List your company
ResearchFormerly WizWand

Long-tailed Image Classification on CIFAR-100 LT (Imbalance ratio 100)

53.41Top-1 AccPrototype Classifier
37.716441.790745.86549.9393Jul 26, 2021Jan 22, 2022Jul 21, 2022Jan 17, 2023Jul 16, 2023Jan 12, 2024Jul 11, 2024
Updated 8mo ago

Evaluation results

MethodLinks
Prototype Classifier2023.0253.41
WD + WD + Maxweight regularization=WD + WD + Max2023.0253.35
APA*+AGLU2024.0752.3
PaCOself-supervised pre-training=true, aggressive data augmentation=true2023.0252
PaCoTraining setup=RandAugment [12] in 400 epochs2021.0752
APA*2024.0751.9
Baseline with SE2024.0750.9
BALMS2024.0750.8
CC-SAM2024.0750.8
Balanced SoftmaxTraining setup=RandAugment [12] in 400 epochs2021.0750.8
RIDE+CMO+CR2024.0750.7
RIDE+CMO2024.0750
TLCnumber of experts=42024.0749.8
ResLTnumber of experts=32024.0749.7
ACEnumber of experts=42024.0749.6
RIDEnumber of experts=42024.0749.4
RIDEensemble=true2023.0249.1
AREA2024.0748.9
GCLBackbone Net=ResNet-322023.0548.71
NCM2023.0248.67
MisLASBackbone Net=ResNet-322023.0547.5
DRO-LT2023.0247.31
MisLas2024.0747
MISLAS2021.0747
Contrastive learningBackbone Net=ResNet-322023.0546.72
HSC2024.0746.7
WDweight decay=tuned baseline2023.0246.08
SSD2024.0746
DIVE2024.0745.4
LADE2024.0745.4
ResLT2021.0745.3
CE loss + mixup + cRTBackbone Net=ResNet-322023.0545.12
De-confound-TDEBackbone Net=ResNet-322023.0544.15
Causal Norm2021.0744.1
TSC2024.0743.8
BBN2021.0742.6
BBNBackbone Net=ResNet-322023.0542.56
LDAM-DRWBackbone Net=ResNet-322023.0542.04
LogitAdjust2023.0242.01
LDAM+DRW2021.0742
cRT2023.0241.24
CE loss + mixupBackbone Net=ResNet-322023.0540.01
LDAM Loss2023.0239.6
CE lossBackbone Net=ResNet-322023.0538.43
Focal Loss2023.0238.41
Focal Loss2021.0738.4
CE2023.0238.32