Out-of-Distribution Detection on Textures
0.9943AUROCRODD
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RODDID dataset=CIFAR-102022.04 | 0.9943 | 0.0387 | — | — | — | |
| FSID dataset=CIFAR-102022.04 | 0.9864 | 0.0555 | — | — | — | |
| VIMBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9797 | 10.51 | — | — | — | |
| NACBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.979 | — | — | — | — | |
| OEID dataset=CIFAR-102022.04 | 0.9773 | 0.1294 | — | — | — | |
| MahalanobisID dataset=CIFAR-102022.04 | 0.9733 | 0.15 | — | — | — | |
| KNNBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.9718 | 0.1156 | — | — | — | |
| KNNBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9709 | 17.31 | — | — | — | |
| ASHBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.969 | 15.26 | — | — | — | |
| APEXBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 0.9623 | 0.1659 | — | — | — | |
| APMBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 0.9621 | 0.1673 | — | — | — | |
| WeiPer+KLDBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9618 | 22.17 | — | — | — | |
| VRA+Backbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.9608 | 0.1966 | — | — | — | |
| PALMBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 0.9594 | 0.1748 | — | — | — | |
| VRABackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.9562 | 0.2147 | — | — | — | |
| CIDERBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 0.9544 | 0.1931 | — | — | — | |
| VRA+Backbone=ResNetv2-101, Pre-trained=ImageNet2023.02 | 0.9543 | 0.2388 | — | — | — | |
| VRABackbone=ResNetv2-101, Pre-trained=ImageNet2023.02 | 0.9422 | 0.3069 | — | — | — | |
| kNN+Backbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 0.9422 | 0.266 | — | — | — | |
| NACBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.9417 | — | — | — | — | |
| RODDID Dataset=CIFAR-1002022.04 | 0.9414 | 0.2464 | — | — | — | |
| CLELCategory=EBM2023.12 | 0.94 | — | 2 | — | — | |
| SHEBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.936 | 35.27 | — | — | — | |
| DynProtoBackbone=ViT-B/162026.04 | 0.9314 | 0.208 | — | — | — | |
| ReActBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9279 | 29.64 | — | — | — | |
| APEXBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9269 | 0.3331 | — | — | — | |
| SHEBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.9265 | 25.63 | — | — | — | |
| APMBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9244 | 0.3367 | — | — | — | |
| DynProtoBackbone=ConvNeXt-B2026.04 | 0.9213 | 0.2312 | — | — | — | |
| GradNormBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9205 | 43.27 | — | — | — | |
| DICEBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9204 | 44.28 | — | — | — | |
| IDLikeBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.9189 | 0.3282 | — | — | — | |
| WeiPer+ReActBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.9188 | 34.95 | — | — | — | |
| PALMBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9143 | 0.4406 | — | — | — | |
| NPOSBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9135 | 0.2492 | — | — | — | |
| KNNBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.9112 | 33.23 | — | — | — | |
| Projection RegretCategory=Diffusion model2023.12 | 0.91 | — | 1.5 | — | — | |
| DMPLBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9093 | 0.3509 | — | — | — | |
| VIMBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.9061 | 40.35 | — | — | — | |
| MahalanobisID Dataset=CIFAR-1002022.04 | 0.9057 | 0.3939 | — | — | — | |
| CIDERBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.9042 | 0.4394 | — | — | — | |
| CoOpBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.904 | 0.4186 | — | — | — | |
| DICEBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.903 | 0.3172 | — | — | — | |
| GENBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.9023 | 38.3 | — | — | — | |
| GCOS + LregTraining Dataset=CIFAR-102026.03 | 0.9019 | 0.505 | — | — | — | |
| ReActBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.898 | 0.473 | — | — | — | |
| MDSBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.898 | 42.79 | — | — | — | |
| IDLikeBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=1-shot2024.11 | 0.8967 | 0.3834 | — | — | — | |
| CADRefBackbone=ViT-B/162026.04 | 0.8959 | 0.4978 | — | — | — | |
| MDSBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8941 | 38.91 | — | — | — | |
| RMDSBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8938 | 37.22 | — | — | — | |
| LSNBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.8938 | 0.4643 | — | — | — | |
| WeiPer+KLDBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8936 | 46.1 | — | — | — | |
| CADRefBackbone=ConvNeXt-B2026.04 | 0.8927 | 0.4917 | — | — | — | |
| SCTBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.8906 | 0.4151 | — | — | — | |
| ODINBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.89 | 49.24 | — | — | — | |
| VIMBackbone=ViT-B/162026.04 | 0.8892 | 0.4648 | — | — | — | |
| LoCoOpBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.8886 | 0.4527 | — | — | — | |
| NPOSEvaluation Protocol=fine-tuned2023.03 | 0.888 | 0.4612 | — | — | — | |
| EBOBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.887 | 45.77 | — | — | — | |
| MaxLogitBackbone=CLIP-ViT-B/16, Setting=Post-hoc2024.11 | 0.8863 | 0.4872 | — | — | — | |
| MSPID dataset=CIFAR-102022.04 | 0.885 | 0.5928 | — | — | — | |
| MLSBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.8839 | 46.17 | — | — | — | |
| NegPromptBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=16-shot2024.11 | 0.8838 | 0.4393 | — | — | — | |
| ODIN+UMAPBackbone=ResNet-50, Pre-trained=PyTorch2023.06 | 0.8835 | 0.4202 | — | — | — | |
| kNN+Backbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.8835 | 0.5715 | — | — | — | |
| CoOpBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=1-shot2024.11 | 0.8834 | 0.4961 | — | — | — | |
| VOSTraining Dataset=CIFAR-102026.03 | 0.8834 | 0.49 | — | — | — | |
| WeiPer+ReActBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8831 | 47.37 | — | — | — | |
| EnergyBackbone=CLIP-ViT-B/16, Setting=Post-hoc2024.11 | 0.8822 | 0.5039 | — | — | — | |
| ReActBackbone=CLIP-ViT-B/16, Setting=Post-hoc2024.11 | 0.8813 | 0.4988 | — | — | — | |
| OpenMaxBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.881 | 40.26 | — | — | — | |
| WeiPer+MSPBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8808 | 48.62 | — | — | — | |
| GramBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.8802 | 58.8 | — | — | — | |
| EnergyEvaluation Protocol=fine-tuned2023.03 | 0.88 | 0.5135 | — | — | — | |
| Improved CDCategory=EBM2023.12 | 0.88 | — | 3.25 | — | — | |
| SHEBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.8793 | 0.4509 | — | — | — | |
| ODINEvaluation Protocol=fine-tuned2023.03 | 0.8785 | 0.5167 | — | — | — | |
| GENBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.8759 | 46.22 | — | — | — | |
| MSP+UMAPBackbone=ResNet-50, Pre-trained=PyTorch2023.06 | 0.8737 | 0.5161 | — | — | — | |
| VOSBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.8731 | 0.5757 | — | — | — | |
| ViMEvaluation Protocol=fine-tuned2023.03 | 0.8718 | 0.5394 | — | — | — | |
| LoCoOpBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=1-shot2024.11 | 0.8685 | 0.5153 | — | — | — | |
| VIMBackbone=ConvNeXt-B2026.04 | 0.8679 | 0.5241 | — | — | — | |
| VOSEvaluation Protocol=fine-tuned2023.03 | 0.8674 | 0.5667 | — | — | — | |
| ReActBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8666 | 55.88 | — | — | — | |
| SCTBackbone=CLIP-ViT-B/16, Setting=Prompt tuning, Shots=1-shot2024.11 | 0.8666 | 0.4887 | — | — | — | |
| VOSID dataset=CIFAR-102022.04 | 0.8664 | 0.4709 | — | — | — | |
| FSID Dataset=CIFAR-1002022.04 | 0.8664 | 0.4709 | — | — | — | |
| WeiPer+MSPBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.8662 | 55.16 | — | — | — | |
| MSP+UMBackbone=ResNet-50, Pre-trained=PyTorch2023.06 | 0.865 | 0.4622 | — | — | — | |
| KLMBackbone=ViT16-B, Training Objective=Cross Entropy, Preprocessing=standard2024.05 | 0.8649 | 50.12 | — | — | — | |
| CSIBackbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.8647 | 0.6161 | — | — | — | |
| OptFSBackbone=ViT-B/162026.04 | 0.8646 | 0.566 | — | — | — | |
| VOS+Evaluation Protocol=fine-tuned2023.03 | 0.8633 | 0.6102 | — | — | — | |
| SSD+Backbone=ResNet-34, ID Dataset=CIFAR-100, Training Supervision=Labeled2026.02 | 0.8618 | 0.6663 | — | — | — | |
| OptFSBackbone=ConvNeXt-B2026.04 | 0.8611 | 0.5432 | — | — | — | |
| RMDSBackbone=ResNet50, Training Loss=Cross Entropy, Preprocessing=Standard Preprocessing2024.05 | 0.8608 | 48.8 | — | — | — | |
| ASHTraining Dataset=CIFAR-102026.03 | 0.8607 | 0.509 | — | — | — | |
| EnergyBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 0.8599 | 0.5372 | — | — | — |