RandAugment

The Pytorch Image Model framework provides another useful augmentation class called RandAugment. As the name implies, it performs random augmentations on the image datasets.

The RandAugment class

Here’s a list of image transformations in the RandAugment class:

Transformation	Description
Invert	Invert the colors of the input image.
AutoContrast	Maximize the contrast of the input image by remapping its pixels per channel.
Equalize	Equalize the histogram of the input image.
Rotate	Rotate the input image by a certain angle.
Solarize	Solarize the input image by inverting all pixel values above a threshold.
Color	Colorize the input greyscale image.
Posterize	Reduce the number of bits for each color channel of the input image.
Contrast	Adjust the contrast of the input image.
Brightness	Adjust the brightness of the input image.
Sharpness	Adjust the sharpness of the input image.
ShearX	Slant the input image in the x-direction.
ShearY	Slant the input image in the y-direction.
TranslateX	Slide the input image in the x-direction.
TranslateY	Slide the input image in the y-direction.