Object Discovery on Pascal VOC 2012 (val)
50.2CORLOCEgoViTWT-all
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EgoViTWT-allBackbone=ViT-S, Pre-training Dataset=WT-all2026.03 | 50.2 | — | |
| EgoViTWT-Sub5Backbone=ViT-S, Pre-training Dataset=WT-Sub52026.03 | 48.9 | — | |
| EgoViTVeniceBackbone=ViT-S, Pre-training Dataset=Venice video2026.03 | 45.4 | — | |
| EgoViTBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 45.2 | — | |
| MoCo-v3Backbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 43.1 | — | |
| AttMaskBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 37.8 | — | |
| DINOBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 37.2 | — | |
| iBOTBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 36.3 | — | |
| MAEBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 35.9 | — | |
| SimCLRBackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 35.5 | — | |
| SAVi++Backbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 34.2 | — | |
| PooDLeVeniceBackbone=ResNet-50, Pre-training Dataset=Venice video2026.03 | 32.7 | — | |
| DORABackbone=ViT-S, Pre-training Dataset=Zurich video2026.03 | 24.1 | — | |
| DORAVeniceBackbone=ViT-S, Pre-training Dataset=Venice video2026.03 | 23.9 | — | |
| Cyclic walksNumber of slots=4, Image Size=2242023.02 | — | 29.6 | |
| DINOSAURImplementation=Original, Number of slots=4, Image Size=2242023.02 | — | 24.6 | |
| DINOSAURImplementation=Re-implemented, Number of slots=4, Image Size=2242023.02 | — | 27.5 |