Mastering Computer Vision with TensorFlow 2.x

Training data generator

Let's have a look at the following code for the training data generator:

from keras.preprocessing.image import ImageDataGeneratortrain_dir = '/home/…/visual_search/furniture_images/train'train_datagen =  ImageDataGenerator(preprocessing_function=preprocess_input,rotation_range=90,horizontal_flip=True,vertical_flip=True)

Flow for directory API—this is used to import data from directories. It has the following parameters:

• Directory: This is a folder path, and should be set to the path where images for all three classes are present. An example, in this case, would be a path for the train directory. To get the path, you can drag the folder to the Terminal and it will show you the path, which you can then copy and paste. ...