Semantic Segmentation on Coco-Stuff (test)
52mIoUGMMSeg
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| GMMSegBackbone=SwinLarge, Base Architecture=Mask2Former2022.10 | 52 | — | — | — | — | — | — | — | — | — | |
| Mask2FormerBackbone=SwinLarge2022.10 | 51 | — | — | — | — | — | — | — | — | — | |
| UPerNet+IDRNetBackbone=Swin-Large, Stride=32x2023.10 | 50.5 | — | — | — | — | — | — | — | — | — | |
| UPerNet+MCIBI++Backbone=Swin-Large, Stride=32x2023.10 | 50.27 | — | — | — | — | — | — | — | — | — | |
| DINOv3-LResolution=7682026.05 | 50.12 | — | — | — | — | — | — | — | — | — | |
| VECA-LResolution=7682026.05 | 50.06 | — | — | — | — | — | — | — | — | — | |
| Mask2Former+IDRNetBackbone=Swin-Large, Stride=32x2023.10 | 49.96 | — | — | — | — | — | — | — | — | — | |
| DINOv3-LResolution=5122026.05 | 49.71 | — | — | — | — | — | — | — | — | — | |
| VECA-LResolution=5122026.05 | 49.53 | — | — | — | — | — | — | — | — | — | |
| DINOv3-BResolution=7682026.05 | 49.04 | — | — | — | — | — | — | — | — | — | |
| DINOv3-LResolution=3842026.05 | 48.93 | — | — | — | — | — | — | — | — | — | |
| VECA-LResolution=3842026.05 | 48.92 | — | — | — | — | — | — | — | — | — | |
| DINOv3-BResolution=5122026.05 | 48.41 | — | — | — | — | — | — | — | — | — | |
| VECA-BResolution=7682026.05 | 48.37 | — | — | — | — | — | — | — | — | — | |
| DINOv3-BResolution=3842026.05 | 47.97 | — | — | — | — | — | — | — | — | — | |
| VECA-BResolution=47.922026.05 | 47.92 | — | — | — | — | — | — | — | — | — | |
| DINOv3-SResolution=2562026.05 | 47.74 | — | — | — | — | — | — | — | — | — | |
| DINOv3-LResolution=2562026.05 | 47.36 | — | — | — | — | — | — | — | — | — | |
| VECA-LResolution=2562026.05 | 47.36 | — | — | — | — | — | — | — | — | — | |
| VECA-BResolution=3842026.05 | 47.27 | — | — | — | — | — | — | — | — | — | |
| SegNeXt-LBackbone=MSCAN-Large, Stride=32x2023.10 | 47.2 | — | — | — | — | — | — | — | — | — | |
| SegformerBackbone=MiT-B5, Stride=32x2023.10 | 46.7 | — | — | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=7682026.05 | 46.27 | — | — | — | — | — | — | — | — | — | |
| DINOv3-BResolution=2562026.05 | 45.87 | — | — | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=5122026.05 | 45.86 | — | — | — | — | — | — | — | — | — | |
| VECA-BResolution=2562026.05 | 45.43 | — | — | — | — | — | — | — | — | — | |
| DINOv3-SResolution=7682026.05 | 45.04 | — | — | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=3842026.05 | 44.95 | — | — | — | — | — | — | — | — | — | |
| GMMSegBackbone=MiT-B5, Base Model=SegFormer2022.10 | 44.7 | — | — | — | — | — | — | — | — | — | |
| DINOv3-SResolution=5122026.05 | 44.67 | — | — | — | — | — | — | — | — | — | |
| VECA-S+Resolution=5122026.05 | 44.33 | — | — | — | — | — | — | — | — | — | |
| VECA-S+Resolution=7682026.05 | 44.33 | — | — | — | — | — | — | — | — | — | |
| GMMSegBackbone=Swin-Base, Base Model=UPerNet2022.10 | 44.3 | — | — | — | — | — | — | — | — | — | |
| SegFormerBackbone=MiT-B52022.10 | 44 | — | — | — | — | — | — | — | — | — | |
