Semantic-level object discovery on COCO
46.7mIoUGLASS-Sem
Evaluation Results
| Method | Links | |
|---|---|---|
| GLASS-SemDownstream tasks=iOD + PP + CG + CPG, Input=I, Pre-trained models=SD + DINOv2 + BLIP-22024.07 | 46.7 | |
| SPOTDownstream tasks=iOD + PP, Input=I, Pre-trained models=DINO2024.07 | 44.6 | |
| ToCoDownstream tasks=sOD, Input=I + L2024.07 | 41.3 | |
| DINOSAUR-TransformerDownstream tasks=iOD + PP, Input=I, Pre-trained models=DINO2024.07 | 40.6 | |
| AFADownstream tasks=sOD, Input=I + L2024.07 | 38.9 | |
| OVDiffDownstream tasks=sOD, Input=I + C, Pre-trained models=SD + DINO + CLIP2024.07 | 34.6 | |
| Dataset DiffusionDownstream tasks=sOD, Input=I, Pre-trained models=SD + BLIP-22024.07 | 34.2 | |
| DiffCutDownstream tasks=sOD, Input=I, Pre-trained models=SD2024.07 | 34.1 | |
| EmerDiffDownstream tasks=sOD, Input=I, Pre-trained models=SD2024.07 | 33.1 | |
| CLIPPYDownstream tasks=sOD, Input=I + C, Pre-trained models=CLIP2024.07 | 32 | |
| DINOSAUR-MLPDownstream tasks=iOD + PP, Input=I, Pre-trained models=DINO2024.07 | 31.7 | |
| StableLSDDownstream tasks=iOD + PP + CG, Input=I, Pre-trained models=SD + DINOv22024.07 | 29.5 | |
| SegCLIPDownstream tasks=sOD, Input=I + C, Pre-trained models=CLIP2024.07 | 26.5 | |
| OVSegDownstream tasks=sOD, Input=I + C, Pre-trained models=CLIP2024.07 | 25.1 | |
| MaskCLIPDownstream tasks=sOD, Input=I + C, Pre-trained models=CLIP2024.07 | 20.6 |