Image data augmentation is a method used by scientists to increase the size of a dataset artificially by creating modified versions of images in the dataset. Only training dataset images can be used for augmentation. If deep learning neural network models are trained on more data then the Artificial Intelligence models become more skillful and robust, and the augmentation techniques might create variations of the training dataset images which may develop the ability of the trained models to generalize better by what they have learned from new augmented data.
The Keras deep learning neural network library is a famous library which provides methods to apply image data augmentation on images. ‘ImageDataGenerator’ class. In this example, we will use image data augmentation when training deep learning neural networks. We will do Horizontal and Vertical Shift, Horizontal and Vertical Flip, Random Rotation, Random Brightness, Random Zoom.
