List your company
ResearchFormerly WizWand

Image Classification on ImageNet LT 2018 (test)

56.6Top-1 AccACE (3 experts)
19.47229.11138.7548.389Aug 5, 2021
Updated 8mo ago

Evaluation results

MethodLinks
ACE (3 experts)Backbone=ResNeXt-502021.0856.6
RIDE (3 experts)Backbone=ResNeXt-50, Number of experts=32021.0856.4
RIDE (2 experts)Backbone=ResNeXt-50, Number of experts=22021.0855.9
RIDE (3 experts)Backbone=ResNet-50, Number of experts=32021.0854.9
ACE (3 experts)Backbone=ResNet-502021.0854.7
RIDE (2 experts)Backbone=ResNet-50, Number of experts=22021.0854.4
Logit Adj.Backbone=ResNet-502021.0851.1
LWSBackbone=ResNeXt-502021.0849.9
cRTBackbone=ResNeXt-502021.0849.5
τ-normBackbone=ResNeXt-502021.0849.4
BBNBackbone=ResNeXt-502021.0849.3
BBNBackbone=ResNet-502021.0848.3
LWSBackbone=ResNet-502021.0847.7
cRTBackbone=ResNet-502021.0847.3
NCMBackbone=ResNeXt-502021.0847.3
τ-normBackbone=ResNet-502021.0846.7
OLTRBackbone=ResNeXt-502021.0846.3
BaselineBackbone=ResNeXt-502021.0844.4
NCMBackbone=ResNet-502021.0844.3
ACE (3 experts)Backbone=ResNet-102021.0844
CAMBackbone=ResNet-102021.0843.1
cRTBackbone=ResNet-102021.0841.8
BaselineBackbone=ResNet-502021.0841.6
LWSBackbone=ResNet-102021.0841.4
τ-normBackbone=ResNet-102021.0840.6
LFMEBackbone=ResNet-102021.0838.8
LDAM+DRWBackbone=ResNet-102021.0836
NCMBackbone=ResNet-102021.0835.5
OLTRBackbone=ResNet-102021.0834.1
Lifted LossBackbone=ResNet-102021.0830.8
Range LossBackbone=ResNet-102021.0830.7
Focal lossBackbone=ResNet-102021.0830.5
FSLwFBackbone=ResNet-102021.0828.4
BaselineBackbone=ResNet-102021.0820.9