Out-of-Distribution Detection on CIFAR-10 (in-dist) vs iSUN (out-of-dist) (test)
99.77AUROCDynProto
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DynProtoBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 99.77 | — | — | — | 0.75 | |
| DynProtoBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 99.62 | — | — | — | 2.17 | |
| AdaNDBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 99.55 | — | — | — | 0.76 | |
| VIMBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 99.13 | — | — | — | 4.52 | |
| CADRefBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 99.13 | — | — | — | 4.35 | |
| DICEBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 98.99 | — | — | — | 5.15 | |
| ASH-SBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 98.88 | — | — | — | 5.14 | |
| EnergyBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 98.11 | — | — | — | 9.49 | |
| OptFSBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 97.98 | — | — | — | 10.49 | |
| CSPBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 97.95 | — | — | — | 9.06 | |
| NegLabelBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 97.7 | — | — | — | 8.87 | |
| MCMBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 97.31 | — | — | — | 12.5 | |
| GLMCMBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 95.92 | — | — | — | 18.97 | |
| NegRefineBackbone=CLIP-B/16, Method Category=Vision Language Model based Methods2026.04 | 95 | — | — | — | 28.26 | |
| MSPBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 94.49 | — | — | — | 42.54 | |
| ReActBackbone=ResNet50, Method Category=Vision Model based Methods2026.04 | 83.78 | — | — | — | 73.66 | |
| OursBackbone=ResNet2019.12 | 0.998 | 99.3 | 98.1 | — | — | |
| OursBackbone=DenseNet2019.12 | 0.998 | 99 | 97.9 | — | — | |
| OursModel Architecture=ResNet2019.12 | 0.998 | 99.3 | 98.1 | — | — | |
| OursModel Architecture=DenseNet2019.12 | 0.998 | 99 | 97.9 | — | — | |
| MahalanobisBackbone=ResNet2019.12 | 0.995 | 97.8 | 96.7 | — | — | |
| MahalanobisModel Architecture=ResNet2019.12 | 0.995 | 97.8 | 96.7 | — | — | |
| MahalanobisBackbone=DenseNet2019.12 | 0.989 | 95.3 | 95.2 | — | — | |
| MahalanobisModel Architecture=DenseNet2019.12 | 0.989 | 95.3 | 95.2 | — | — | |
| ODIN_OODbaseline=ODIN2021.06 | 0.988 | — | — | 6.3 | — | |
| ODINBackbone=DenseNet2019.12 | 0.987 | 93.2 | 94.3 | — | — | |
| ODINModel Architecture=DenseNet2019.12 | 0.987 | 93.2 | 94.3 | — | — | |
| D_alphadetector=DOCTOR, variant=alpha2021.06 | 0.981 | — | — | 8 | — | |
| D_betadetector=DOCTOR, variant=beta2021.06 | 0.979 | — | — | 9.1 | — | |
| BaselineBackbone=DenseNet2019.12 | 0.947 | 62.5 | 89.2 | — | — | |
| BaselineModel Architecture=DenseNet2019.12 | 0.947 | 62.5 | 89.2 | — | — | |
| ODINBackbone=ResNet2019.12 | 0.94 | 73.2 | 86.5 | — | — | |
| ODINModel Architecture=ResNet2019.12 | 0.94 | 73.2 | 86.5 | — | — | |
| BaselineBackbone=ResNet2019.12 | 0.91 | 44.6 | 85 | — | — | |
| BaselineModel Architecture=ResNet2019.12 | 0.91 | 44.6 | 85 | — | — |