Out-of-Distribution Detection on Textures (test)
0.9309AUROCVIM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| VIM2026.03 | 0.9309 | — | 33.64 | 55.92 | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 3 + 42022.09 | 0.917 | 0.3729 | — | — | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 32022.09 | 0.9144 | 0.4184 | — | — | |
| KNN2026.03 | 0.9029 | — | 42.59 | 40.79 | |
| ASH2026.03 | 0.8953 | — | 38.34 | 34.78 | |
| RankFeatBackbone=T2T-ViT-242022.09 | 0.8936 | 0.3264 | — | — | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 42022.09 | 0.8933 | 0.4606 | — | — | |
| RankFeatBackbone=SqueezeNet, Block=Block 42022.09 | 0.8882 | 0.3804 | — | — | |
| SHE2026.03 | 0.8748 | — | 52.17 | 33.56 | |
| KL MatchingBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=20.6, Fine-tuned=true2021.05 | 0.8707 | 0.497 | — | — | |
| MDS2026.03 | 0.8626 | — | 55.95 | 38.55 | |
| GradNorm2026.03 | 0.8599 | — | 55.84 | 29.3 | |
| ReActBackbone=T2T-ViT-242022.09 | 0.8546 | 0.5254 | — | — | |
| RankFeatBackbone=SqueezeNet, Block=Block 3 + 42022.09 | 0.85 | 0.4254 | — | — | |
| Gram2026.03 | 0.8483 | — | 64.65 | 36.83 | |
| DICE2026.03 | 0.8389 | — | 63.84 | 28.6 | |
| ODINBackbone=T2T-ViT-242022.09 | 0.8363 | 0.5427 | — | — | |
| EnergyBackbone=T2T-ViT-242022.09 | 0.8325 | 0.5105 | — | — | |
| MSPBackbone=T2T-ViT-242022.09 | 0.8231 | 0.6291 | — | — | |
| MOSBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=3.2, Fine-tuned=true2021.05 | 0.8123 | 0.6043 | — | — | |
| ReAct2026.03 | 0.8108 | — | 53.47 | 19.39 | |
| GradNormBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.8107 | 0.6142 | — | — | |
| MahalanobisBackbone=T2T-ViT-242022.09 | 0.8067 | 0.496 | — | — | |
| RankFeatBackbone=SqueezeNet, Block=Block 32022.09 | 0.7932 | 0.5014 | — | — | |
| RAS2026.03 | 0.7809 | — | 54.23 | 12.8 | |
| RAS HL2026.03 | 0.7736 | — | 54.44 | 11.69 | |
| ReActBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.7685 | 0.7073 | — | — | |
| ReActBackbone=SqueezeNet2022.09 | 0.7657 | 0.5105 | — | — | |
| ODINBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=23.6, Fine-tuned=true2021.05 | 0.763 | 0.8131 | — | — | |
| ODINBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.763 | 0.8131 | — | — | |
| ODIN2026.03 | 0.7604 | — | 67.34 | 12.43 | |
| EnergyBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=3.1, Fine-tuned=true2021.05 | 0.7579 | 0.8087 | — | — | |
| EnergyBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.7579 | 0.8087 | — | — | |
| MSPBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=3.1, Fine-tuned=true2021.05 | 0.7445 | 0.8273 | — | — | |
| MSPBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.7445 | 0.8273 | — | — | |
| RMDS2026.03 | 0.7425 | — | 67.83 | 16.92 | |
| EBO2026.03 | 0.7389 | — | 66.05 | 12.09 | |
| OpenMax2026.03 | 0.7354 | — | 62.28 | 9.84 | |
| MLS2026.03 | 0.7342 | — | 66.34 | 11.26 | |
| MDSEns2026.03 | 0.7339 | — | 78.56 | 15.05 | |
| MahalanobisBackbone=BiT-S-R101x1, In-distribution dataset=ImageNet-1k, Test Time (min)=145.4, Fine-tuned=true2021.05 | 0.721 | 0.5223 | — | — | |
| MahalanobisBackbone=ResNetv2-101, Pre-training=ImageNet-1k2022.09 | 0.721 | 0.5223 | — | — | |
| GEN2026.03 | 0.7182 | — | 66.4 | 9.61 | |
| TempScale2026.03 | 0.7084 | — | 72.87 | 10.03 | |
| KLM2026.03 | 0.7073 | — | 68.69 | 8.98 | |
| MSP2026.03 | 0.6933 | — | 75.27 | 9.57 | |
| GradNormBackbone=SqueezeNet2022.09 | 0.6807 | 0.6872 | — | — | |
| MahalanobisBackbone=SqueezeNet2022.09 | 0.6716 | 0.586 | — | — | |
| RankFeat2026.03 | 0.6629 | — | 83.49 | 11.92 | |
| EnergyBackbone=SqueezeNet2022.09 | 0.6451 | 0.6716 | — | — | |
| ODINBackbone=SqueezeNet2022.09 | 0.4343 | 0.9225 | — | — | |
| MSPBackbone=SqueezeNet2022.09 | 0.4184 | 0.9461 | — | — | |
| GradNormBackbone=T2T-ViT-242022.09 | 0.388 | 0.9268 | — | — | |
| KNN (non-parametric)Backbone=ViT-B/16, Fine-tuned on=ImageNet-1k2022.04 | — | 0.3991 | — | — | |
| Mahalanobis (parametric)Backbone=ViT-B/16, Fine-tuned on=ImageNet-1k2022.04 | — | 0.7051 | — | — |