Depth classification on 3D physical scene properties (test)
99.7ROC AUCDINOv2
Evaluation Results
| Method | Links | |
|---|---|---|
| DINOv2Backbone=ViT-G, Pre-training dataset=LVD-142M, Evaluation protocol=SVM grid search (best layer and C)2023.10 | 99.7 | |
| Stable DiffusionEvaluation protocol=SVM grid search (best time step, layer and C)2023.10 | 99.6 | |
| OpenCLIPBackbone=ViT-G, Pre-training dataset=LAION, Evaluation protocol=SVM grid search (best layer and C)2023.10 | 99.2 | |
| DINOv1Backbone=ViT-B, Pre-training dataset=ImageNet, Evaluation protocol=SVM grid search (best layer and C)2023.10 | 97.1 | |
| VQGANBackbone=48-layer transformer, Pre-training dataset=ImageNet, Evaluation protocol=SVM grid search (best layer and C)2023.10 | 91.8 | |
| RandomDescription=Random classifier2023.10 | 50 |