Mastering Computer Vision with TensorFlow 2.x

Defining the train and validation generators

The train and validation generators use a method called image data generation and flow. They use this on a directory to input batches of the image as a tensor. For details of this process, please refer to the Keras documentation at: https://keras.io/preprocessing/image/.

A typical example of this is as follows. As explained in the Keras documentation, the image data generator has many parameters, but we're only using a few of them here. The preprocessing input converts the image into a tensor. The input rotation range rotates the image by 90 degrees and flips it vertically for image augmentation. We can use what we learned about image transformation in Chapter 1, Computer Vision and TensorFlow ...