List your company
ResearchFormerly WizWand

Long-tailed Image Recognition on CIFAR-10 long-tailed (test)

88.1Accuracy (New Classes)BaCon-S
17.48435.81754.1572.483Oct 2, 2023Mar 9, 2024Aug 15, 2024Jan 21, 2025Jun 29, 2025Dec 5, 2025May 13, 2026
Updated 5mo ago

Evaluation results

MethodLinks
BaCon-SBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block, pseudo-labeling branch=SimGCD-style2023.1088.194.291.1——
BaCon-OBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block, pseudo-labeling branch=ORCA-style2023.107883.380.7——
RobustLTBase algorithm=AT-BSL, Backbone=WRN-28-102026.0573.83—77.6142.1135.98
CFABase algorithm=AT-BSL, Backbone=WRN-28-102026.0573.67—77.9138.8930.27
UDRBase algorithm=AT-BSL, Backbone=WRN-28-102026.0573.4—77.6638.429.78
DAFABase algorithm=AT-BSL, Backbone=WRN-28-102026.0573.23—77.3138.1130.3
AT-BSLBase algorithm=AT-BSL, Backbone=WRN-28-102026.0572.48—77.0937.9828.6
GCDBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1071.778.575.1——
RobustLTBase algorithm=RoBal, Backbone=WRN-28-102026.0570.32—74.6336.0829.19
CFABase algorithm=RoBal, Backbone=WRN-28-102026.0567.42—72.8133.3523.52
RoBalBase algorithm=RoBal, Backbone=WRN-28-102026.0567.15—72.7332.2922.19
TRSSLBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block, Adapted=true2023.1066.878.472.6——
DAFABase algorithm=RoBal, Backbone=WRN-28-102026.0566.78—72.2532.8723.22
RobustLTBase algorithm=REAT, Backbone=WRN-28-102026.0566.58—71.9833.0124.21
RobustLTBase algorithm=TAET, Backbone=WRN-28-102026.0564.36—68.0534.0529.33
CFABase algorithm=TAET, Backbone=WRN-28-102026.0564.25—68.8332.725.75
DAFABase algorithm=TAET, Backbone=WRN-28-102026.0562.66—67.0932.1225.91
UDRBase algorithm=TAET, Backbone=WRN-28-102026.0562.57—67.1232.2225.44
CFABase algorithm=REAT, Backbone=WRN-28-102026.0562.2—68.6830.6619.71
DAFABase algorithm=REAT, Backbone=WRN-28-102026.0561.45—67.8429.3219.01
UDRBase algorithm=REAT, Backbone=WRN-28-102026.0561.38—68.2229.2817.6
REATBase algorithm=REAT, Backbone=WRN-28-102026.0560.81—67.7828.6717.12
TAETBase algorithm=TAET, Backbone=WRN-28-102026.0560.44—65.3231.6824.67
RobustLTBase algorithm=AWP, Backbone=WRN-28-102026.0557.05—65.2229.1116.46
ORCABackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1055.677.566.6——
UDRBase algorithm=RoBal, Backbone=WRN-28-102026.0553.6—62.3831.1118.28
DAFABase algorithm=AT, Backbone=WRN-28-102026.0553.2—61.8227.4914.9
RobustLTBase algorithm=AT, Backbone=WRN-28-102026.0552.67—61.5928.9716.36
DAFABase algorithm=AWP, Backbone=WRN-28-102026.0551.43—60.6328.7815.37
AWPBase algorithm=AWP, Backbone=WRN-28-102026.0550.17—59.6628.514.9
UDRBase algorithm=AWP, Backbone=WRN-28-102026.0550.1—59.6528.5914.92
ATBase algorithm=AT, Backbone=WRN-28-102026.0548.56—58.2527.2813.71
UDRBase algorithm=AT, Backbone=WRN-28-102026.0548.42—58.1427.2813.72
OpenConBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1047.287.267.2——
CFABase algorithm=AT, Backbone=WRN-28-102026.0546.72—56.828.514.7
CFABase algorithm=AWP, Backbone=WRN-28-102026.0546.71—56.929.3115.45
SimGCDBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1041.575.158.3——
DARPBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1035.277.656.4——
ABCBackbone=ViT-B/16, Pre-trained=DINO, Fine-tuning protocol=fine-tune last block2023.1020.277.748.9——