深度学习网络中numpy多维数组的说明

目前在计算机视觉中应用的数组维度最多有四维，可以表示为 (Batch_size, Row, Column, Channel)

以下将要从二维数组到四维数组进行代码的简单说明：

Tips：

1) 在numpy中所有的index都是从0开始。

2) axis = 0 对Cloumn(Width)操作； axis = 1 对Row(Height)操作； axis = 2 or -1 对Channel(Depth)操作

1. 二维数组 (Row, Column)

import numpy as np

# Set a matrix with (2*3)

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

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

print(array.shape) # (Row, Column)
(2, 3) 

print(array[0,1])
2   

2. 三维数组 (Row, Column, Channel)

import numpy as np

# Set a matrix with (2*3*4)
array = np.array([
    [[1,2,3,4],[5,6,7,8],[9,10,11,12]],
    [[13,14,15,16],[17,18,19,20],[21,22,23,24]]
                 ])

print(array)
[[[ 1  2  3  4]
  [ 5  6  7  8]
  [ 9 10 11 12]]

 [[13 14 15 16]
  [17 18 19 20]
  [21 22 23 24]]]

print(array.shape)
(2, 3, 4)  #(Row, Column, Channel)

print(array[0,1,2])
7

3. 四维数组(Batch_size, Row, Column, Channel)

import numpy as np
# Set a matrix with (2*2*3*4)
array = np.array([
    [[[1,2,3,4],[5,6,7,8],[9,10,11,12]],[[13,14,15,16],[17,18,19,20],[21,22,23,24]]],
    [[[21,22,23,24],[17,18,19,20],[13,14,15,16]],[[9,10,11,12],[5,6,7,8],[1,2,3,4]]]
                ])

print(array)
[[[[ 1  2  3  4]
   [ 5  6  7  8]
   [ 9 10 11 12]]

  [[13 14 15 16]
   [17 18 19 20]
   [21 22 23 24]]]

 [[[21 22 23 24]
   [17 18 19 20]
   [13 14 15 16]]

  [[ 9 10 11 12]
   [ 5  6  7  8]
   [ 1  2  3  4]]]]

print(array.shape) #(Batch_size, Row, Column, Channel)
(2, 2, 3, 4)

print(array[1,0,1,2])
19

print(array[1]) # Choice Batch_size 1
[[[21 22 23 24]
  [17 18 19 20]
  [13 14 15 16]]

 [[ 9 10 11 12]
  [ 5  6  7  8]
  [ 1  2  3  4]]]

以上。

原文地址：https://www.cnblogs.com/godislight/p/10789642.html