1. np.expand_dims

用于扩展数组的维度

执行程序后注意观察中括号[ ]的位置和数量

np.expand_dims(a, axis=0)表示在axis=0维度处扩展维度，加一层中括号[ ];

np.expand_dims(a, axis=1)表示在axis=1维度处扩展维度，加一层中括号[ ];

np.expand_dims(a, axis=2)表示在axis=2维度处扩展维度，加一层中括号[ ];

np.expand_dims(a, axis=-1)表示在axis=-1(最后)维度处扩展维度，加一层中括号[ ];

(py3.6) E:\PYTHON>ipython
Python 3.6.13 |Anaconda, Inc.| (default, Mar 16 2021, 11:37:27) [MSC v.1916 64 bit (AMD64)]
Type 'copyright', 'credits' or 'license' for more information
IPython 7.16.3 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import numpy as np

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

In [3]: a
Out[3]:
array([[1, 2, 3],
       [4, 5, 6]])

In [4]: a.shape
Out[4]: (2, 3)

1.1 axis=0

In [5]: b = np.expand_dims(a, axis=0)

In [6]: b
Out[6]:
array([[[1, 2, 3],
        [4, 5, 6]]])

In [7]: b.shape
Out[7]: (1, 2, 3)

1.2 axis=1

In [8]: c = np.expand_dims(a, axis=1)

In [9]: c
Out[9]:
array([[[1, 2, 3]],

       [[4, 5, 6]]])

In [10]: c.shape
Out[10]: (2, 1, 3)

1.3 axis=2

In [11]: d = np.expand_dims(a, axis=2)

In [12]: d
Out[12]:
array([[[1],
        [2],
        [3]],

       [[4],
        [5],
        [6]]])

In [13]: d.shape
Out[13]: (2, 3, 1)

1.4 axis=-1 

In [14]: e = np.expand_dims(a, axis=-1)

In [15]: e
Out[15]:
array([[[1],
        [2],
        [3]],

       [[4],
        [5],
        [6]]])
In [16]: e.shape
Out[16]: (2, 3, 1)

1.5 axis=3

In [17]: f = np.expand_dims(a, axis=3)
---------------------------------------------------------------------------
AxisError                                 Traceback (most recent call last)
<ipython-input-16-d7316647942f> in <module>
----> 1 f = np.expand_dims(a, axis=3)

<__array_function__ internals> in expand_dims(*args, **kwargs)

D:\RuanJianAnZhunangWeiZhi\anaconda\anaconda3\envs\py3.6\lib\site-packages\numpy\lib\shape_base.py in expand_dims(a, axis)
    595
    596     out_ndim = len(axis) + a.ndim
--> 597     axis = normalize_axis_tuple(axis, out_ndim)
    598
    599     shape_it = iter(a.shape)

D:\RuanJianAnZhunangWeiZhi\anaconda\anaconda3\envs\py3.6\lib\site-packages\numpy\core\numeric.py in normalize_axis_tuple(axis, ndim, argname, allow_duplicate)
   1325             pass
   1326     # Going via an iterator directly is slower than via list comprehension.
-> 1327     axis = tuple([normalize_axis_index(ax, ndim, argname) for ax in axis])
   1328     if not allow_duplicate and len(set(axis)) != len(axis):
   1329         if argname:

D:\RuanJianAnZhunangWeiZhi\anaconda\anaconda3\envs\py3.6\lib\site-packages\numpy\core\numeric.py in <listcomp>(.0)
   1325             pass
   1326     # Going via an iterator directly is slower than via list comprehension.
-> 1327     axis = tuple([normalize_axis_index(ax, ndim, argname) for ax in axis])
   1328     if not allow_duplicate and len(set(axis)) != len(axis):
   1329         if argname:

AxisError: axis 3 is out of bounds for array of dimension 3

实例：

素材：注意程序中用的是相对路径

import tensorflow as tf
tf.compat.v1.disable_eager_execution()
import numpy as np
from tensorflow.keras.preprocessing import image
from skimage import io

image_path="../images/lena.jpg"

# 加载图像
img = io.imread(image_path)

# 显示图像
io.imshow(img)
io.show()

# 图像预处理
x = image.img_to_array(img)
y = np.expand_dims(x, axis = 0)

print("x:", x)
print("y:", y)

print(x.shape)
print(y.shape)

结果：

D:\RuanJianAnZhunangWeiZhi\anaconda\anaconda3\envs\py3.6\python.exe E:\Python_files\Tensorflow_dl\TensorFlowProgramming\TensorFlowProgramming\ch6CNN\pool_op_demo2.py 
2023-04-04 08:24:33.571665: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found
2023-04-04 08:24:33.571878: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
x: [[[221. 139. 125.]
  [221. 139. 125.]
  [222. 139. 125.]
  ...
  [223. 136. 108.]
  [216. 131. 102.]
  [211. 126.  95.]]

 [[221. 139. 125.]
  [221. 139. 125.]
  [222. 139. 125.]
  ...
  [225. 139. 112.]
  [217. 132. 103.]
  [211. 126.  97.]]

 [[223. 140. 124.]
  [223. 140. 124.]
  [223. 140. 124.]
  ...
  [225. 141. 117.]
  [214. 130. 104.]
  [206. 122.  96.]]

 ...

 [[ 81.  26.  58.]
  [ 83.  28.  60.]
  [ 84.  27.  59.]
  ...
  [161.  72.  90.]
  [163.  75.  91.]
  [163.  75.  91.]]

 [[ 80.  25.  56.]
  [ 82.  27.  58.]
  [ 83.  26.  58.]
  ...
  [164.  76.  92.]
  [167.  76.  93.]
  [167.  76.  93.]]

 [[ 80.  25.  56.]
  [ 81.  26.  57.]
  [ 82.  25.  57.]
  ...
  [166.  78.  94.]
  [169.  78.  95.]
  [169.  78.  95.]]]
y: [[[[221. 139. 125.]
   [221. 139. 125.]
   [222. 139. 125.]
   ...
   [223. 136. 108.]
   [216. 131. 102.]
   [211. 126.  95.]]

  [[221. 139. 125.]
   [221. 139. 125.]
   [222. 139. 125.]
   ...
   [225. 139. 112.]
   [217. 132. 103.]
   [211. 126.  97.]]

  [[223. 140. 124.]
   [223. 140. 124.]
   [223. 140. 124.]
   ...
   [225. 141. 117.]
   [214. 130. 104.]
   [206. 122.  96.]]

  ...

  [[ 81.  26.  58.]
   [ 83.  28.  60.]
   [ 84.  27.  59.]
   ...
   [161.  72.  90.]
   [163.  75.  91.]
   [163.  75.  91.]]

  [[ 80.  25.  56.]
   [ 82.  27.  58.]
   [ 83.  26.  58.]
   ...
   [164.  76.  92.]
   [167.  76.  93.]
   [167.  76.  93.]]

  [[ 80.  25.  56.]
   [ 81.  26.  57.]
   [ 82.  25.  57.]
   ...
   [166.  78.  94.]
   [169.  78.  95.]
   [169.  78.  95.]]]]
(512, 512, 3)
(1, 512, 512, 3)

Process finished with exit code 0

参考文章：

https://blog.csdn.net/qq_37924224/article/details/119816771

https://blog.csdn.net/hong615771420/article/details/83448878