MixPro: Data Augmentation with MaskMix and Progressive Attention Labeling for Vision Transformer

Open paper
Benchmark Model Rank Results
data-augmentation-on-imagenetDeiT-B (+MixPro)#1Accuracy (%): 82.9
data-augmentation-on-imagenetDeiT-S (+MixPro)#3Accuracy (%): 81.3
data-augmentation-on-imagenetDeiT-T (+MixPro)#17Accuracy (%): 73.8
image-classification-on-imagenetXCiT-M (+MixPro)#347Top 1 Accuracy: 84.1%
image-classification-on-imagenetCA-Swin-S (+MixPro)#388Top 1 Accuracy: 83.7%
image-classification-on-imagenetDeiT-B (+MixPro)#474Top 1 Accuracy: 82.9%
image-classification-on-imagenetCA-Swin-T (+MixPro)#486Top 1 Accuracy: 82.8%
image-classification-on-imagenetPVT-M (+MixPro)#499Top 1 Accuracy: 82.7%
image-classification-on-imagenetPVT-S (+MixPro)#630Top 1 Accuracy: 81.2%
image-classification-on-imagenetCaiT-XXS (+MixPro)#662Top 1 Accuracy: 80.6%
image-classification-on-imagenetPVT-T (+MixPro)#850Top 1 Accuracy: 76.7%
image-classification-on-imagenetDeiT-T (+MixPro)#921Top 1 Accuracy: 73.8%

Rank counts only results with a code link.