Semantic Segmentation on COCO 20i (4-Fold CV mIoU)
76.1mIoU (Mean)SegIC
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SegICEncoder=DINOv2, #Param=310 M, Training Protocol=Task-specific fine-tuning, Supervision Type=Semantic + mask supervision2026.03 | 76.1 | — | — | — | — | — | — | — | — | |
| SegIC (COCO)Encoder=DINOv2, #Param=310 M, Training Protocol=Task-specific fine-tuning, Supervision Type=Semantic + mask supervision, dataset split=COCO2026.03 | 75.6 | — | — | — | — | — | — | — | — | |
| DSV-LFSBackbone=SAM, Setting=5-shot, Multimodal=true2026.07 | 71.9 | — | — | — | — | 71 | 73.8 | 71.3 | 71.4 | |
| DiffewSEncoder=Stable Diffusion, #Param=890 M, Training Protocol=Task-specific fine-tuning, Supervision Type=Semantic + mask supervision2026.03 | 71.3 | — | — | — | — | — | — | — | — | |
| DSV-LFSBackbone=SAM, Setting=1-shot, Multimodal=true2026.07 | 71.3 | — | — | — | — | 69.9 | 73.3 | 70.6 | 71.3 | |
| SINEEncoder=DINOv2, #Param=373 M, Training Protocol=Task-specific fine-tuning, Supervision Type=Semantic + mask supervision2026.03 | 64.5 | — | — | — | — | — | — | — | — | |
| CLIPixPrompt=text2026.07 | 61.8 | 58.8 | 64.9 | 57.8 | 65.7 | — | — | — | — | |
| CLIPixBackbone=ResNet, Setting=0-shot, Multimodal=true2026.07 | 61.8 | — | — | — | — | 58.8 | 64.9 | 57.8 | 65.7 | |
| LLaFS++Backbone=ResNet, Setting=5-shot, Multimodal=true2026.07 | 60.9 | — | — | — | — | 53.9 | 64.9 | 63.8 | 61.1 | |
| DCAMAVersion=TaP2025.12 | 59.81 | 59.83 | 58.93 | 60.94 | 59.53 | — | — | — | — | |
| PI_CLIPBackbone=ResNet, Setting=5-shot, Multimodal=true2026.07 | 59.1 | — | — | — | — | 56.4 | 66.2 | 55.9 | 58 | |
| CLIPixBackbone=EfficientNet, Setting=0-shot, Multimodal=true2026.07 | 59.1 | — | — | — | — | 56.4 | 62.1 | 55.1 | 62.9 | |
| DCAMAVersion=AdaptiveFSS2025.12 | 59.04 | 58.17 | 58.25 | 60.4 | 59.36 | — | — | — | — | |
| HMNetBackbone=ResNet, Setting=5-shot2026.07 | 58.9 | — | — | — | — | 53.4 | 64.6 | 60.8 | 56.8 | |
| ABCBBackbone=ResNet, Setting=5-shot2026.07 | 58.8 | — | — | — | — | 51.6 | 63.5 | 62.8 | 57.2 | |
| GF-SAMEncoder=DINOv2 + SAM, #Param=945 M, Training Protocol=Training free, Supervision Type=Mask-supervised pre-training2026.03 | 58.7 | — | — | — | — | — | — | — | — | |
| DCAMAVersion=Vanilla2025.12 | 58.07 | 56.39 | 57.63 | 59.84 | 58.43 | — | — | — | — | |
| INSID3Encoder=DINOv3, #Param=304 M, Training Protocol=Training free, Supervision Type=Unsupervised pre-training2026.03 | 57.6 | — | — | — | — | — | — | — | — | |
| LLaFS++Backbone=ResNet, Setting=1-shot, Multimodal=true2026.07 | 57.5 | — | — | — | — | 50.8 | 62.7 | 60.2 | 56.4 | |
| CLIPixBackbone=MobileNet, Setting=0-shot, Multimodal=true2026.07 | 57.5 | — | — | — | — | 54.9 | 60.7 | 53.8 | 60.4 | |
