`data_augmenter()` correct but does not change image to the naked eye

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi everyone!

It looks to me that the original image is identical – at least to the naked eye – to the augmented image. Here’s what I mean:

  1. my tests pass
  2. I run this and the plots returned are identical.
for image, ii in train_dataset.take(1):
    
    first_image = image[0]
    plt.figure(figsize=(3,3))
    plt.imshow(first_image/255)
    
    """Take just one transformation of that original image"""
    plt.figure(figsize=(3,3))
    augmented_image = data_augmentation(
        tf.expand_dims(first_image, 0)
    )
    plt.imshow(augmented_image[0] / 255)
2

Hello,
Here you are just expanding the dimensions and that will change the shape of your image dimensions that is you get one more dimension
To explain a×b were the original dimensions and now they are a×b×c where c is the new dimension
In doing so,the image is unaffected and hence you dont see any change to naked eye
Hope it helps
Thanks and Regards,
Mayank Ghogale

3

Nevermind! I just restarted the kernel and ran everything and it fixed the problem. Thanks for your help @MayankGhogale

4

Glad it helped sir
Regards,
Mayank Ghogale