[FIXED] new numpy np.array assigned wrong data from old np.array, expected that the values will be the same

Issue

This is my first question here and it is about using numpy library in Python. I have a function which get three dimensional numpy array (np.array) with some x,y coordinates. The logic of that function - reordering this array from smaller to bigger coordinate

import numpy as np
def reorder(points):
    # reshaping array
    points = points.reshape((4, 2))
    # creating empty output array
    points_new = np.zeros((4, 1, 2), np.uint8)
    # summing the values
    add = points.sum(1)
    # find difference
    diff = np.diff(points, axis=1)
    # with the smaller sum will be first, the maximum will be last
    points_new[0] = points[np.argmin(add)]
    points_new[3] = points[np.argmax(add)]
    # the smaller difference will be the second, the max difference - the third
    points_new[1] = points[np.argmin(diff)]
    points_new[2] = points[np.argmax(diff)]
    return points_new

But after using this function

input_data = np.array([[[ 573,  148]], [[  25,  223]], [[ 153, 1023]], [[ 730,  863]]])
output_data = reorder(data)

I got the strange result

np.array([[[ 25, 223]], [[ 61, 148]], [[153, 255]], [[218,  95]]])

As you can see, in new array outputdata the data was changed, but should be contained the same values as inputdata

I rewrite this function by using simple list

def reorder_by_lst(points):
    # reshaping array
    points = points.reshape((4, 2))
    # summing the values
    add = points.sum(1)
    # find difference
    diff = np.diff(points, axis=1)
    # with the smaller sum will be first, the maximum will be last
    a = points[np.argmin(add)]
    d = points[np.argmax(add)]
    # the smaller difference will be the second, the max difference - the third
    b = points[np.argmin(diff)]
    c = points[np.argmax(diff)]
    lst = [a, b, c, d]
    return np.array(lst)

And after using it

input_data = np.array([[[ 573,  148]], [[  25,  223]], [[ 153, 1023]], [[ 730,  863]]])
output_data = reorder_by_lst(data)

I got this result

np.array([[  25,  223], [ 730,  863], [ 573,  148], [ 153, 1023]])

Now the results are the same, only dimensional should be fixed. It seems like a bug, but maybe it is deep feature which I dont know. I need a comment from professional community.

P.S. Python version - 3.10, Numpy version - 1.23.5

Solution

You set your points_new array to be of type np.unint8, which is an unsigned 8-bit integer. The range of values is therefore 0 to 255, which can be checked as follows:

>>> np.iinfo(np.uint8)
iinfo(min=0, max=255, dtype=uint8)

Your points array has values that exceed 255, so they are overflowing and wrapping back around. That is why you are getting "new data". You can check to see what it is actually working with by casting the points array to be np.uint8 as well.

>>> np.array([[[573, 148]],
              [[25, 223]],
              [[153, 1023]],
              [[730, 863]]],
             dtype=np.uint8)
array([[[ 61, 148]],
       [[ 25, 223]],
       [[153, 255]],
       [[218,  95]]], dtype=uint8)

The reason your list version worked is because you never set the return array to be of type np.uint8, so it defaulted to np.int32.

Answered By - jared