Out-of-Distribution Detection on Places
9.59FPR95RODD
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RODDID dataset=CIFAR-102022.04 | 9.59 | 98.47 | — | — | |
| LoCoOpScoring Function=MaxLogit, Variant=CLS-M2024.05 | 11.05 | — | — | -16.52 | |
| FSID dataset=CIFAR-102022.04 | 11.56 | 96.42 | — | — | |
| KgCoOpScoring Function=MaxLogit, Variant=CLS-M2024.05 | 12.46 | — | — | -1.97 | |
| CoCoOpScoring Function=MaxLogit, Variant=CLS-M2024.05 | 12.92 | — | — | -0.09 | |
| CoCoOpScoring Function=MaxLogit, Variant=Original2024.05 | 13.01 | — | — | — | |
| KgCoOpScoring Function=Energy, Variant=CLS-E2024.05 | 13.33 | — | — | -4.99 | |
| CoOpScoring Function=MaxLogit, Variant=CLS-M2024.05 | 13.56 | — | — | -4.01 | |
| PromptSRCScoring Function=MaxLogit, Variant=CLS-M2024.05 | 13.77 | — | — | -1.11 | |
| MaPLeScoring Function=MaxLogit, Variant=CLS-M2024.05 | 14.01 | — | — | -1.37 | |
| CoCoOpScoring Function=Energy, Variant=CLS-E2024.05 | 14.09 | — | — | -1.33 | |
| LoCoOpScoring Function=Energy, Variant=CLS-E2024.05 | 14.25 | — | — | -39.33 | |
| CoOpScoring Function=Energy, Variant=CLS-E2024.05 | 14.42 | — | — | -6.48 | |
| KgCoOpScoring Function=MaxLogit, Variant=Original2024.05 | 14.43 | — | — | — | |
| PromptSRCScoring Function=MaxLogit, Variant=Original2024.05 | 14.88 | — | — | — | |
| IVLPScoring Function=MaxLogit, Variant=CLS-M2024.05 | 15.32 | — | — | -2.5 | |
| MaPLeScoring Function=MaxLogit, Variant=Original2024.05 | 15.37 | — | — | — | |
| CoCoOpScoring Function=Energy, Variant=Original2024.05 | 15.41 | — | — | — | |
| PromptSRCScoring Function=Energy, Variant=CLS-E2024.05 | 16.04 | — | — | -2.66 | |
| ProGradScoring Function=MaxLogit, Variant=CLS-M2024.05 | 16.28 | — | — | -6.19 | |
| IVLPScoring Function=Energy, Variant=CLS-E2024.05 | 16.67 | — | — | -5.47 | |
| MaPLeScoring Function=Energy, Variant=CLS-E2024.05 | 16.74 | — | — | -2.62 | |
| CoOpScoring Function=MaxLogit, Variant=Original2024.05 | 17.58 | — | — | — | |
| IVLPScoring Function=MaxLogit, Variant=Original2024.05 | 17.82 | — | — | — | |
| ProGradScoring Function=Energy, Variant=CLS-E2024.05 | 17.85 | — | — | -10.52 | |
| KgCoOpScoring Function=Energy, Variant=Original2024.05 | 18.32 | — | — | — | |
| PromptSRCScoring Function=Energy, Variant=Original2024.05 | 18.7 | — | — | — | |
| OEID dataset=CIFAR-102022.04 | 19.07 | 96.16 | — | — | |
| MaPLeScoring Function=Energy, Variant=Original2024.05 | 19.35 | — | — | — | |
| CoOpScoring Function=Energy, Variant=Original2024.05 | 20.9 | — | — | — | |
| IVLPScoring Function=Energy, Variant=Original2024.05 | 22.14 | — | — | — | |
| ProGradScoring Function=MaxLogit, Variant=Original2024.05 | 22.47 | — | — | — | |
| LAPSBackbone=ResNet-502026.02 | 24.71 | 93.64 | — | — | |
| LoCoOpScoring Function=MaxLogit, Variant=Original2024.05 | 27.58 | — | — | — | |
| ProGradScoring Function=Energy, Variant=Original2024.05 | 28.38 | — | — | — | |
| Catalyst(μ) + ReActBackbone=ResNet-502026.02 | 28.98 | 93.31 | — | — | |
| Catalyst(σ) + ReActBackbone=ResNet-502026.02 | 29.58 | 93.04 | — | — | |
| Catalyst(m) + ReActBackbone=ResNet-502026.02 | 29.77 | 92.92 | — | — | |
| DynProtoBackbone=ConvNeXt-B2026.04 | 30.06 | 91.44 | — | — | |
| Adaptive Multi-prompt Contrastive NetworkShot=82025.06 | 32.45 | 93.96 | — | — | |
| Adaptive Multi-prompt Contrastive NetworkShot=12025.06 | 32.76 | 92.78 | — | — | |
| SCTShot=12025.06 | 32.81 | 91.23 | — | — | |
| ReActBackbone=ResNet-502026.02 | 33.33 | 91.96 | — | — | |
| GradOrthBackbone=ResNet-502026.02 | 33.67 | 91.78 | — | — | |
| ReActBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 33.85 | 91.58 | — | — | |
| ReAct+DICEBackbone=ResNet-502026.02 | 34.28 | 91.71 | — | — | |
| BATSBackbone=ResNet-502026.02 | 34.34 | 91.83 | — | — | |
