Question

我有一个来自api的数组响应，我得到了这个多边形：

poly = [(0,525),(961,525),(961,1003),(0,1003)]

我需要将每个项目乘以0.35，我的代码就是这样：

poly = ak.Array(poly) * (35/100)

我有一个<class 'awkward1.highlevel.Array'>对象，然后将其更改为np array

poly = np.array(poly)

但是这被转换为错误的格式

[(  0.  , 183.75) (336.35, 183.75) (336.35, 351.05) (  0.  , 351.05)]

我需要我的结果数组为np.array int32：

[[0, 184], [336, 184], [336, 351], [0, 351]]

有人可以帮我吗？

Answer 1

poly数组中的每个子数组的类型均为numpy.void。直接将numpy.void转换为numpy.array是行不通的，即，它没有提供所需的数组形状。

遍历每个numpy.void数组，将它们转换为元组，然后将它们转换为numpy.array，同时将float转换为np.int32。

这有效，如上所述：

import numpy as np
import awkward1.highlevel as ak

poly = [(0,525),(961,525),(961,1003),(0,1003)]
poly = ak.Array(poly) * (35/100)
poly = np.array(poly)

# Initializing a list, as NumPy was throwing some error that I haven't resolved.
poly_list = list()
for i, tup in enumerate(poly.copy()):
    # Converting to tuple and then to np.array to fix an error.
    poly_list.append(np.array(tuple(tup)))

# Converting back to a NumPy array.
poly = np.array(poly_list)

print(poly)

# Rounding and converting elements to np.int32 type
poly = poly.astype(np.int32)

我对awkward1软件包不熟悉，因此这可能是一个hack解决方案。但这有效。

Answer 2

我做了^^

import numpy as np
import awkward1 as ak




poly = [(0,525),(961,525),(961,1003),(0,1003)]
poly = ak.Array(poly) * (35/100)
print(poly)
print(type(poly))



poly = np.array(poly)
poly = [[int(i), int(j)] for i, j in poly]

print(poly)
print(type(poly))

我找到了与朋友交谈的解决方案。

转换列表列表中的元组列表（NP数组）

2 个答案: