Clustering on ImageNet 1k (val)
58ACCMAE-CTmin
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MAE-CTminBackbone=ViT-H/16, Augmentation Strategy=minimal image augmentations2023.04 | 58 | 81.8 | 69.3 | 36.8 | |
| MAE-CTaugBackbone=ViT-L/16, Augmentation Strategy=extensive image augmentations2023.04 | 56.9 | 81.7 | 68.4 | 40.5 | |
| MAE-CTminBackbone=ViT-L/16, Augmentation Strategy=minimal image augmentations2023.04 | 54.9 | 80.6 | 66.8 | 34.5 | |
| MAE-CTaugBackbone=ViT-H/16, Augmentation Strategy=extensive image augmentations2023.04 | 54.8 | 81.1 | 67.6 | 38.9 | |
| MSNBackbone=ViT-B/16, Augmentation Strategy=extensive image augmentations2023.04 | 54.2 | 79.9 | 66.6 | 30.2 | |
| iBOTBackbone=ViT-L/16, Augmentation Strategy=extensive image augmentations2023.04 | 52 | 80.5 | 66.8 | 35 | |
| iBOTBackbone=ViT-B/16, Augmentation Strategy=extensive image augmentations2023.04 | 50 | 79.1 | 64.9 | 32.6 | |
| DINOBackbone=ViT-B/16, Augmentation Strategy=extensive image augmentations2023.04 | 48 | 77.9 | 63.2 | 31.2 | |
| MAE-CTaugBackbone=ViT-B/16, Augmentation Strategy=extensive image augmentations2023.04 | 46.2 | 76 | 59 | 29.8 | |
| MSNBackbone=ViT-L/16, Augmentation Strategy=extensive image augmentations2023.04 | 45.4 | 76.2 | 63.2 | 19.5 | |
| Moco v3Backbone=ViT-B/16, Augmentation Strategy=extensive image augmentations2023.04 | 43 | 73 | 55.5 | 23.1 | |
| MAE-CTminBackbone=ViT-B/16, Augmentation Strategy=minimal image augmentations2023.04 | 35.3 | 68.2 | 45.8 | 19.7 | |
| MAE (reimpl.)Backbone=ViT-L/16, Augmentation Strategy=minimal image augmentations2023.04 | 14.3 | 51.9 | 22.8 | 4.6 | |
| MAE (reimpl.)Backbone=ViT-B/16, Augmentation Strategy=minimal image augmentations2023.04 | 13.8 | 51.4 | 20.7 | 4.4 | |
| MAE (reimpl.)Backbone=ViT-H/16, Augmentation Strategy=minimal image augmentations2023.04 | 11.1 | 48.2 | 17.1 | 2.9 |