Variable size of numpy array

Question

I am loading an image using skimage.io.imread and saving the image as numpy array. Following is the inferance:

Original image: (256x512x3)

Following is the code that I execute:

img = io.imread(img_file) # 48.1 kB
i1, i2 = img[:, :256], img[:, 256:]
np.save('i1', i1) # 196.7 kB
np.save('i2', i2) # 196.7 kB
final_image = np.empty([1, 2, 256, 256, 3])
final_image[0, 0], final_image[0, 1] = i1, i2
np.save('final', final_image) # 3.1 MB

Can anyone explain why such a huge difference in the size of the image?

EDIT: dtype of i1, i2, final_image is np.float64

Well, for starters, final_image is of dtype np.float_, which is likely 8 bytes per item. — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Dec 19, 2017 at 0:08
The dtype of i1 and i2 is np.float_ as well. But I don't see such a surge in the size of the saved array — harman
– harman, Commented Dec 19, 2017 at 0:10
Um, pretty sure it isn't. It should be defaulting to reading in as np.uint8, at least it does on my system. — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Dec 19, 2017 at 0:14
Can you actually print out final_mage.shape, final_image.dtype — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Dec 19, 2017 at 0:15
Yes, that's right, i1 and i2 have dtype of uint8 whereas final_image has float64 — harman
– harman, Commented Dec 19, 2017 at 0:18

juanpa.arrivillaga · Accepted Answer · 2017-12-19 00:22:52Z

2

numpy.empty will default to whatever np.float_ is on your system, however, your image should have been read in as np.uint8, so provide the corresponding dtype to empty:

final_image = np.empty([1, 2, 256, 256, 3], dtype=np.uint8)

answered Dec 19, 2017 at 0:22

juanpa.arrivillaga

97.7k14 gold badges141 silver badges190 bronze badges

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

3 Comments

harman Over a year ago

This doesn't seem to solve the problem. Even though specifying dtype to be of uint8, it changes to float64

harman Over a year ago

It's changing at least on my system

harman Over a year ago

Sorry, my bad, I was making changes to the code in a different file and executing another file. I am sorry

Collectives™ on Stack Overflow

Variable size of numpy array

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related