List your company
ResearchFormerly WizWand

Image Classification on ImageNet and ImageNet-V2

81.9ImageNet AccuracyAugReg-ViT + DAT + DAD
72.43674.89377.3579.807Nov 2, 2023
Updated 8mo ago

Evaluation results

MethodLinks
AugReg-ViT + DAT + DADBackbone=ViT-B/16, Resolution=224x2242023.1181.971.776.8
AugReg-ViT + Pyramid ATBackbone=ViT-B/16, Resolution=224x2242023.1181.770.376
AugReg-ViT + DATBackbone=ViT-B/16, Resolution=224x2242023.1181.570.876.2
CLIPBackbone=ViT-L/14, Resolution=224x2242023.1179.972.976.4
AugReg-ViTBackbone=ViT-B/16, Resolution=224x2242023.1179.967.973.9
AugReg-ViT + DADBackbone=ViT-B/16, Resolution=224x2242023.1179.669.974.8
AdvpropBackbone=ViT-B/16, Resolution=224x2242023.1179.568.774.1
DebiasedBackbone=ViT-B/16, Resolution=224x2242023.1179.367.673.5
Fast AdvpropBackbone=ViT-B/16, Resolution=224x2242023.11796773
DAT + DADBackbone=ResNet50, Resolution=224x2242023.1177.86671.9
AdvpropBackbone=ResNet50, Resolution=224x2242023.1177.665.571.6
DebiasedBackbone=ResNet50, Resolution=224x2242023.1176.96571
DATBackbone=ResNet50, Resolution=224x2242023.1176.56570.8
ResNet50Backbone=ResNet50, Resolution=224x2242023.1176.163.269.7
DADBackbone=ResNet50, Resolution=224x2242023.1175.76570.4
Pyramid ATBackbone=ResNet50, Resolution=224x2242023.1175.562.569
ViTBackbone=ViT-B/16, Resolution=224x2242023.1172.858.765.8