Semantic Segmentation on Pascal Context 59 without background (val)
63.4Pixel AccuracyProxyCLIP
Evaluation Results
| Method | Links | |
|---|---|---|
| ProxyCLIPPub. & Year=ECCV’24, Extra backbone=DINOv2† (ViT-B/14), Protocol=Training-free2026.03 | 63.4 | |
| PEARLPub. & Year=–, Extra backbone=✗, Protocol=Training-free2026.03 | 62.9 | |
| CASSPub. & Year=CVPR’25, Extra backbone=DINOv3 (ViT-B/16), Protocol=Training-free2026.03 | 62.3 | |
| SFPPub. & Year=ICCV’25, Extra backbone=✗, Protocol=Training-free2026.03 | 60.7 | |
| CASSPub. & Year=CVPR’25, Extra backbone=DINOv2† (ViT-B/14), Protocol=Training-free2026.03 | 60.1 | |
| NACLIPPub. & Year=WACV’25, Extra backbone=✗, Protocol=Training-free2026.03 | 59.8 | |
| LaVGPub. & Year=ECCV’24, Extra backbone=DINOv1 (ViT-B/8), Protocol=Training-free2026.03 | 58.9 | |
| CLIPtrasePub. & Year=ECCV’24, Extra backbone=✗, Protocol=Training-free2026.03 | 58.9 | |
| SCLIPPub. & Year=ECCV’24, Extra backbone=✗, Protocol=Training-free2026.03 | 58.3 |