tf.transpose 直覺感受

tensorflow cifar10的cifar10_input.py中read_cifar10（）最後有一個轉置語句

result.uint8image = tf.transpose(depth_major, [1, 2, 0])

用以下語句直覺感受一下

import as 
      

import numpy as np

1231231231231231231231231231'depth'2'height'3'width'dtypeshape333perm120with as feed_dictprintprintfor in rangelen0for in rangelen1for in rangelen2for in rangelen0for in rangelen1for in rangelen2printprint'before_trans:\n'print
      

print('after_trans:\n',np.array(after_trans))

# 結果

before_trans:

[[['depth' 'height' 'width']

['depth' 'height' 'width']

['depth' 'height' 'width']]

[['depth' 'height' 'width']

['depth' 'height' 'width']]]

after_trans:

[[['depth' 'depth' 'depth']

['height' 'height' 'height']

['width' 'width' 'width']]

[['depth' 'depth' 'depth']

['width' 'width' 'width']]]