numpy 学习笔记

1. np.array([127, 127, 127]) 生成一个一维数组 [127, 127, 127]。

参考: https://numpy.org/doc/stable/reference/generated/numpy.array.html

2. a = np.reshape(list(range(24)), (2, 3, 4)) reshape 的第二个参数是维度参数，比如这个是一个三维数组，行数 2 行，列数 3 行，每个单位 4 个元素。

参考: https://numpy.org/doc/stable/reference/generated /numpy.reshape.html?highlight=reshape#numpy.reshape

3. image = np.transpose(image, [2, 0, 1]) 矩阵转置，第二个参数是相应的维度变化。 原始的维度是 [0, 1, 2] 其他的都是需要转换的。比如这个 [2, 0, 1] 意思是 把 第三维的数据提取出来作为第一维的开头数据，第一维的数据提取出来放到第二维，第二维的数据放到第三维。

arr = np.arange(16)
matrix = arr.reshape(2, 2, 4)
print("matrix: ", matrix)

matrix_4 = matrix.transpose((2, 0, 1))
print("matrix 4: ", matrix_4)

输出结果为

matrix:  [ [ [ 0  1  2  3]
             [ 4  5  6  7] ]

           [ [ 8  9 10 11]
             [12 13 14 15] ] ]

matrix 4:  [ [ [ 0  4]
               [ 8 12]]

             [ [ 1  5]
               [ 9 13] ]

             [ [ 2  6]
               [10 14] ]

             [ [ 3  7]
               [11 15] ] ]

参考: https://blog.csdn.net/u012762410/article/details/78912667

4. np.expand_dims(a, axis) 第二个参数表示具体再那个地方增加一个维度。

import numpy as np

a = np.reshape(list(range(24)), (2, 3, 4))
a_new = np.expand_dims(a, axis=0)
print('a =', a)
print('a_new =', a_new)
print('a.shape = ', a.shape)
print('a_new.shape = ', a_new.shape)

a = [[[ 0  1  2  3]
      [ 4  5  6  7]
      [ 8  9 10 11]]

     [[12 13 14 15]
      [16 17 18 19]
      [20 21 22 23]]]

a_new = [[[[ 0  1  2  3]
           [ 4  5  6  7]
           [ 8  9 10 11]]

          [[12 13 14 15]
           [16 17 18 19]
           [20 21 22 23]]]]
a.shape =  (2, 3, 4)
a_new.shape =  (1, 2, 3, 4)

import numpy as np

a = np.reshape(list(range(24)), (2, 3, 4))
print('a =', a)
print('np.expand_dims(a, axis=1) =', np.expand_dims(a, axis=1))
print('np.expand_dims(a, axis=2) =', np.expand_dims(a, axis=2))
print('np.expand_dims(a, axis=3) =', np.expand_dims(a, axis=3))
print('a.shape = ', a.shape)
print('np.expand_dims(a, axis=1).shape =', np.expand_dims(a, axis=1).shape)
print('np.expand_dims(a, axis=2).shape =', np.expand_dims(a, axis=2).shape)
print('np.expand_dims(a, axis=3).shape =', np.expand_dims(a, axis=3).shape)

a = [[[ 0  1  2  3]
      [ 4  5  6  7]
      [ 8  9 10 11]]

     [[12 13 14 15]
      [16 17 18 19]
      [20 21 22 23]]]
np.expand_dims(a, axis=1) = [[[[ 0  1  2  3]
                               [ 4  5  6  7]
                               [ 8  9 10 11]]]

                             [[[12 13 14 15]
                               [16 17 18 19]
                               [20 21 22 23]]]]
np.expand_dims(a, axis=2) = [[[[ 0  1  2  3]]
                              [[ 4  5  6  7]]
                              [[ 8  9 10 11]]]

                             [[[12 13 14 15]]
                              [[16 17 18 19]]
                              [[20 21 22 23]]]]
np.expand_dims(a, axis=3) = [[[[ 0]
                               [ 1]
                               [ 2]
                               [ 3]]

                              [[ 4]
                               [ 5]
                               [ 6]
                               [ 7]]

                              [[ 8]
                               [ 9]
                               [10]
                               [11]]]

                             [[[12]
                               [13]
                               [14]
                               [15]]

                              [[16]
                               [17]
                               [18]
                               [19]]

                              [[20]
                               [21]
                               [22]
                               [23]]]]
a.shape =  (2, 3, 4)
np.expand_dims(a, axis=1).shape = (2, 1, 3, 4)
np.expand_dims(a, axis=2).shape = (2, 3, 1, 4)
np.expand_dims(a, axis=3).shape = (2, 3, 4, 1)

参考: https://blog.csdn.net/weixin_41560402/article/details/105289015

5. .astype(np.float32) 转换数据类型。

参考: https://blog.csdn.net/qq_34638161/article/details/102853276

6. np.concatenate() 对数列或者矩阵进行合并。

import numpy as np
a=[1,2,3]
b=[4,5,6]
np.concatenate((a,b),axis=0)

array([1, 2, 3, 4, 5, 6])

参考: https://www.jianshu.com/p/a094a954ff61

7. numpy 和 list 互转

numpy -> list: array.tolist() list -> numpy: numpy.array(list) 参考: https://www.cnblogs.com/WMT-Azura/p/11138084.html

参考数目: https://www.runoob.com/numpy/numpy-array-manipulation.html https://numpy.org/doc/stable/user/c-info.html https://numpy.org.cn/article/advanced/numpy_exercises_for_data_analysis.html#numpy%E6%95%B0%E6%8D%AE%E5%88%86%E6%9E%90%E9%97%AE%E7%AD%94