How are images resized in ImageDataGenerator

Question

In my understanding, when using tensorflow keras, we can flow data from directories using ImageDataGenerator. When the data is flowed in, they are resized to target_size. But how is this transformation done?

An example is this:

from tensorflow.keras.preprocessing.image import ImageDataGenerator
trainDir = 'train'
dataGen = ImageDataGenerator(rescale=1/255)
gen = dataGen.flow_from_directories(
  trainDir,
  batch_size=64,
  target_size=(150, 150),
  class_mode='binary'
)

Here, all the images from trainDir would be resized to 150x150. But how are they resized?

I've checked the documentation on tensorflow ImageDataGenerator. It only says that target_size is "The dimensions to which all images found will be resized." But I didn't find an explanation on how the resizing is done.

Answer 1

It uses the interpolation parameter when the target_size is different from the size of input image. The default interpolation method used is nearest but others are available to use as well.

Source: https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image/ImageDataGenerator