Object-Centric Learning on PASCAL
50.4MBO^iSlotDiffusion
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SlotDiffusion2023.12 | 50.4 | 55.3 | — | |
| QASADec.=Trans., K-type=adaptive2026.01 | 49.7 | 57.9 | 47.9 | |
| SPOT+FS+RCDec.=Trans., K-type=fixed2026.01 | 49.3 | 56.5 | — | |
| SPOTEnsemble=true, Seeds=32023.12 | 48.3 | 55.6 | — | |
| SPOTDec.=Trans., K-type=fixed2026.01 | 48.3 | 55.6 | — | |
| SPOTEnsemble=false, Seeds=32023.12 | 48.1 | 55.3 | — | |
| QASA-MLPDec.=MLP, K-type=adaptive2026.01 | 47.3 | 53.8 | 45.5 | |
| DINOSAUR+FS+RCDec.=MLP, K-type=fixed2026.01 | 45.2 | 52.9 | — | |
| DINOSAUREncoder=ViT-B/16, Decoder=Autoregressive2023.12 | 44 | 51.2 | — | |
| DINOSAURDec.=Trans., K-type=fixed2026.01 | 44 | 51.2 | — | |
| MetaSlotDec.=MLP, K-type=adaptive2026.01 | 42.1 | 43.9 | — | |
| Rotating Features2023.12 | 40.7 | 46 | — | |
| DINOSAUR-MLPEncoder=ViT-B/16, Decoder=MLP2023.12 | 39.5 | 40.9 | — | |
| DINOSAUR-MLPDec.=MLP, K-type=fixed2026.01 | 39.5 | 40.9 | — | |
| SLATE2023.12 | 35.9 | 41.5 | — | |
| SLATE2026.01 | 35.9 | 41.5 | — | |
| CAE2023.12 | 32.9 | 37.4 | — | |
| SA2023.12 | 24.6 | 24.9 | — |