List your company
ResearchFormerly WizWand

Open-Vocabulary Segmentation on Potsdam (test)

60.5mIoUCONCEPTBANK
13.80425.92738.0550.173Feb 6, 2026Feb 20, 2026Mar 7, 2026Mar 21, 2026Apr 5, 2026Apr 19, 2026May 4, 2026
Updated 6mo ago

Evaluation results

MethodLinks
CONCEPTBANKBackbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background2026.0260.5
SegEarth-OV3Backbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=arXiv’252026.0257.8
CAFe-DINORS Training=No, Background Inclusion=Yes, Trained on target dataset=false2026.0556.4
SkySense-OBackbone (Size)=CLIP (ViT-L/14@512px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’252026.0254.1
SAM3Backbone (Size)=SAM3 (PE-L+/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICLR’262026.0254.1
CorrCLIPBackbone (Size)=CLIP + DINO (ViT-B/8) + SAM2 (Hiera-L), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICCV’252026.0251.9
ProxyCLIPBackbone (Size)=CLIP + DINOv2 (ViT-B/14), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’242026.0249
SegEarth-OVBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=CVPR’252026.0248.5
SegEarth-OVRS Training=Self-Supervised, Background Inclusion=Yes, Trained on target dataset=false2026.0546.2
SCLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’242026.0239.6
GEMBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=CVPR’242026.0239.1
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’262026.0238.4
SANBackbone (Size)=CLIP (ViT-L/14@336px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’232026.0237.3
Cat-SegBackbone (Size)=CLIP (ViT-L/14@336px), Training Paradigm=Training-based, Evaluation Protocol=with background, Pub. & Year=CVPR’242026.0235.8
GSNetRS Training=Yes, Background Inclusion=Yes, Trained on target dataset=false2026.0534.9
MaskCLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ECCV’222026.0233.9
DINOv3.txtRS Training=No, Background Inclusion=Yes, Trained on target dataset=false2026.0523.8
OVRSRS Training=Yes, Background Inclusion=Yes, Trained on target dataset=false2026.0517.9
CLIPBackbone (Size)=CLIP (ViT-B/16), Training Paradigm=Parameter-free, Evaluation Protocol=with background, Pub. & Year=ICML’212026.0215.6