| DINO + EAGLEBackbone=ViT-S/8, Evaluation Protocol=Linear2024.03 | 43.9 | — | 76.8 | — | — | — | — | — | — | — | |
| DINOv3-SResolution=3842026.05 | 43.82 | — | — | — | — | — | — | — | — | — | |
| VECA-SResolution=7682026.05 | 43.77 | — | — | — | — | — | — | — | — | — | |
| VECA-S+Resolution=3842026.05 | 43.53 | — | — | — | — | — | — | — | — | — | |
| VECA-SResolution=5122026.05 | 43.45 | — | — | — | — | — | — | — | — | — | |
| UPerNetBackbone=Swin-Base2022.10 | 43.4 | — | — | — | — | — | — | — | — | — | |
| VECA-SResolution=3842026.05 | 42.86 | — | — | — | — | — | — | — | — | — | |
| DINO + HPBackbone=ViT-S/8, Evaluation Protocol=Linear2023.03 | 42.7 | — | 75.6 | — | — | — | — | — | — | — | |
| DINO + HPBackbone=ViT-S/8, Evaluation Protocol=Linear2024.03 | 42.7 | — | 75.6 | — | — | — | — | — | — | — | |
| DINO + EAGLEBackbone=ViT-S/16, Evaluation Protocol=Linear2024.03 | 42.5 | — | 75.2 | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=2562026.05 | 42.13 | — | — | — | — | — | — | — | — | — | |
| UPerNet+MCIBI++Backbone=ResNet-101, Stride=8x2023.10 | 41.84 | — | — | — | — | — | — | — | — | — | |
| ISNetBackbone=ResNet-101, Stride=8x2023.10 | 41.6 | — | — | — | — | — | — | — | — | — | |
| VECA-S+Resolution=2562026.05 | 41.32 | — | — | — | — | — | — | — | — | — | |
| SelfPatch + HPBackbone=ViT-S/16, Evaluation Protocol=Linear2023.03 | 41.3 | — | 74.9 | — | — | — | — | — | — | — | |
| OCRNetBackbone=HRNetV2-W48, Stride=4x2023.10 | 40.5 | — | — | — | — | — | — | — | — | — | |
| VECA-SResolution=2562026.05 | 40.49 | — | — | — | — | — | — | — | — | — | |
| ACNetBackbone=Res-1012019.11 | 40.1 | — | — | — | — | — | — | — | — | — | |
| ACNetBackbone=ResNet-1012018.09 | 40.1 | — | — | — | — | — | — | — | — | — | |
| ACNetBackbone=ResNet-101, Stride=8x2023.10 | 40.1 | — | — | — | — | — | — | — | — | — | |
| OCNet (w/ Base-OC)Backbone=HRNetV2-482018.09 | 40 | — | — | — | — | — | — | — | — | — | |
| EMANetBackbone=ResNet-1012018.09 | 39.9 | — | — | — | — | — | — | — | — | — | |
| MVP-SEG+Learning Setting=Transductive, ST (Self-Training)=true2023.04 | 39.9 | — | — | — | 55.8 | 38.3 | 45.5 | — | — | — | |
| Fully SupST (Self-Training)=false2023.04 | 39.9 | — | — | — | — | — | — | — | — | — | |
| EMANetBackbone=ResNet-101, Stride=8x2023.10 | 39.9 | — | — | — | — | — | — | — | — | — | |
| EMANetBackbone=ResNet-1012019.07 | 39.9 | — | — | — | — | — | — | — | — | — | |
| OCNet (w/ ASP-OC)Backbone=HRNetV2-482018.09 | 39.8 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-1012018.09 | 39.7 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=Res1012018.09 | 39.7 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=Res-1012019.11 | 39.7 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-1012018.09 | 39.7 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-1012019.07 | 39.7 | — | — | — | — | — | — | — | — | — | |
| SVCNet2019.09 | 39.6 | 51.5 | — | 69.2 | — | — | — | — | — | — | |
| SVCNetBackbone=ResNet-1012018.09 | 39.6 | — | — | — | — | — | — | — | — | — | |
| MaskCLIP+Learning Setting=Transductive, ST (Self-Training)=false2023.04 | 39.6 | — | — | — | 54.7 | — | 45 | — | — | — | |
