Open-Vocabulary Segmentation on USIS16K
60.45Average Accuracy (aAcc)Earth2Ocean
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Earth2OceanBackbone=ViT-H/14, Publication=–2025.11 | 60.45 | 45.68 | 59.99 | |
| Earth2OceanBackbone=ViT-L/14, Publication=–2025.11 | 59.97 | 45.13 | 58.04 | |
| TridentBackbone=ViT-H/14, Publication=ICCV20252025.11 | 57.94 | 42.43 | 56.46 | |
| CorrCLIPBackbone=ViT-H/14, Publication=ICCV20252025.11 | 57.82 | 42.35 | 56.01 | |
| ProxyCLIPBackbone=ViT-H/14, Publication=ECCV20242025.11 | 53.12 | 37.89 | 52.42 | |
| ClearCLIPBackbone=ViT-H/14, Publication=ECCV20242025.11 | 51.34 | 36.08 | 49.07 | |
| TridentBackbone=ViT-L/14, Publication=ICCV20252025.11 | 47.22 | 32.13 | 43.86 | |
| CorrCLIPBackbone=ViT-L/14, Publication=ICCV20252025.11 | 46.99 | 31.32 | 43.59 | |
| Earth2OceanBackbone=ViT-B/16, Publication=–2025.11 | 45.42 | 29.02 | 43.03 | |
| ProxyCLIPBackbone=ViT-L/14, Publication=ECCV20242025.11 | 45.23 | 30.55 | 42.35 | |
| CorrCLIPBackbone=ViT-B/16, Publication=ICCV20252025.11 | 42.44 | 25.95 | 39.54 | |
| ClearCLIPBackbone=ViT-L/14, Publication=ECCV20242025.11 | 39.96 | 25.46 | 36.52 | |
| SCLIPBackbone=ViT-H/14, Publication=ECCV20242025.11 | 38.9 | 25.32 | 39.92 | |
| TridentBackbone=ViT-B/16, Publication=ICCV20252025.11 | 37.88 | 22.86 | 35.06 | |
| ProxyCLIPBackbone=ViT-B/16, Publication=ECCV20242025.11 | 35.34 | 20.45 | 32.66 | |
| SCLIPBackbone=ViT-L/14, Publication=ECCV20242025.11 | 33.84 | 20.46 | 32.71 | |
| ClearCLIPBackbone=ViT-B/16, Publication=ECCV20242025.11 | 29.31 | 16.78 | 26.29 | |
| SCLIPBackbone=ViT-B/16, Publication=ECCV20242025.11 | 27.26 | 14.26 | 25.69 |