List your company
ResearchFormerly WizWand

Out-of-Distribution Detection on ImageNet (ID) vs SSB-hard OOD (test)

37.51FPR@95%TPRDynProto
35.316850.120964.92579.7291Apr 26, 2026Apr 29, 2026May 2, 2026May 5, 2026May 8, 2026May 11, 2026May 14, 2026
Updated 5mo ago

Evaluation results

MethodLinks
DynProtoExternal Text=false, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0437.5186.78
DynProtoExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0449.586.25
FAExternal Text=false, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0470.8378.21
AdaNegExternal Text=true, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0474.9175.11
LINeExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0475.7877.14
MahaVarBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0575.8773.02
Mahalanobis++Backbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0578.4571.5
CADRefExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0478.7974.58
NegRefineExternal Text=true, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0478.7974.59
CSPExternal Text=true, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0478.8873.61
ReActExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0479.0772.81
OODDExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0480.3565.9
AdaNDExternal Text=false, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0481.2266.64
KL-matchingBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0582.1669.38
ASH-SExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0482.5370.49
MSPBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0583.1570.83
GENBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0583.4872.02
MaxLogitBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0583.667.47
EnergyExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0483.8772.35
ViMBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0583.9269.63
NNGuideBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0584.1171.67
MSPExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0484.5472.16
ReActBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0584.5668.29
NCI+filterBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0584.9669.56
MahalanobisBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0585.0364.46
LoCoOpExternal Text=false, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0485.6168.22
NegLabelExternal Text=true, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0485.6869.39
OptFSExternal Text=false, Method Category=Vision Model based Methods, Backbone=ResNet-502026.0485.7769.73
SynOODExternal Text=true, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0485.9469.58
EnergyBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0585.9963.59
fDBDBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0586.3268.64
SCALEBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0586.5363.33
NCIBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0586.9167.75
KNNBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0588.7166.29
MCMExternal Text=false, Method Category=Vision Language Model based Methods, Backbone=CLIP-B/162026.0489.962.95
ASH-SBackbone architectures=Average of ResNet-50, Swin-B, and ViT-B2026.0592.3448.18