Convert numpy ndarray to tuple of tuples in optimize method

Question

I have numpy ndarray, result of opencv findContours().
I want to convert each element of the result from numpy array to tuple of tuples efficiently.
Tried tolist(), asarray() etc but none of them give me the exact result.

example

numpy array:

    [[[191 307]]

     [[190 308]]

     [[181 308]]]

to tuple of tuples:

((191,307),(190,308),(181,308))

update
tuple(elements[0]) return

(array([[191 ,307]], dtype=int32),array([[190, 308]], dtype=int32),array([[181,308]], dtype=int32))

Im not sure. assume I write this line x=((0,0),(1,1)). x is list of tuples right? @MatinaG — rety, Aug 16 '18 at 13:01
try `my_list = numpy.ndarray.tolist(my_array)` first and then `tuple(my_list)` — U3.1415926, Aug 16 '18 at 13:03

score 2 · Accepted Answer · answered Aug 16 '18 at 13:28

2

In [9]: a = numpy.array([[[191, 307]],
   ...:                  [[190, 308]],
   ...:                  [[181, 308]]])

In [10]: tuple(tuple(row[0]) for row in a)
Out[10]: ((191, 307), (190, 308), (181, 308))

answered Aug 16 '18 at 13:28

kuppern87

1,125
9
14

score 1 · Answer 2 · answered Aug 16 '18 at 17:05

1

Your array is 3d:

In [356]: a.shape
Out[356]: (3, 1, 2)

If you remove the middle dimension, it's easy to iterate on the rest:

In [357]: tuple(tuple(i) for i in a[:,0,:])
Out[357]: ((191, 307), (190, 308), (181, 308))

If it doesn't have to be tuples, tolist is enough:

In [358]: a[:,0,:].tolist()
Out[358]: [[191, 307], [190, 308], [181, 308]]

answered Aug 16 '18 at 17:05

hpaulj

221,503
14
230
353

I've just asked [Convert any multi-dimensional numpy array to tuple of tuples of tuples… agnostic of number of dimensions](https://stackoverflow.com/q/68361779/3904031) – uhoh Jul 13 '21 at 11:46

Convert numpy ndarray to tuple of tuples in optimize method

2 Answers2

Linked