Semantic Segmentation on VOC21
84.6mIoUODISE
Evaluation Results
| Method | Links | |
|---|---|---|
| ODISEBackbone=ViT-L/14, Training Set=COCO-Panoptic2025.05 | 84.6 | |
| CAT-Seg+DeCLIPBackbone=ViT-L/14, Training Set=COCO-Stuff2025.05 | 83.9 | |
| FrozenSegBackbone=ConvNeXt-L, Training Set=COCO-Panoptic2025.05 | 82.5 | |
| CAT-SegBackbone=ViT-L/14, Training Set=COCO-Stuff2025.05 | 82.5 | |
| FC-CLIPBackbone=ConvNeXt-L, Training Set=COCO-Panoptic2025.05 | 81.8 | |
| CAT-Seg+DeCLIPBackbone=ViT-B/16, Training Set=COCO-Stuff2025.05 | 81.3 | |
| CAT-SegBackbone=ViT-B/16, Training Set=COCO-Stuff2025.05 | 77.3 | |
| CorrCLIPBackbone=ViT-L/14, Zero-shot=true2025.12 | 76.7 | |
| CorrCLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 74.8 | |
| CLIPer + ARMBackbone=ViT-B/16, Zero-shot=true2025.12 | 73.2 | |
| CLIPer + ARMBackbone=ViT-L/14, Zero-shot=true2025.12 | 71.6 | |
| CLIPerBackbone=ViT-L/14, Zero-shot=true2025.12 | 69.8 | |
| SCLIP + ARMBackbone=ViT-B/16, Zero-shot=true2025.12 | 69.5 | |
| DSLO (solve maximum velocity)Logits Model=CLIP ViT-L/142026.04 | 68.9 | |
| DSLO (solve optimal path)Logits Model=CLIP ViT-L/142026.04 | 68.2 | |
| DSLO (solve maximum velocity)Logits Model=CLIP ViT-B/162026.04 | 67.8 | |
| DSLO (solve optimal path)Logits Model=CLIP ViT-B/162026.04 | 66.9 | |
| OVDiffVenue=ECCV'242025.12 | 66.3 | |
| CLIPerBackbone=ViT-B/16, Zero-shot=true2025.12 | 65.9 | |
| SCLIP + ARMBackbone=ViT-L/14, Zero-shot=true2025.12 | 65.8 | |
| CASSLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 65.8 | |
| ZegFormer+Backbone=ViT-B/16, Training Set=COCO-Stuff2025.05 | 65.5 | |
| SC-CLIPLogits Model=CLIP ViT-L/14, Paradigm=M.M.2026.04 | 65 | |
| SC-CLIPLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 64.6 | |
| NACLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 64.1 | |
| ProxyCLIP + RAG-OVSVenue=This Paper2025.12 | 62.9 | |
| LaVGLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 62.1 | |
| CLIP-DINOiser†Logits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 62.1 | |
| SCLIP + RAG-OVSVenue=This Paper2025.12 | 61.9 | |
| ProxyCLIPVenue=ECCV'242025.12 | 61.3 | |
| NACLIP + ResCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 61.1 | |
| ResCLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 61.1 | |
| CLIP-Surgery + RAG-OVSVenue=This Paper2025.12 | 61.1 | |
| SCLIP + ResCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 60.7 | |
| NACLIP + RAG-OVSVenue=This Paper2025.12 | 60.3 | |
| SCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 59.1 | |
| SCLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 59.1 | |
| SCLIPVenue=ECCV'242025.12 | 59.1 | |
| SCLIPLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 59.1 | |
| ProxyCLIP‡Logits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 59.1 | |
| ClearCLIP + ResCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 59 | |
| CLIP-SurgeryVenue=Pattern Recognition2025.12 | 59 | |
| NACLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 58.9 | |
| NACLIPVenue=WACV'252025.12 | 58.9 | |
| NACLIPLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 58.9 | |
| GEMLogits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 58.7 | |
| ProxyCLIP‡Logits Model=CLIP ViT-L/14, Paradigm=M.M.2026.04 | 58.1 | |
| CLIPtrace + RAG-OVSVenue=This Paper2025.12 | 56.2 | |
| CLIPSurgeryLogits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 55.2 | |
