variable_resolution3.py

#initialize our generators; specifying data directories, batch size, and dimension threshold

train_image_directory = 'imagenette2/train'
test_image_directory = 'imagenette2/val'

n_classes = 10
batch_size = 16
max_dimension = 512

#create generators for training and generating

train_generator = ImageGenerator(train_image_directory, batch_size=batch_size, shuffle=True, max_dimension=max_dimension)
test_generator = ImageGenerator(test_image_directory, batch_size=batch_size, max_dimension=max_dimension)

#convert generators into tf.data.Dataset objects for optimization with keras model fit method

train_dataset = tf.data.Dataset.from_generator(train_generator,
     (tf.float32, tf.int32),
    (tf.TensorShape([None,None,None,3]), tf.TensorShape([None])))

test_dataset = tf.data.Dataset.from_generator(test_generator,
     (tf.float32, tf.int32),
    (tf.TensorShape([None,None,None,3]), tf.TensorShape([None])))


#train and evaluate model
model.fit(train_dataset,validation_data=test_dataset,epochs=10,verbose=1,workers=2,max_queue_size=20)