Image Data Generator

[pablowilks]

I have a question about image data generator that has bugged me for a while now :slight_smile.

Take the following code:

train_datagen = ImageDataGenerator(rescale=1./255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode=‘nearest’)

In train_datagen above, there are several augmentation techniques applied (horizontal_flip, shear_range etc) for augmenting images.
My question is, how many techniques get applied at a time to each image?
is it all of them?
just one of them?
or a random amount?

[gent.spah]

Hi there,

I briefly went through the source code in github and it seems to me that the transformation goes through every image: Imagedatagenerator check apply_transform line 831.

[ashish23]

All of them gets applied to the image