Out-of-Distribution Detection on CIFAR-100 ID Near-OOD Average OpenOOD v1.5
89.36AUROCOnline Prototype Learning
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Online Prototype LearningTraining Protocol=Zero-Shot Training-free Methods2026.05 | 89.36 | 45.7 | |
| ViMBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 88.5 | 41.31 | |
| ViMBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 88.47 | 40.22 | |
| OE + MSPTraining Protocol=Methods requiring training (or fine-tuning)2026.05 | 88.3 | — | |
| GradNormBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.94 | 49.13 | |
| ReActBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.92 | 47.7 | |
| KNNBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.72 | 42.57 | |
| KNNBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.55 | 44.95 | |
| EPDBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.37 | 43.76 | |
| SHEBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.23 | 57.54 | |
| EPDBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.16 | 45.39 | |
| ReActBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 87.06 | 46.44 | |
| MDSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.85 | 44.73 | |
| SHEBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.69 | 48.26 | |
| RMDSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.66 | 44.42 | |
| GradNormBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.57 | 46.97 | |
| RMDSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.41 | 43.85 | |
| EBOBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.36 | 55.23 | |
| GENBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.35 | 55.14 | |
| MLSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 86.07 | 55.23 | |
| GENBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 85.72 | 56.22 | |
| EBOBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 85.6 | 56.74 | |
| MLSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 85.39 | 56.61 | |
| MDSBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 84.91 | 46.99 | |
| AdaNeg+NeglabelTraining Protocol=Zero-Shot Training-free Methods2026.05 | 84.62 | 59.07 | |
| DICEBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 84.29 | 62.54 | |
| TempScaleBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 84.28 | 56.69 | |
| TempScaleBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 84.1 | 56.67 | |
| MSPBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 82.76 | 57.22 | |
| MSPBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 82.68 | 58.17 | |
| DICEBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 82.63 | 65.08 | |
| OpenMaxBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 82.26 | 58.93 | |
| OpenMaxBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 81.98 | 61.74 | |
| GENTraining Protocol=Methods requiring training (or fine-tuning)2026.05 | 81.31 | — | |
| SCALETraining Protocol=Methods requiring training (or fine-tuning)2026.05 | 80.99 | — | |
| VOS + EBOTraining Protocol=Methods requiring training (or fine-tuning)2026.05 | 80.93 | — | |
| MCMTraining Protocol=Zero-Shot Training-free Methods2026.05 | 71 | 75.2 | |
| NegLabelTraining Protocol=Zero-Shot Training-free Methods2026.05 | 70.58 | 71.44 | |
| ASHBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 55.38 | 86.68 | |
| ASHBackbone=ViT-B/16, Pre-trained Model=google/vit-base-patch16-224-in21k, ID Accuracy=84.78%, Evaluation Protocol=fine-tuned2026.04 | 54.09 | 89.01 |