List your company
ResearchFormerly WizWand

Image Classification on ILSVRC (val)

15.47Top-1 ErrorGoogLeNet-SPG
14.444821.364928.28535.2051Sep 4, 2014Apr 30, 2015Dec 24, 2015Aug 18, 2016Apr 13, 2017Dec 7, 2017Aug 3, 2018
Updated 8mo ago

Evaluation results

MethodLinks
GoogLeNet-SPGClassification Backbone=DPN-ensemble2018.0715.472.7
ACoLLocalization Architecture=VGGnet, Classification Architecture=DPN-ensemble, ensemble=true2018.0415.472.7
ACoLLocalization Architecture=VGGnet, Classification Architecture=DPN-ensemble, ensemble=true, Top-3 selection strategy=true2018.0415.472.7
GoogLeNet-SPGClassification Backbone=DPN-1312018.0717.083.42
ACoLLocalization Architecture=VGGnet, Classification Architecture=DPN-1312018.0417.083.42
GoogLeNet-SPGClassification Backbone=DPN-982018.0717.423.67
ACoLLocalization Architecture=VGGnet, Classification Architecture=DPN-982018.0417.423.67
GoogLeNet-SPGClassification Backbone=DPN-922018.0717.73.83
ACoLLocalization Architecture=VGGnet, Classification Architecture=DPN-922018.0417.73.83
VGGNumber of networks (ensemble size)=2 nets, Evaluation protocol (multi-crop/dense evaluation)=multi-crop & dense eval.2014.0923.76.8
GoogLeNet-SPGClassification Backbone=ResNet-1522018.0724.397.59
ACoLLocalization Architecture=VGGnet, Classification Architecture=ResNet-1522018.0424.397.59
VGGNumber of networks (ensemble size)=1 net, Evaluation protocol (multi-crop/dense evaluation)=multi-crop & dense eval.2014.0924.47.1
VGGNumber of networks (ensemble size)=7 nets, Evaluation protocol (multi-crop/dense evaluation)=dense eval. (ILSVRC submission)2014.0924.77.5
GoogLeNet-SPGClassification Backbone=ResNet-1012018.0724.97.8
ACoLLocalization Architecture=VGGnet, Classification Architecture=ResNet-1012018.0424.97.8
GoogLeNet-SPGClassification Backbone=ResNet-502018.0726.228.47
ACoLLocalization Architecture=VGGnet, Classification Architecture=ResNet-502018.0426.228.47
MSRANumber of networks (ensemble size)=1 net2014.0927.99.1
OverFeatNumber of networks (ensemble size)=7 nets2014.093413.2
CurriculumNetTraining Data=WebVision training set2018.0835.215.1
OverFeatNumber of networks (ensemble size)=1 net2014.0935.714.2
Zeiler & FergusNumber of networks (ensemble size)=6 nets2014.093614.7
CleanNetTraining Data=WebVision training set2018.0836.615.4
Our BaselineTraining Data=WebVision training set2018.0837.116.4
Zeiler & FergusNumber of networks (ensemble size)=1 net2014.0937.516
MentorNetTraining Data=WebVision training set2018.0837.517
Krizhevsky et al.Number of networks (ensemble size)=5 nets2014.0938.116.4
Krizhevsky et al.Number of networks (ensemble size)=1 net2014.0940.718.2
BaselineTraining Data=WebVision training set2018.0841.120.2