| TCLLogits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 55 | |
| NACLIP + ResCLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 54.1 | |
| ResCLIPBackbone=ViT-L/14, Zero-shot=true2025.12 | 54.1 | |
| OVSegmentorTraining-free?=false, Encoder=ViT-B/162024.11 | 53.8 | |
| PGSegTraining-free?=false, Encoder=ViT-S/162024.11 | 53.2 | |
| CLIPtraceVenue=ECCV'242025.12 | 53 | |
| SCLIP + ResCLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 52.8 | |
| SegCLIPTraining-free?=false, Encoder=ViT-S/162024.11 | 52.6 | |
| ViewCoTraining-free?=false, Encoder=ViT-S/162024.11 | 52.4 | |
| GroupViTVenue=CVPR'222025.12 | 52.3 | |
| NACLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 52.2 | |
| NACLIPLogits Model=CLIP ViT-L/14, Paradigm=M.M.2026.04 | 52.1 | |
| ClearCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 51.8 | |
| ClearCLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 51.8 | |
| ClearCLIPLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 51.8 | |
| TCLTraining-free?=false, Encoder=ViT-B/162024.11 | 51.2 | |
| CLIPtraseLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 50.9 | |
| ClearCLIP + ResCLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 50.7 | |
| GroupViTTraining-free?=false, Encoder=ViT-S/162024.11 | 50.4 | |
| GroupViTLogits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 50.4 | |
| ClearCLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 48.7 | |
| CaRLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 48.6 | |
| ClearCLIPLogits Model=CLIP ViT-L/14, Paradigm=M.M.2026.04 | 48.6 | |
| CLIPSurgeryLogits Model=CLIP ViT-L/14, Paradigm=I.T.2026.04 | 47.9 | |
| GEMTraining-free?=true, Encoder=ViT-B/162024.11 | 46.2 | |
| ClearCLIPBackbone=ViT-L/14, Zero-shot=true2025.12 | 46.1 | |
| GEMLogits Model=CLIP ViT-L/14, Paradigm=I.T.2026.04 | 45.2 | |
| SCLIPTraining-free?=true, Encoder=ViT-L/142024.11 | 44.4 | |
| SCLIPLogits Model=CLIP ViT-L/14, Paradigm=M.M.2026.04 | 44.4 | |
| TIPSv2 L/14Model Scale=L, Patch Size=14, Zero-shot=true2026.04 | 44.4 | |
| SCLIPBackbone=ViT-L/14, Zero-shot=true2025.12 | 43.5 | |
| MaskCLIPVenue=ECCV'222025.12 | 43.4 | |
| CoCuTraining-free?=false, Encoder=ViT-S/162024.11 | 40.9 | |
| UNCHABackbone=ViT-S/16, Zero-shot=true2026.03 | 39.03 | |
| MaskCLIPTraining-free?=true, Encoder=ViT-B/162024.11 | 38.8 | |
| MaskCLIPBackbone=ViT-B/16, Zero-shot=true2025.12 | 38.8 | |
| MaskCLIPLogits Model=CLIP ViT-B/16, Paradigm=I.T.2026.04 | 38.3 | |
| HyCoCLIPBackbone=ViT-S/16, Zero-shot=true2026.03 | 36.79 | |
| MERUBackbone=ViT-S/16, Zero-shot=true2026.03 | 36.18 | |
| CLIPBackbone=ViT-S/16, Zero-shot=true2026.03 | 36.02 | |
| UNCHABackbone=ViT-B/16, Zero-shot=true2026.03 | 32.28 | |
| TIPS L/14Model Scale=L, Patch Size=14, Zero-shot=true2026.04 | 30.5 | |
| CLIPBackbone=ViT-B/16, Zero-shot=true2026.03 | 28.47 | |
| SigLIP2 SO/14Model Scale=SO, Patch Size=14, Zero-shot=true2026.04 | 26.8 | |
| MERUBackbone=ViT-B/16, Zero-shot=true2026.03 | 26.05 | |
| HyCoCLIPBackbone=ViT-B/16, Zero-shot=true2026.03 | 26.03 | |
| ReCoTraining-free?=true, Encoder=ViT-B/162024.11 | 25.1 | |
| RecoVenue=NeurIPS'222025.12 | 25.1 | |
| ReCoLogits Model=CLIP ViT-B/16, Paradigm=M.M.2026.04 | 25.1 | |
| MaskCLIPLogits Model=CLIP ViT-L/14, Paradigm=I.T.2026.04 | 24.8 | |
| MaskCLIPBackbone=ViT-L/14, Zero-shot=true2025.12 | 23.3 |