | 52 | 73 | 15 | 6 | 8 | 7 | 14 | 2.6 | 389.24 | |
| Del-AnyBackbone=YOLOv11, Inference Protocol=zero-shot, Input channels=3ch, Window setting=win A2026.03 | 37 | 53 | 56 | 25 | 5 | 9 | 5 | 10 | 56.9 | 87.32 | |
| SAMBackbone=ViT-Huge, Inference Protocol=zero-shot, Input channels=3ch2026.03 | 32 | 36 | 82 | 14 | 34 | 17 | 6 | 12 | 641.1 | 0.17 |