Semantic Segmentation on Pascal VOC 2012 (mIoU)
85.6mIoU*RADIOv2.5-L
Evaluation Results
| Method | Links | |
|---|---|---|
| *RADIOv2.5-LParams (M)=320, Evaluation=Linear probe2025.08 | 85.6 | |
| RADIOv2.5-LParams (M)=320, Evaluation Protocol=Linear probe2025.08 | 85.6 | |
| UniUGG Enc.Params (M)=320, Evaluation=Linear probe2025.08 | 85.43 | |
| UniUGG Enc.Params (M)=320, Evaluation Protocol=Linear probe2025.08 | 85.43 | |
| DINOv2-g/14-regParams (M)=1,137, Evaluation=Linear probe2025.08 | 82.78 | |
| DINOv2-g/14-regParams (M)=1,137, Evaluation Protocol=Linear probe2025.08 | 82.78 | |
| EVACLIP (+LazyStrike)Backbone=ViT-B/16, Zero-shot=true2026.02 | 79.6 | |
| EVACLIP (+LazyStrike)Backbone=ViT-L/14, Zero-shot=true2026.02 | 76.4 | |
| MetaCLIP (+LazyStrike)Backbone=ViT-L/14, Zero-shot=true2026.02 | 75.6 | |
| CLIP (+LazyStrike)Backbone=ViT-B/16, Zero-shot=true2026.02 | 75 | |
| MetaCLIP (+LazyStrike)Backbone=ViT-B/16, Zero-shot=true2026.02 | 72.8 | |
| CLIP (+LazyStrike)Backbone=ViT-L/14, Zero-shot=true2026.02 | 72.4 | |
| GoCAType=Ours, Backbone=Flux2026.03 | 70.7 | |
| SigLIP-L/14Params (M)=428, Evaluation=Linear probe2025.08 | 70.31 | |
| SigLIP-L/14Params (M)=428, Evaluation Protocol=Linear probe2025.08 | 70.31 | |
| OpenAI CLIP-L/14Params (M)=305, Evaluation=Linear probe2025.08 | 67.04 | |
| OpenAI CLIP-L/14Params (M)=305, Evaluation Protocol=Linear probe2025.08 | 67.04 | |
| GoCAType=Ours, Backbone=SD XL2026.03 | 65.6 | |
| GoCAType=Ours, Backbone=Pixart-Sigma2026.03 | 63.6 | |
| GoCAType=Ours, Backbone=SD v1.52026.03 | 60.7 | |
| DiffSegmentorType=Pre-Trained DM2026.03 | 60.1 | |
| EVACLIPBackbone=ViT-B/16, Zero-shot=true2026.02 | 56.5 | |
| FluxType=Vanilla2026.03 | 55.7 | |
| EVACLIPBackbone=ViT-L/14, Zero-shot=true2026.02 | 53.8 | |
| SD XLType=Vanilla2026.03 | 51.1 | |
| SD v1.5Type=Baseline2026.03 | 51.1 | |
| CLIPBackbone=ViT-B/16, Zero-shot=true2026.02 | 49 | |
| FTTMType=Pre-Trained DM, Note=Reproduced2026.03 | 48.9 | |
| MASt3R Enc.Params (M)=303, Evaluation=Linear probe2025.08 | 48.58 | |
| MASt3R Enc.Params (M)=303, Evaluation Protocol=Linear probe2025.08 | 48.58 | |
| DUSt3R Enc.Params (M)=303, Evaluation=Linear probe2025.08 | 46.02 | |
| DUSt3R Enc.Params (M)=303, Evaluation Protocol=Linear probe2025.08 | 46.02 | |
| Pixart-SigmaType=Vanilla2026.03 | 45.2 | |
| SD v1.5Type=Vanilla2026.03 | 44.3 | |
| MetaCLIPBackbone=ViT-B/16, Zero-shot=true2026.02 | 39.6 | |
| MaskCLIPType=Non-DM2026.03 | 38.8 | |
| SAM-H/16Params (M)=637, Evaluation=Linear probe2025.08 | 34.34 | |
| SAM-H/16Params (M)=637, Evaluation Protocol=Linear probe2025.08 | 34.34 | |
| MaskDiffusionType=Pre-Trained DM2026.03 | 29.9 | |
| MetaCLIPBackbone=ViT-L/14, Zero-shot=true2026.02 | 25.7 | |
| ReCOType=Non-DM2026.03 | 25.1 | |
| CLIPBackbone=ViT-L/14, Zero-shot=true2026.02 | 17.1 |