Data augmentation is the technique of increasing the size of data used for training a model. Some of position augmentation includes scaling, cropping, flipping, padding, rotation, translation, affine transformation. Color augmentation includes brightness, contrast, saturation and hue.