Image Data Augmentation

Image Data Augmentation refers to a class of methods that augment an image dataset to increase the effective size of the training set, or as a form of regularization to help the network learn more effective representations. Below you can find a continuously updating list of image data augmentation methods.

METHOD YEAR PAPERS
Mixup
2017 99
Random Resized Crop
2000 77
Random Horizontal Flip
2000 77
ColorJitter
2000 35
AutoAugment
2018 35
Cutout
2017 27
CutMix
2019 22
Random Gaussian Blur
2000 20
RandAugment
2019 9
Random Scaling
2000 7
Fast AutoAugment
2019 4
Population Based Augmentation
2019 3
AugMix
2019 3
Image Scale Augmentation
2000 3
Random Grayscale
2000 1
RandomRotate
2000 1
InstaBoost
2019 1
DiffAugment
2020 1
Batchboost
2020 1
MaxUp
2020 1