Open-Vocabulary Segmentation on iSAID (test)
54.3mIoURSKT-Seg
Evaluation Results
| Method | Links | |
|---|---|---|
| RSKT-SegBackbone (Size)=CLIP (ViT-L/14@336px) + RemoteCLIP (ViT-B/16) + DINO (ViT-B/32), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=AAAI’262026.02 | 54.3 | |
| Cat-SegBackbone (Size)=CLIP (ViT-L/14@336px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’242026.02 | 53.3 | |
| SANBackbone (Size)=CLIP (ViT-L/14@336px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’232026.02 | 49.6 | |
| SkySense-OBackbone (Size)=CLIP (ViT-L/14@512px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’252026.02 | 43.9 | |
| CONCEPTBANKBackbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background2026.02 | 35.4 | |
| SegEarth-OV3Backbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=arXiv’252026.02 | 27.6 | |
| CorrCLIPBackbone (Size)=CLIP + DINO (ViT-B/8) + SAM2 (Hiera-L), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICCV’252026.02 | 25.5 | |
| ProxyCLIPBackbone (Size)=CLIP + DINOv2 (ViT-B/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’242026.02 | 21.8 | |
| SegEarth-OVBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=CVPR’252026.02 | 21.7 | |
| GEMBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=CVPR’242026.02 | 17.7 | |
| SAM3Backbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICLR’262026.02 | 17.7 | |
| SCLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’242026.02 | 16.1 | |
| MaskCLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’222026.02 | 14.5 | |
| CLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICML’212026.02 | 7.5 |