| HDMNetVersion=TaP2025.12 | 57.19 | 54.74 | 61.4 | 56.72 | 55.92 | — | — | — | — | |
| HDMNetVersion=Decoder FT2025.12 | 57.16 | 53.83 | 63.62 | 55.38 | 55.8 | — | — | — | — | |
| AENetBackbone=ResNet, Setting=5-shot2026.07 | 57.1 | — | — | — | — | 52.7 | 62.6 | 56.8 | 56.1 | |
| PI_CLIPBackbone=ResNet, Setting=1-shot, Multimodal=true2026.07 | 56.8 | — | — | — | — | 49.3 | 65.7 | 55.8 | 56.3 | |
| DCAMAVersion=Decoder FT2025.12 | 56.36 | 55.5 | 55.76 | 57.84 | 56.35 | — | — | — | — | |
| DGPNetBackbone=ResNet-50, Shots=5-shot2023.03 | 56.2 | 54.7 | 59.1 | 56.8 | 54.4 | — | — | — | — | |
| SegGPTEncoder=ViT, #Param=354 M, Training Protocol=Task-specific fine-tuning, Supervision Type=Semantic + mask supervision2026.03 | 56.1 | — | — | — | — | — | — | — | — | |
| SegGPTPrompt=mask2026.07 | 56.1 | 56.3 | 57.4 | 58.9 | 51.7 | — | — | — | — | |
| HDMNetBackbone=ResNet-50, Shots=5-shot2023.03 | 56 | 50.6 | 61.6 | 55.7 | 56 | — | — | — | — | |
| HDMNetBackbone=ResNet, Setting=5-shot2026.07 | 56 | — | — | — | — | 50.6 | 61.6 | 55.7 | 56 | |
| GF-SAM† + debiasEncoder=DINOv3 + SAM, #Param=945 M, Training Protocol=Training free, Supervision Type=Mask-supervised pre-training, Debias=True2026.03 | 55.9 | — | — | — | — | — | — | — | — | |
| HDMNetVersion=Vanilla2025.12 | 55.53 | 51.81 | 62.24 | 53.01 | 55.07 | — | — | — | — | |
| HDMNetVersion=TaP2025.12 | 55.51 | 51.2 | 60.21 | 57.46 | 53.19 | — | — | — | — | |
| FPTransVersion=TaP2025.12 | 55.38 | 53.6 | 57.9 | 56 | 54 | — | — | — | — | |
| BAMVersion=TaP2025.12 | 55.05 | 50.89 | 58.9 | 56.77 | 53.65 | — | — | — | — | |
| GF-SAM†Encoder=DINOv3 + SAM, #Param=945 M, Training Protocol=Training free, Supervision Type=Mask-supervised pre-training2026.03 | 54.8 | — | — | — | — | — | — | — | — | |
| FPTransVersion=Vanilla2025.12 | 54.72 | 52.3 | 57.4 | 55.7 | 53.5 | — | — | — | — | |
| FPTransVersion=Decoder FT2025.12 | 54.72 | 52.38 | 57.46 | 55.63 | 53.41 | — | — | — | — | |
| VRP-SAMPrompt=mask2026.07 | 53.9 | 48.1 | 55.8 | 60 | 51.6 | — | — | — | — | |
| LLaFSPrompt=mask2026.07 | 53.9 | 47.5 | 58.8 | 56.2 | 53 | — | — | — | — | |
| HDMNetVersion=Decoder FT2025.12 | 53.17 | 49.93 | 58.54 | 50.54 | 53.66 | — | — | — | — | |
| MatcherEncoder=DINOv2 + SAM, #Param=945 M, Training Protocol=Training free, Supervision Type=Mask-supervised pre-training2026.03 | 52.7 | — | — | — | — | — | — | — | — | |
| MatcherPrompt=mask2026.07 | 52.7 | 52.7 | 53.5 | 52.6 | 52.1 | — | — | — | — | |
| FCPPrompt=mask2026.07 | 52.5 | 46.4 | 56.4 | 55.3 | 51.8 | — | — | — | — | |
| HDMNetBackbone=VGG-16, Shots=5-shot2023.03 | 52.4 | 47 | 56.5 | 54.1 | 51.9 | — | — | — | — | |
