I need to convert array like this:
[[1527 1369 86 86]
[ 573 590 709 709]
[1417 1000 68 68]
[1361 1194 86 86]]
to like this:
[(726, 1219, 1281, 664),
(1208, 1440, 1283, 1365),
(1006, 1483, 1069, 1421),
(999, 1414, 1062, 1351),]
I tried using convert diretly to tuple but got this:
( array([1527, 1369, 86, 86], dtype=int32),
array([573, 590, 709, 709], dtype=int32),
array([1417, 1000, 68, 68], dtype=int32),
array([1361, 1194, 86, 86], dtype=int32))
(array([701, 899, 671, 671], dtype=int32),)
解决方案
The array method tolist is a easy and fast way of converting an array to a list. It handles multiple dimensions correctly:
In [92]: arr = np.arange(12).reshape(3,4)
In [93]: arr
Out[93]:
array([[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]])
In [94]: arr.tolist()
Out[94]: [[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]
For most purposes such as list of lists is just as good as a list of tuples, or tuple of tuples. They differ only in mutability.
But if you must have a tuples, a list comprehension does the conversion nicely.
In [95]: [tuple(x) for x in arr.tolist()]
Out[95]: [(0, 1, 2, 3), (4, 5, 6, 7), (8, 9, 10, 11)]
An alternative [tuple(x) for x in arr] is a bit slower, because it is iterating on the array rather than on a list. It also produces a different result - though you have to examine the type of the tuple elements to see that.
I strongly recommend starting with the tolist method, and doing any list to tuple conversions after.