Expeditious Saliency-guided Mix-up through Random Gradient Thresholding

Open paper
Benchmark Model Rank Results
image-classification-on-cifar-100R-Mix (WideResNet 28-10)#68Percentage correct: 85
image-classification-on-cifar-100RL-Mix (WideResNet 28-10)#70Percentage correct: 84.9
image-classification-on-cifar-100WideResNet 28-10 + CutMix (OneCycleLR scheduler)#78Percentage correct: 83.97
image-classification-on-cifar-100R-Mix (ResNeXt 29-4-24)#89Percentage correct: 83.02
image-classification-on-cifar-100RL-Mix (ResNeXt 29-4-24)#100Percentage correct: 82.43
image-classification-on-cifar-100R-Mix (WideResNet 16-8)#102Percentage correct: 82.32
image-classification-on-cifar-100ResNeXt 29-4-24 + CutMix (OneCycleLR scheduler)#103Percentage correct: 82.3
image-classification-on-cifar-100RL-Mix (WideResNet 16-8)#105Percentage correct: 82.16
image-classification-on-cifar-100WideResNet 16-8 + CutMix (OneCycleLR scheduler)#109Percentage correct: 81.79
image-classification-on-cifar-100R-Mix (PreActResNet-18)#113Percentage correct: 81.49
image-classification-on-cifar-100RL-Mix (PreActResNet-18)#119Percentage correct: 80.75
image-classification-on-cifar-100PreActResNet-18 + CutMix (OneCycleLR scheduler)#120Percentage correct: 80.6
image-classification-on-imagenetR-Mix (ResNet-50)#825Top 1 Accuracy: 77.39%
weakly-supervised-object-localization-on-2R-Mix (ResNet-50)#6Top-1 Localization Accuracy: 55.58

Rank counts only results with a code link.