| FPTransVersion=AdaptiveFSS2025.12 | 52.3 | 50.35 | 54.87 | 52.64 | 51.35 | — | — | — | — | |
| HMNetBackbone=ResNet, Setting=1-shot2026.07 | 52.1 | — | — | — | — | 45.5 | 58.7 | 52.9 | 51.4 | |
| FPTransVersion=TaP2025.12 | 52 | 47.8 | 53.78 | 54.64 | 51.8 | — | — | — | — | |
| DCAMAVersion=TaP2025.12 | 51.91 | 51.21 | 48.36 | 56.82 | 51.25 | — | — | — | — | |
| QCLNetBackbone=ResNet-101, Shot=5-shot2023.05 | 51.9 | — | — | — | — | — | — | — | — | |
| MIANetBackbone=ResNet50, Shot=5-shot2023.05 | 51.65 | 45.84 | 58.18 | 51.29 | 51.9 | — | — | — | — | |
| MIANetBackbone=ResNet, Setting=5-shot2026.07 | 51.6 | — | — | — | — | 45.8 | 58.1 | 51.2 | 51.9 | |
| HDMNetVersion=Vanilla2025.12 | 51.54 | 47.99 | 57.46 | 49.23 | 51.49 | — | — | — | — | |
| ABCBBackbone=ResNet, Setting=1-shot2026.07 | 51.5 | — | — | — | — | 46 | 56.3 | 54.3 | 51.3 | |
| AENetBackbone=ResNet, Setting=1-shot2026.07 | 51.3 | — | — | — | — | 45.4 | 57.1 | 52.6 | 50 | |
| BAMBackbone=ResNet50, Shot=5-shot2023.05 | 51.16 | 49.26 | 54.2 | 51.63 | 49.55 | — | — | — | — | |
| BAMBackbone=ResNet, Setting=5-shot2026.07 | 51.1 | — | — | — | — | 49.2 | 54.2 | 51.6 | 49.5 | |
| MIANetBackbone=VGG16, Shot=5-shot2023.05 | 51.03 | 46.18 | 56.09 | 52.33 | 49.54 | — | — | — | — | |
| CAPLBackbone=ResNet-101, Shot=5-shot2023.05 | 50.4 | — | — | — | — | — | — | — | — | |
| SSPBackbone=ResNet, Setting=5-shot2026.07 | 50.2 | — | — | — | — | 47.4 | 54.5 | 50.4 | 49.6 | |
| BAMVersion=Decoder FT2025.12 | 50.19 | 46.41 | 52.72 | 51.77 | 49.86 | — | — | — | — | |
| DCAMAVersion=Decoder FT2025.12 | 50.08 | 46.27 | 48.47 | 53.84 | 51.74 | — | — | — | — | |
| HDMNetBackbone=ResNet-50, Shots=1-shot2023.03 | 50 | 43.8 | 55.3 | 51.6 | 49.4 | — | — | — | — | |
| QCLNetBackbone=ResNet-50, Shot=5-shot2023.05 | 50 | — | — | — | — | — | — | — | — | |
| HDMNetBackbone=ResNet, Setting=1-shot2026.07 | 50 | — | — | — | — | 43.8 | 55.3 | 51.6 | 49.4 | |
| DPCNBackbone=ResNet50, Shot=5-shot2023.05 | 49.8 | 46 | 54.9 | 50.8 | 47.4 | — | — | — | — | |
| HSNetBackbone=ResNet-101, Shot=5-shot2023.05 | 49.5 | — | — | — | — | — | — | — | — | |
| HSNetBackbone=ResNet, Setting=5-shot2026.07 | 49.5 | — | — | — | — | 45.9 | 53 | 51.8 | 47.1 | |
| BAMBackbone=VGG16, Shot=5-shot2023.05 | 49.34 | 47.02 | 52.62 | 48.59 | 49.11 | — | — | — | — | |
| BAMVersion=TaP2025.12 | 48.98 | 44.3 | 50.93 | 52.67 | 48.02 | — | — | — | — | |
| DCAMAVersion=AdaptiveFSS2025.12 | 48.72 | 47.37 | 46.19 | 50.53 | 50.77 | — | — | — | — | |
| FPTransVersion=Decoder FT2025.12 | 48.51 | 44.49 | 49.2 | 51.99 | 48.37 | — | — | — | — | |
| BAMBackbone=ResNet-50, Shots=5-shot2023.03 | 48.5 | 43.2 | 53.4 | 49.4 | 48.1 | — | — | — | — | |
