List your company
ResearchFormerly WizWand

Open Vocabulary Semantic Segmentation on COCO-Stuff

44.4mIoUActiveSAM
2.813.624.435.2Feb 5, 2025Apr 29, 2025Jul 22, 2025Oct 14, 2025Jan 5, 2026Mar 30, 2026Jun 22, 2026
Updated 4mo ago

Evaluation results

MethodLinks
ActiveSAMVenue=Ours, Backbone (Size)=SAM 3 (PE-L+/14), Training Strategy=Training-free2026.0644.4——
ActiveSAM2026.0644.41.66—
SegEarth-OV3Venue=arXiv’25, Backbone (Size)=SAM 3 (PE-L+/14), Training Strategy=Training-free2026.0642.8——
SegEarth-OV32026.0642.80.31—
OVS-DINOBackbone=DINOv2 B/142026.0434.2——
CorrCLIPVenue=ICCV’25, Backbone (Size)=CLIP + †, Training Strategy=Training-free2026.0634——
ReMEVenue=ICCV’25, Backbone (Size)=CLIP + §, Training Strategy=Training-free2026.0633.3——
SAM 3Venue=ICLR’26, Backbone (Size)=SAM 3 (PE-L+/14), Training Strategy=Training-free2026.0633.3——
SAM^32026.0633.30.11—
SAM-CLIPVenue=CVPR’24, Backbone (Size)=CLIP + SAM (ViT-B/16), Training Strategy=Training-based2026.0631.5——
Talk2DINOVenue=ICCV’25, Backbone (Size)=CLIP + DINOv2†(ViT-B/14), Training Strategy=Training-based2026.0630.2——
TridentVenue=ICCV’25, Backbone (Size)=CLIP + ‡, Training Strategy=Training-free2026.0628.3——
Talk2DINOBackbone=DINOv2 B/142026.0428.1——
LoftUp2026.0627.54—48.38
RF-CLIPVenue=AAAI’26, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0627.5——
AnyUp2026.0627.3—48.62
JAFAR2026.0627.13—48.47
RaysUp2026.0627.11—48.57
FreeDAPub. & Year=CVPR’24, Extra data=COCO Captions, Extra backbone=DINOv2 (ViT-B/14), Training free=✓, Mask Refinement=true2026.0327.1——
CASSVenue=CVPR’25, Backbone (Size)=CLIP + DINO (ViT-B/8), Training Strategy=Training-free2026.0626.7——
DIH-CLIPVenue=ICCV’25, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0626.7——
CASS2026.0626.70.63—
FeatUp2026.0626.67—48.13
ProxyCLIP*Pub. & Year=ECCV’24, Extra data=✗, Extra backbone=DINOv2† (ViT-B/14), Training free=✓, Mask Refinement=true2026.0326.6——
SFPVenue=ICCV’25, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0626.4——
PEARLExtra data=✗, Extra backbone=✗, Training free=✓, Mask Refinement=false2026.0326.3——
SFP*Pub. & Year=ICCV’25, Extra data=✗, Extra backbone=✗, Training free=✓, Mask Refinement=true2026.0325.9——
LPOSSVenue=CVPR’25, Backbone (Size)=CLIP + DINO (ViT-B/16), Training Strategy=Training-free2026.0625.9——
Bilinear2026.0625.73—47.91
NACLIP*Pub. & Year=WACV’25, Extra data=✗, Extra backbone=✗, Training free=✓, Mask Refinement=true2026.0325.7——
DSGSBackbone=CLIP B/162026.0425.7——
FreeDAVenue=CVPR’24, Backbone (Size)=CLIP + DINOv2 (ViT-B/14), Training Strategy=Training-free2026.0625.7——
CASS*Pub. & Year=CVPR’25, Extra data=✗, Extra backbone=DINOv3 (ViT-B/16), Training free=✓, Mask Refinement=true2026.0325.6——
ProxyCLIPVenue=ECCV’24, Backbone (Size)=CLIP + DINOv2†(ViT-B/14), Training Strategy=Training-free2026.0625.4——
CLIP-DINOiserPub. & Year=ECCV’24, Extra data=ImageNet-1K, Extra backbone=DINOv1 (ViT-B/16), Training free=✗, Mask Refinement=true2026.0325.3——
+CS & AMF (GCLIP)Pub. & Year=Ours, Setting=TF-OVSS2025.0224.8——
CLIP-DINOiserParadigm=Use Frozen CLIP, Background Prompt=No, Protocol=Rerun (*)2026.0124.6——
CLIP-DINOiserBackbone=CLIP B/162026.0424.6——
CLIP-DINOiserVenue=ECCV’24, Backbone (Size)=CLIP + DINO (ViT-B/16), Training Strategy=Training-based2026.0624.6——
CLIPtrasePub. & Year=ECCV’24, Setting=TF-OVSS2025.0224.1——
