Why is seeding mandatory for stateless random flipping?

TensorFlow General Discussion
, ,
1

Why is seeding mandatory for stateless random flipping?

I am implementing custom data augmentation layer in my model where I am trying to use the function
tf.image.stateless_random_flip_left_right.

Documentation says this function can,

Randomly flip an image horizontally (left to right) deterministically.

Why do we need deterministic flip? Also, for this function seed is a mandatory argument.

In data augmentation, I understand that from epoch to epoch, a particular image is fed in different forms. Wouldn’t this force images of the same kind to be supplied?

3

Stateless means results won’t have any state. So it’s just going to randomize the images and give different results in terms of the different seeds. For image augmentation you can use the random_flip_left_right. Thank you.