| CAPLBackbone=ResNet-50, Shots=5-shot2023.03 | 48.3 | 43.1 | 53.8 | 48.7 | 47.4 | — | — | — | — | |
| CAPLBackbone=ResNet-50, Shot=5-shot2023.05 | 48.3 | — | — | — | — | — | — | — | — | |
| FPTransVersion=Vanilla2025.12 | 48.04 | 44.1 | 48.79 | 51.49 | 47.78 | — | — | — | — | |
| BAMVersion=Vanilla2025.12 | 47.91 | 44.11 | 52.28 | 48.77 | 46.47 | — | — | — | — | |
| MIANetBackbone=ResNet50, Shot=1-shot2023.05 | 47.66 | 42.49 | 52.95 | 47.77 | 47.42 | — | — | — | — | |
| MIANetBackbone=ResNet, Setting=1-shot2026.07 | 47.6 | — | — | — | — | 42.5 | 52.9 | 47.7 | 47.7 | |
| BAMBackbone=VGG-16, Shots=5-shot2023.03 | 47.3 | 42.9 | 51.4 | 48.3 | 46.6 | — | — | — | — | |
| HSNetBackbone=ResNet-50, Shots=5-shot2023.03 | 46.9 | 43.3 | 51.3 | 48.2 | 45 | — | — | — | — | |
| HSNetBackbone=ResNet50, Shot=5-shot2023.05 | 46.9 | 43.3 | 51.3 | 48.2 | 45 | — | — | — | — | |
| HSNetBackbone=ResNet-50, Shot=5-shot2023.05 | 46.9 | — | — | — | — | — | — | — | — | |
| DCPBackbone=ResNet-50, Shots=5-shot2023.03 | 46.5 | 45.8 | 49.7 | 43.7 | 46.6 | — | — | — | — | |
| DCAMAVersion=Vanilla2025.12 | 46.47 | 44.39 | 45.39 | 48.87 | 47.24 | — | — | — | — | |
| BaselineBackbone=ResNet50, Shot=5-shot2023.05 | 46.43 | 42.79 | 49.42 | 47.41 | 46.08 | — | — | — | — | |
| BAMBackbone=ResNet50, Shot=1-shot2023.05 | 46.23 | 43.41 | 50.59 | 47.49 | 43.42 | — | — | — | — | |
| DPCNBackbone=VGG16, Shot=5-shot2023.05 | 46.2 | 42.7 | 51.6 | 45.7 | 44.6 | — | — | — | — | |
| BAMBackbone=ResNet, Setting=1-shot2026.07 | 46.2 | — | — | — | — | 43.4 | 50.6 | 47.5 | 43.4 | |
| HDMNetBackbone=VGG-16, Shots=1-shot2023.03 | 45.9 | 40.7 | 50.6 | 48.2 | 44 | — | — | — | — | |
| CWTn-shot=5-shot, Backbone=ResNet-50, Cross-domain setting=PASCAL → COCO2021.08 | 45.8 | 40.6 | 48.6 | 51.9 | 41.9 | — | — | — | — | |
| MIANetBackbone=VGG16, Shot=1-shot2023.05 | 45.69 | 40.56 | 50.53 | 46.5 | 45.18 | — | — | — | — | |
| CyCTRBackbone=ResNet-50, Shots=5-shot2023.03 | 45.6 | 41.1 | 48.9 | 45.2 | 47 | — | — | — | — | |
| BAMBackbone=ResNet-50, Shots=1-shot2023.03 | 45.2 | 39.4 | 49.9 | 46.2 | 45.2 | — | — | — | — | |
| DGPNetBackbone=ResNet-50, Shots=1-shot2023.03 | 45 | 43.6 | 47.8 | 44.5 | 44.2 | — | — | — | — | |
| SSPBackbone=ResNet50, Shot=5-shot2023.05 | 44.1 | 40.6 | 47 | 45.1 | 43.9 | — | — | — | — | |
| Label AnythingVersion=TaP2025.12 | 43.98 | 41.63 | 44.16 | 45.65 | 44.46 | — | — | — | — | |
| QCLNetBackbone=ResNet-101, Shot=1-shot2023.05 | 43.6 | — | — | — | — | — | — | — | — | |
| BAMBackbone=VGG16, Shot=1-shot2023.05 | 43.5 | 38.96 | 47.04 | 46.41 | 41.57 | — | — | — | — | |
| FPTransVersion=AdaptiveFSS2025.12 | 43.12 | 37.44 | 44.63 | 47.13 | 43.26 | — | — | — | — |