| OCRNetBackbone=ResNet-101, Stride=8x2023.10 | 39.5 | — | — | — | — | — | — | — | — | — | |
| GFFNetBackbone=ResNet1012019.04 | 39.2 | — | — | — | — | — | — | — | — | — | |
| OCNet (w/ Base-OC)Backbone=ResNet-1012018.09 | 39.2 | — | — | — | — | — | — | — | — | — | |
| GMMSegBackbone=HRNetv2-W48, Base Model=OCRNet2022.10 | 39.2 | — | — | — | — | — | — | — | — | — | |
| SGRBackbone=Res-1012019.11 | 39.1 | — | — | — | — | — | — | — | — | — | |
| OCNet (w/ ASP-OC)Backbone=ResNet-1012018.09 | 39.1 | — | — | — | — | — | — | — | — | — | |
| DINO + HPBackbone=ViT-S/16, Evaluation Protocol=Linear2023.03 | 39.1 | — | 74.1 | — | — | — | — | — | — | — | |
| DINO + HPBackbone=ViT-S/16, Evaluation Protocol=Linear2024.03 | 39.1 | — | 74.1 | — | — | — | — | — | — | — | |
| SGRBackbone=ResNet-1012019.07 | 39.1 | — | — | — | — | — | — | — | — | — | |
| OCRNet + BANELoss=L_CE + L_PA (Ours), Sampling=Boundary-aware (Ours)2024.04 | 39.08 | — | — | — | — | — | — | — | — | — | |
| OCRNetLoss=L_CE, Sampling=None2024.04 | 39 | — | — | — | — | — | — | — | — | — | |
| DSSPNBackbone=ResNet-101, Evaluation Protocol=Universal2018.03 | 38.9 | 50.3 | 70.7 | — | — | — | — | — | — | — | |
| DSSPNBackbone=Res-1012019.11 | 38.9 | — | — | — | — | — | — | — | — | — | |
| OCRNet + [31]Loss=L_CE + L_ems + L_ccs, Sampling=Random2024.04 | 38.55 | — | — | — | — | — | — | — | — | — | |
| DINO + STEGOBackbone=ViT-S/8, Evaluation Protocol=Linear2023.03 | 38.3 | — | 74.4 | — | — | — | — | — | — | — | |
| DINO + STEGOBackbone=ViT-S/8, Evaluation Protocol=Linear2024.03 | 38.3 | — | 74.4 | — | — | — | — | — | — | — | |
| DeepLabV3 + BANELoss=L_CE + L_PA (Ours), Sampling=Boundary-aware (Ours)2024.04 | 38.12 | — | — | — | — | — | — | — | — | — | |
| DeepLabV3Loss=L_CE, Sampling=None2024.04 | 37.92 | — | — | — | — | — | — | — | — | — | |
| PSPNetBackbone=ResNet1012022.10 | 37.8 | — | — | — | — | — | — | — | — | — | |
| OCRNetBackbone=HRNetv2-W482022.10 | 37.6 | — | — | — | — | — | — | — | — | — | |
| EMANetBackbone=ResNet-502019.07 | 37.6 | — | — | — | — | — | — | — | — | — | |
| DeepLabV3 + [31]Loss=L_CE + L_ems + L_ccs, Sampling=Random2024.04 | 37.39 | — | — | — | — | — | — | — | — | — | |
| DSSPNBackbone=ResNet-101, Evaluation Protocol=finetune2018.03 | 37.3 | 48.1 | 69.4 | — | — | — | — | — | — | — | |
| DSSPNBackbone=ResNet-1012019.07 | 37.3 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-502018.09 | 37.2 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=Res502018.09 | 37.2 | — | — | — | — | — | — | — | — | — | |
| DANetBackbone=ResNet-502019.07 | 37.2 | — | — | — | — | — | — | — | — | — | |
| HRNet + [31]Loss=L_CE + L_ems + L_ccs, Sampling=Random2024.04 | 36.35 | — | — | — | — | — | — | — | — | — | |
| HRNet + BANELoss=L_CE + L_PA (Ours), Sampling=Boundary-aware (Ours)2024.04 | 36.34 | — | — | — | — | — | — | — | — | — | |
| DSSPNBackbone=ResNet-101, Evaluation Protocol=standard2018.03 | 36.2 | 47 | 68.5 | — | — | — | — | — | — | — | |
| DSSPN2019.09 | 36.2 | 48.1 | — | 68.5 | — | — | — | — | — | — |