CODEWITHSUNDEEP
pythontensorflowkerasconv-neural-networkpreprocessor
BlueBen
BlueBen

Reputation: 11

how to know how many images are augmented in tensorflow.keras.layers.experimental.preprocessing?

I'm studying CNN to image classification and using data augmentation for prohibiting overfitting. Data augmentation works which the loss function go decrease but, I can't find out how many training images are augmented. Many research papers talk about the augmentation factor(I understood 'augmentation factor' means how many times the data has been multiplied. If it is fault, plz tell me what is the exact meaning of it. One of the references is below.). I also want to know about it.

※ ref. (A. Krizhevsky, I. Sutskever, G. E. Hinton, ImageNet classification with deep convolutional neural networks, in Advances in Neural Information Processing Systems 25 (NIPS 2012) (NIPS, 2012), pp.1097–1105.)

And, this is the part of my code. How many images are augmented?? I can't find it out from tensorflow and any site(Maybe it's because of my lack of search skills...) help me!!!

data_augmentation = keras.Sequential(
  [
    layers.experimental.preprocessing.RandomFlip("horizontal_and_vertical", 
                                                  input_shape=(32, 
                                                              32,
                                                              1)),
    layers.experimental.preprocessing.RandomRotation(0.2),
    layers.experimental.preprocessing.RandomZoom(0.2),
    layers.experimental.preprocessing.RandomTranslation(height_factor=0.1, width_factor=0.1)  
  ])

thank you!

Upvotes: 1

Views: 831

Answers (1)

Whereismywall
Whereismywall

Reputation: 197

You need to mention the batch size, which is the number of images selected each epoch. You can read the link https://www.tensorflow.org/tutorials/images/data_augmentation . It provides some insight. They use batch size 32.

Upvotes: 2

Related Questions