List your company
ResearchFormerly WizWand

Out-of-Distribution Detection on ImageNet-1k V2 (test)

21.97FPR@95NNGuide
19.249637.612355.97574.3377Sep 26, 2023
Updated 8mo ago

Evaluation results

MethodLinks
NNGuideTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0921.9794.1793.44
ViMTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0928.0292.892.09
NNGuideTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0930.7891.790.08
GradNormTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0932.6691.8790.62
KLTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0932.6989.7487.87
EnergyTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0932.8689.7587.89
ViMTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0933.1191.8890.75
KNNTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0933.7491.3490.54
MaxLogitTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0934.1789.6787.97
MahalanobisTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0936.5891.7291.87
ViMTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0939.1389.3988.74
NNGuideTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0941.7390.0889.95
MahalanobisTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0942.9990.2990.76
SSDTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0943.2190.3690.09
SSDTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0944.1988.7388.65
KNNTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0944.5789.2288.64
KLTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0944.6788.9788.24
EnergyTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0944.6888.9788.24
MaxLogitTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0945.9786.3284.75
KLTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0947.0385.9184.41
EnergyTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0947.0385.9184.41
GradNormTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0947.6286.0783.75
MahalanobisTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0947.8589.4589.65
MaxLogitTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0947.9788.5488.01
MSPTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0948.9986.3885.93
MSPTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0954.6184.9384.74
MSPTraining scheme=From scratch, Model=ResNet-50, ID accuracy=74.432023.0955.5485.384.57
KNNTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0957.9886.5286.73
SSDTraining scheme=Transfer learning, Model=ViT, ID accuracy=81.162023.0958.2881.2181.39
MahalanobisTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0965.2779.3778.87
NNGuideTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0967.879.478.93
KNNTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0975.4974.3375.2
ViMTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0976.972.3272.16
SSDTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0977.2469.3267.66
MSPTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0979.476.6477.14
MaxLogitTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0980.1975.1775.06
GradNormTraining scheme=Transfer learning, Model=RegNet, ID accuracy=82.792023.0984.4761.0259.56
KLTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0989.5766.1767.92
EnergyTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0989.5766.1767.92
GradNormTraining scheme=From scratch, Model=MobileNet, ID accuracy=67.782023.0989.9864.565.27