Multi-object representation on SimCo
79.72Accuracy (2 objects)UNCHA
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| UNCHABackbone=ViT-B/162026.03 | 79.72 | 86.93 | 89.75 | 90.65 | — | — | — | — | |
| MERUBackbone=ViT-B/162026.03 | 77.82 | 83.91 | 85.79 | 86.9 | — | — | — | — | |
| CLIPBackbone=ViT-B/162026.03 | 77.15 | 84.58 | 87.4 | 88.48 | — | — | — | — | |
| HyCoCLIPBackbone=ViT-B/162026.03 | 75.71 | 81.13 | 82.41 | 82.85 | — | — | — | — | |
| UNCHABackbone=ViT-S/162026.03 | 74.41 | 81.79 | 83.55 | 83.13 | — | — | — | — | |
| HyCoCLIPBackbone=ViT-S/162026.03 | 72.91 | 78.25 | 79.55 | 79.43 | — | — | — | — | |
| CLIPBackbone=ViT-S/162026.03 | 72.49 | 80.05 | 82.45 | 82.65 | — | — | — | — | |
| MERUBackbone=ViT-S/162026.03 | 71.69 | 78.56 | 80.65 | 81.2 | — | — | — | — | |
| ATMGBackbone=ViT-B/162026.03 | 65.52 | 65.32 | 65.28 | 65.12 | — | — | — | — | |
| ATMGBackbone=ViT-S/162026.03 | 62.17 | 63.02 | 61.83 | 62 | — | — | — | — | |
| ATMGBackbone=ViT-B/16, Zero-shot=true2026.03 | — | — | — | — | 65.52 | 65.32 | 65.28 | 65.12 | |
| CLIPBackbone=ViT-B/16, Zero-shot=true2026.03 | — | — | — | — | 77.15 | 84.58 | 87.4 | 88.48 | |
| HyCoCLIPBackbone=ViT-B/16, Zero-shot=true2026.03 | — | — | — | — | 75.71 | 81.13 | 82.41 | 82.85 | |
| MERUBackbone=ViT-B/16, Zero-shot=true2026.03 | — | — | — | — | 77.82 | 83.91 | 85.79 | 86.9 | |
| UNCHABackbone=ViT-B/16, Zero-shot=true2026.03 | — | — | — | — | 79.72 | 86.93 | 89.75 | 90.65 |