INSIGHT_thresParadigm=Use Frozen CLIP with Interpretability, Background Prompt=No2026.0124——
CASS*Pub. & Year=CVPR’25, Extra data=✗, Extra backbone=DINOv2† (ViT-B/14), Training free=✓, Mask Refinement=true2026.0324——
+CSPub. & Year=Ours, Setting=TF-OVSS2025.0224——
SCLIP*Pub. & Year=ECCV’24, Extra data=✗, Extra backbone=✗, Training free=✓, Mask Refinement=true2026.0323.9——
ClearCLIPPub. & Year=ECCV’24, Setting=TF-OVSS2025.0223.9——
CoDeVenue=CVPR’24, Backbone (Size)=CLIP, Training Strategy=Training-based2026.0623.9——
ClearCLIPVenue=ECCV’24, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0623.9——
GEM‡Pub. & Year=CVPR’24, Setting=TF-OVSS2025.0223.7——
PixelCLIP-ViTBackbone=CLIP B/162026.0423.6——
SynSegBackbone=CLIP B/162026.0423.6——
NACLIPVenue=WACV’25, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0623.3——
LaVGPub. & Year=ECCV’24, Extra data=✗, Extra backbone=DINOv1 (ViT-B/8), Training free=✓, Mask Refinement=true2026.0323.2——
LaVGVenue=ECCV’24, Backbone (Size)=CLIP + DINO (ViT-B/8), Training Strategy=Training-free2026.0623.2——
CLIPtraseVenue=ECCV’24, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0622.8——
INSIGHTParadigm=Use Frozen CLIP with Interpretability, Background Prompt=No2026.0122.6——
TCLPub. & Year=CVPR’23, Extra data=CC3M+CC12M, Extra backbone=✗, Training free=✗, Mask Refinement=true2026.0322.4——
SCLIPPub. & Year=ECCV’24, Setting=TF-OVSS2025.0222.4——
SCLIPVenue=ECCV’24, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0622.4——
CLIPSurgeryPub. & Year=Arxiv’23, Setting=TF-OVSS2025.0221.9——
dino.txtBackbone=DINOv2 L/142026.0420.9——
ReCLIPPub. & Year=CVPR’24, Setting=USS2025.0220.3——
TCLParadigm=Text/image alignment training with captions, Background Prompt=No2026.0119.6——
TCLBackbone=CLIP B/162026.0419.6——
TCLPub. & Year=CVPR’23, Setting=T-OVSS2025.0219.6——
TCLVenue=CVPR’23, Backbone (Size)=CLIP, Training Strategy=Training-based2026.0619.6——
MaskCLIP+†Pub. & Year=ECCV’22, Setting=USS2025.0219.5——
MaskCLIPParadigm=Use Frozen CLIP, Extra Backbones=ref[9], Background Prompt=No2026.0118.6——
MaskCLIPParadigm=Use Frozen CLIP, Background Prompt=No2026.0117.6——
CoDeBackbone=CLIP B/162026.0417.6——
ReCoPub. & Year=NeurIPS’22, Extra data=ImageNet-1K, Extra backbone=✗, Training free=✓, Mask Refinement=true2026.0316.3——
GEMVenue=CVPR’24, Backbone (Size)=CLIP, Training Strategy=Training-free2026.0615.7——
GroupViTPub. & Year=CVPR’22, Extra data=CC12M+RedCaps, Extra backbone=✗, Training free=✗, Mask Refinement=true2026.0315.4——
GroupViTParadigm=Text/image alignment training with captions, Background Prompt=No2026.0115.3——
GroupViTBackbone=CLIP S/162026.0415.3——
GroupViT‡Pub. & Year=CVPR’22, Setting=T-OVSS2025.0215.3——
GroupViTVenue=CVPR’22, Backbone (Size)=GroupViT (ViT-S/16), Training Strategy=Training-based2026.0615.3——
MaskCLIPPub. & Year=ECCV’22, Extra data=✗, Extra backbone=✗, Training free=✓, Mask Refinement=true2026.0315.1——
ReCoParadigm=Build prototypes per class, Background Prompt=No2026.0114.8——
ReCoBackbone=CLIP B/162026.0414.8——
CoCuPub. & Year=NeurIPS’24, Setting=T-OVSS2025.0213.6——
MaskCLIP†Pub. & Year=ECCV’22, Setting=TF-OVSS2025.0213.6——
CLIP-DIYParadigm=Use Frozen CLIP, Extra Backbones=DINO, Background Prompt=No2026.0113.3——
CLIPVenue=ICML’21, Backbone (Size)=CLIP, Training Strategy=Training-free2026.067.2——
CLIP‡Pub. & Year=ICML’21, Setting=TF-OVSS2025.024.4——