Numpy can't convert list to np array

Question

I have a python list than includes 90900 nparrays shaped (299, 299, 3). I tried to convert this list to numpy array

X_trains = np.asarray(X_train).reshape((len(X_train),299,299,3))

However, this gave me the error:

ValueError: could not broadcast input array from shape (299,299,3) into shape (299,299)

I suppose that the part of the code that cause the error is np.asarray, is there any way to fix that?

Full error code:

ValueErrorTraceback (most recent call last)
<ipython-input-34-2ba5db77f6b1> in <module>()
  1 
  2 
----> 3 X_trains = np.asarray(X_train)

/usr/local/lib/python3.4/dist-packages/numpy/core/numeric.py in asarray(a,          dtype, order)
529 
530     """
--> 531     return array(a, dtype, copy=False, order=order)
532 
533 

ValueError: could not broadcast input array from shape (299,299,3) into shape (299,299)

I want to shape my data as a 4 dimensional array (It's image files) (number of images, height, length, color channels) — Lau
– Lau, Commented Apr 8, 2017 at 13:18
@kmario23 You were right! The problem was that I didn't realize that there were some grayscale images, which messed things up. — Lau
– Lau, Commented Apr 8, 2017 at 13:28

Kasravnd · Accepted Answer · 2017-04-08 13:05:26Z

1

No, the problem is because of the reshape(). The asarray() function will convert your list to a proper Numpy array, you don't need to reshape it.

Here is an example:

In [1]: a = [[1, 2], [4, 5]]

In [2]: import numpy as np

In [3]: np.asarray(a)
Out[3]: 
array([[1, 2],
       [4, 5]])

If you want to reshape the array still after converting to a Numpy array, the new shape should be proper enough for broadcasting the older array to the new one.

You can get the shape and see if it's convertible to your expected one:

X_trains = np.asarray(X_train)
old_shape = X_trains.shape

answered Apr 8, 2017 at 13:05

Kasravnd

108k19 gold badges167 silver badges195 bronze badges

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

2 Comments

Lau Over a year ago

nope the problem is clearly in np.asarray, I included the full error code, hope it would make more sense

Kasravnd Over a year ago

@Wideem This answer is because of the fact that your list contain similar shaped lists. But if the error is because of the asarray this means that you have a mismatch in your sub lists.

kmario23 · Accepted Answer · 2017-04-08 13:23:41Z

0

I understand your problem. There is no need to reshape it. Just np.asarray would do the job

X_trains = np.asarray(X_train)

Example:

In [18]: arr_3d
Out[18]: 
array([[[ 0,  1,  2],
        [ 3,  4,  5],
        [ 6,  7,  8]],

       [[ 9, 10, 11],
        [12, 13, 14],
        [15, 16, 17]]])


In [19]: list_of_arr = [arr_3d, arr_3d]

In [20]: arr_4d = np.asarray(list_of_arr)

In [21]: arr_4d.shape
Out[21]: (2, 2, 3, 3)

answered Apr 8, 2017 at 13:23

kmario23

62.1k17 gold badges174 silver badges159 bronze badges

Collectives™ on Stack Overflow

Numpy can't convert list to np array

2 Answers 2

2 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

2 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related