Class Retrieval on PascalVOC
93.5Top-1 AccuracyMaskInversion
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MaskInversionBackbone=ViT-H/14, Zero-shot status=true2024.07 | 93.5 | 99.4 | 99.7 | |
| MaskInversionBackbone=H/142024.07 | 93.5 | — | — | |
| MaskInversionBackbone=ViT-L/14, Zero-shot status=true2024.07 | 91 | 99.1 | 99.8 | |
| MaskInversionBackbone=L/142024.07 | 91 | — | — | |
| MaskInversionBackbone=ViT-B/16, Zero-shot status=true2024.07 | 85.4 | 96.4 | 98.8 | |
| MaskInversionBackbone=B/162024.07 | 85.4 | — | — | |
| MaskInversionBackbone=ViT-B/32, Zero-shot status=true2024.07 | 79.5 | 96.4 | 98.8 | |
| CLIPSurgeryBackbone=L/142024.07 | 78.3 | — | — | |
| RISBackbone=ViT-B/16, Zero-shot status=true2024.07 | 78 | 95.2 | 98.1 | |
| Masked CropBackbone=ViT-B/16, Zero-shot status=true2024.07 | 75 | 91.4 | 96.4 | |
| MaskCLIPBackbone=B/162024.07 | 74.9 | — | — | |
| FGVPBackbone=ViT-B/16, Zero-shot status=true2024.07 | 71.8 | 93.6 | 98.3 | |
| CLIPSurgeryBackbone=B/162024.07 | 70.8 | — | — | |
| CLIPSurgeryBackbone=H/142024.07 | 68 | — | — | |
| SCLIPBackbone=B/162024.07 | 64.3 | — | — | |
| MaskCLIPBackbone=H/142024.07 | 61.8 | — | — | |
| MaskCLIPBackbone=L/142024.07 | 55.1 | — | — | |
| AlphaCLIPBackbone=ViT-B/16, Zero-shot status=true2024.07 | 52.6 | 85.9 | 93.8 | |
| RedCircleBackbone=ViT-B/16, Zero-shot status=true2024.07 | 47.5 | 92.9 | 97.7 | |
| SCLIPBackbone=L/142024.07 | 43 | — | — | |
| CLIPBackbone=ViT-B/16, Zero-shot status=true2024.07 | 40.1 | 87.2 | 95.6 | |
| SCLIPBackbone=H/142024.07 | 38.2 | — | — | |
| CropBackbone=ViT-B/16, Zero-shot status=true2024.07 | 27.9 | 51.2 | 72.4 |