Reshaping Python list into numpy array for Keras CNN

Question

I have 200000 colored images each of dimensions (50x50x3) stored as a list of list in Python. Because its all in a list, I want to change it to numpy array of dimension (200000,50,50,3). However, using np.toarray(list) returns an array of shape (200000,)

But if I shorten the list (say take only 10000 elements) and then do np.toarray(shortened_list), it returns a correct array of shape (10000,50,50,3)

I want to do this to feed the images to a CNN using Keras.

Daniel Möller · Accepted Answer · 2018-12-05 17:41:02Z

1

One or more images doesn't have the correct shape, or is not an image, thus it's impossible to put them into an array of numbers.

Numpy is then creating an array of objects. Find the wrong image and remove it from the list or resize it correctly.

Hint:

for i, img in enumerate(shortened_list):

    if not hasattr(img, 'shape'):
        print("wrong image at index " + str(i))
    elif img.shape != (50,50,3):
        print("wrong image at index " + str(i))

edited Dec 5, 2018 at 17:41

answered Dec 5, 2018 at 17:32

Daniel Möller

86.8k24 gold badges202 silver badges222 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Bikram Baruah Over a year ago

This helped a lot. There was quiet a few images with different shapes in the data set

Collectives™ on Stack Overflow

Reshaping Python list into numpy array for Keras CNN

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related