| VRA+Backbone=ResNet-50, Pre-trained=ImageNet2023.02 | 34.62 | 91.79 | — | — | |
| DynProtoBackbone=ViT-B/162026.04 | 35.8 | 89.04 | — | — | |
| SCALEBackbone=ResNet-502026.02 | 36.86 | 91.96 | — | — | |
| VOSID dataset=CIFAR-102022.04 | 37.61 | 90.42 | — | — | |
| VRABackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 37.85 | 91.27 | — | — | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 42022.09 | 38.26 | 88.34 | — | — | |
| VOS+Evaluation Protocol=fine-tuned2023.03 | 38.39 | 91.23 | — | — | |
| SCTShot=82025.06 | 38.77 | 92.41 | — | — | |
| GL-MCMShot=02025.06 | 38.85 | 89.9 | — | — | |
| NN-GuideBackbone=ResNet-502026.02 | 38.88 | 90.12 | — | — | |
| LoCoOpShot=12025.06 | 39.23 | 91.07 | — | — | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 3 + 42022.09 | 39.34 | 90.93 | — | — | |
| KNNEvaluation Protocol=fine-tuned2023.03 | 39.61 | 91.02 | — | — | |
| KNNShot=Full2025.06 | 39.61 | 91.02 | — | — | |
| LAPSBackbone=MobileNet-v22026.02 | 39.7 | 90.1 | — | — | |
| ASHBackbone=ResNet-502026.02 | 39.84 | 90.98 | — | — | |
| EnergyID dataset=CIFAR-102022.04 | 40.14 | 89.89 | — | — | |
| GradOrthBackbone=MobileNet-v22026.02 | 40.27 | 89.12 | — | — | |
| LoCoOpShot=82025.06 | 40.53 | 91.53 | — | — | |
| APEXBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 40.56 | 92.06 | — | — | |
| APMBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 40.71 | 92.02 | — | — | |
| CoOpShot=82025.06 | 41.17 | 89.76 | — | — | |
| VOSEvaluation Protocol=fine-tuned2023.03 | 41.62 | 90.23 | — | — | |
| RODDID Dataset=CIFAR-1002022.04 | 41.72 | 89.1 | — | — | |
| PALMBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 42.06 | 91.51 | — | — | |
| Catalyst(μ)Backbone=ResNet-502026.02 | 42.59 | 89.78 | — | — | |
| SeTARShot=02025.06 | 42.64 | 90.16 | — | — | |
| CIDERBackbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 42.81 | 91.39 | — | — | |
| kNN+Backbone=ResNet-50, Evaluation Protocol=fine-tuning, Pre-training=pre-trained, In-Distribution (ID) Dataset=ImageNet-1002026.02 | 44.41 | 90.26 | — | — | |
| MCMShot=02025.06 | 44.69 | 89.77 | — | — | |
| MCMEvaluation Protocol=zero-shot2023.03 | 44.88 | 89.83 | — | — | |
| NPOSEvaluation Protocol=fine-tuned2023.03 | 45.27 | 89.44 | — | — | |
| NPOSShot=Full2025.06 | 45.27 | 89.44 | — | — | |
| SHEBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 45.35 | 90.15 | — | — | |
| VRA+Backbone=ResNetv2-101, Pre-trained=ImageNet2023.02 | 45.83 | 90.01 | — | — | |
| ReAct+DICEBackbone=MobileNet-v22026.02 | 45.93 | 88.29 | — | — | |
| DICEBackbone=ResNet-50, Pre-trained=ImageNet2023.02 | 46.49 | 87.48 | — | — | |
| CoOpShot=12025.06 | 46.68 | 89.09 | — | — | |
| VRABackbone=ResNetv2-101, Pre-trained=ImageNet2023.02 | 47.31 | 90.19 | — | — | |
| FSID Dataset=CIFAR-1002022.04 | 47.61 | 88.42 | — | — | |
| DICEBackbone=ResNet-502026.02 | 47.62 | 87.76 | — | — | |
| Catalyst(σ)Backbone=ResNet-502026.02 | 48.35 | 88.04 | — | — | |
| Catalyst(μ) + ReActBackbone=MobileNet-v22026.02 | 48.62 | 88.59 | — | — | |
| Catalyst(m)Backbone=ResNet-502026.02 | 48.68 | 87.82 | — | — | |
| MOSBackbone=ResNetv2-101, Pre-trained=ImageNet2023.02 | 49.54 | 89.06 | — | — | |
| Catalyst(m) + ReActBackbone=MobileNet-v22026.02 | 49.77 | 88.06 | — | — | |
| Catalyst(σ) + ReActBackbone=MobileNet-v22026.02 | 51.32 | 87.19 | — | — | |
| RankFeatBackbone=ResNetv2-101, Pre-training=ImageNet-1k, Feature Layers=Block 32022.09 | 51.82 | 88.32 | — | — | |
| Catalyst(μ)Backbone=MobileNet-v22026.02 | 52.21 | 87.33 | — | — | |
| Catalyst(m)Backbone=MobileNet-v22026.02 | 52.24 | 86.89 | — | — | |
| DICEBackbone=MobileNet-v22026.02 | 52.35 | 86.17 | — | — | |
| BATSBackbone=MobileNet-v22026.02 | 52.43 | 86.26 | — | — | |
| Catalyst(σ)Backbone=MobileNet-v22026.02 | 53.04 | 86.84 | — | — |