List of 3D Numpy Arrays to Array Conversion

Question

I'm using Image from PIL, to open images and load them into numpy arrays of dimension(300,300,3). These arrays are appended to a list, I then want to convert this list into a numpy array. Everything that should work, hasn't. I keep getting the weird error:

ValueError: could not broadcast input array from shape (300, 300, 3) into shape (300, 300)

Here is a small example code:

before loop

training_data = []

in loop

image = Image.open(path).resize((300,300),Image.ANTIALIAS)

training_data.append(np.asarray(image))

outside loop

training_data = np.array(training_data)

the simple Issue is that I get the error talked about above. Any help is much appreciated.

hpaulj · Accepted Answer · 2020-07-18 07:00:11Z

1

Most likely you have collected a list of arrays of different sizes, perhaps some are b/w and others are color:

In [17]: alist = []                                                                                  
In [18]: alist.append(np.ones((300,300)))        # bw                                                            
In [19]: alist.append(np.ones((300,300,3)))      # color                                               
In [20]: np.array(alist)                                                                             
/usr/local/bin/ipython3:1: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  #!/usr/bin/python3
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-20-7512d762195a> in <module>
----> 1 np.array(alist)

ValueError: could not broadcast input array from shape (300,300,3) into shape (300,300)

v1.19 gives us a warning when we try to make a array from arrays that differ in shape. Sometimes that still gives us an object dtype array, but with this combination of shapes, the result is your error.

===

An equivalent way of combining the arrays into one is with np.stack. If it works the result is the same; if not, the error is different:

In [21]: np.stack(alist)                                                                             
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-21-724d9c1d0554> in <module>
----> 1 np.stack(alist)

<__array_function__ internals> in stack(*args, **kwargs)

/usr/local/lib/python3.6/dist-packages/numpy/core/shape_base.py in stack(arrays, axis, out)
    425     shapes = {arr.shape for arr in arrays}
    426     if len(shapes) != 1:
--> 427         raise ValueError('all input arrays must have the same shape')
    428 
    429     result_ndim = arrays[0].ndim + 1

ValueError: all input arrays must have the same shape

edited Jul 18, 2020 at 7:00

answered Jul 18, 2020 at 5:59

hpaulj

233k14 gold badges260 silver badges392 bronze badges

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

2 Comments

Sharp Dev Over a year ago

Reading your answer, I decided to pull all the shape data into a file to look over the shapes, and I have found 100% consistent shapes between all elements of the list.

hpaulj Over a year ago

You could try np.stack(training_data).

Collectives™ on Stack Overflow

List of 3D Numpy Arrays to Array Conversion

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related