List your company
ResearchFormerly WizWand

Image Classification on Clothing1M 1.0 (test)

81.16AccuracyDeepSelf
58.425664.327870.2376.1322Aug 16, 2019Nov 27, 2019Mar 10, 2020Jun 21, 2020Oct 3, 2020Jan 14, 2021Apr 28, 2021
Updated 8mo ago

Evaluation results

MethodLinks
DeepSelfTraining=clean50k, Pre-training=Initialized from #102020.0381.16
loss cor.Training=clean50k, Pre-training=Initialized from #42020.0380.38
CleanNetTraining=clean50k, Pre-training=Initialized from #82020.0379.9
NoiseRankTraining=clean50k, Pre-training=Initialized from #132020.0379.57
NoneTraining=clean50k, Pre-training=noisy train # 22020.0379.43
NoneTraining=clean50k, Pre-training=ImageNet2020.0375.19
ELR+Pre-training=ImageNet2021.0374.81
DivideMixBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=true2021.0474.8
DivideMixPre-training=ImageNet2021.0374.76
CleanNetTraining=noisy(+verified), Pre-training=ImageNet2020.0374.69
DeepSelfTraining=noisy train, Pre-training=ImageNet2020.0374.45
Self-learningPre-training=ImageNet2021.0374.45
C2DPre-training=None, Initialization=SimCLR, LNL method=DivideMix2021.0374.3
PLCBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=true2021.0474
NoiseRankTraining=cleaned train, Pre-training=noisy train #22020.0373.82
NoiseRankTraining=cleaned train, Pre-training=ImageNet2020.0373.77
C2DPre-training=None, Initialization=BT, LNL method=ELR+2021.0373.52
PENCILBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=true2021.0473.5
MLNTBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0473.5
PENCILTraining=noisy train, Pre-training=ImageNet2020.0373.49
P-correctionPre-training=ImageNet2021.0373.49
Meta-LearningPre-training=ImageNet2021.0373.47
Cross-entropyPre-training=None, Initialization=BT2021.0373.03
DropoutBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=true2021.0472.8
MetaCleanerPre-training=ImageNet2021.0372.5
JOBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0472.2
Joint opt.Training=noisy train, Pre-training=ImageNet2020.0372.16
Joint-OptimPre-training=ImageNet2021.0372.16
Cross-entropyPre-training=None, Initialization=SimCLR2021.0372.05
SLBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0871.02
JoCoRBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0470.3
ForwardBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0869.84
loss cor.Training=noisy(+verified), Pre-training=ImageNet2020.0369.84
F-correctionPre-training=ImageNet2021.0369.84
GCEBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0869.75
D2LBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0869.47
Cross-entropyPre-training=ImageNet2021.0369.21
Co-teachingBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0469.2
NoneTraining=noisy train, Pre-training=ImageNet2020.0368.94
BootstrapBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0868.91
F-correctionBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0468.9
CEBackbone=ResNet-50, Pre-trained=ImageNet, Batch Size=64, Optimizer=SGD, Initial Learning Rate=0.0012019.0868.8
DecouplingBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0468.5
CEBackbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0467.2
Co-teaching+Backbone=ResNet-18, Pre-trained=ImageNet, Balanced subset or loss=false2021.0459.3