OOD Detection on OpenOOD CIFAR100 v1.5 (test)
83.06AUROCMahaVar
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MahaVarBackbone=ResNet-502026.05 | 83.06 | — | — | — | — | 68.5 | |
| Mahalanobis++Backbone=ResNet-502026.05 | 82.01 | — | — | — | — | 70.72 | |
| ASH-SBackbone=ResNet-502026.05 | 80.87 | — | — | — | — | 75.46 | |
| SCALEBackbone=ResNet-502026.05 | 80.64 | — | — | — | — | 76.17 | |
| NCIBackbone=ResNet-502026.05 | 80.61 | — | — | — | — | 78.44 | |
| NCI+filterBackbone=ResNet-502026.05 | 80.61 | — | — | — | — | 78.44 | |
| IDBDBackbone=ResNet-502026.05 | 79.99 | — | — | — | — | 78.43 | |
| NNGuideBackbone=ResNet-502026.05 | 79.88 | — | — | — | — | 78.76 | |
| ReActBackbone=ResNet-502026.05 | 79.56 | — | — | — | — | 78.93 | |
| EnergyBackbone=ResNet-502026.05 | 79.48 | — | — | — | — | 78.91 | |
| MaxLogitBackbone=ResNet-502026.05 | 79.46 | — | — | — | — | 79.41 | |
| GENBackbone=ResNet-502026.05 | 79.42 | — | — | — | — | 79.32 | |
| ViMBackbone=ResNet-502026.05 | 79.05 | — | — | — | — | 78.88 | |
| KL-matchingBackbone=ResNet-502026.05 | 78.27 | — | — | — | — | 78.04 | |
| MSPBackbone=ResNet-502026.05 | 77.95 | — | — | — | — | 80.53 | |
| KNNBackbone=ResNet-502026.05 | 77.63 | — | — | — | — | 79.04 | |
| MahalanobisBackbone=ResNet-502026.05 | 66.67 | — | — | — | — | 90.68 | |
| AdaNegTraining Type=Zero-shot Training-free Methods2024.10 | — | 59.07 | 29.35 | 84.62 | 95.25 | — | |
| AdaNegTraining Requirement=Training-free & test-time adaptation methods2026.03 | — | 59.07 | 29.35 | 84.62 | 95.25 | — | |
| GENTraining Type=Methods requiring training (or fine-tuning)2024.10 | — | — | — | 81.31 | 79.68 | — | |
| GENTraining Requirement=Methods requiring training (or fine-tuning)2026.03 | — | — | — | 81.31 | 79.68 | — | |
| MCMTraining Type=Zero-shot Training-free Methods2024.10 | — | 75.2 | 59.32 | 71 | 76 | — | |
| MCMTraining Requirement=Training-free & non-adaptive methods2026.03 | — | 75.2 | 59.32 | 71 | 76 | — | |
| NegLabelTraining Type=Zero-shot Training-free Methods2024.10 | — | 71.44 | 40.92 | 70.58 | 89.68 | — | |
| NegLabelTraining Requirement=Training-free & non-adaptive methods2026.03 | — | 71.44 | 40.92 | 70.58 | 89.68 | — | |
| OE + MSPTraining Type=Methods requiring training (or fine-tuning)2024.10 | — | — | — | 88.3 | 81.41 | — | |
| OE + MSPTraining Requirement=Methods requiring training (or fine-tuning)2026.03 | — | — | — | 88.3 | 81.41 | — | |
| SCALETraining Type=Methods requiring training (or fine-tuning)2024.10 | — | — | — | 80.99 | 81.42 | — | |
| SCALETraining Requirement=Methods requiring training (or fine-tuning)2026.03 | — | — | — | 80.99 | 81.42 | — | |
| TANLTraining Requirement=Training-free & test-time adaptation methods2026.03 | — | 57.74 | 24.55 | 85.06 | 95.41 | — | |
| VOS + EBOTraining Type=Methods requiring training (or fine-tuning)2024.10 | — | — | — | 80.93 | 81.32 | — | |
| VOS + EBOTraining Requirement=Methods requiring training (or fine-tuning)2026.03 | — | — | — | 80.93 | 81.32 | — |