tf.transpose

 tf.transpose(X,perm = [2,0,1])就是对数组维数进行转换，X[i][j][k]=Y[k][i][j]
import tensorflow as tf
import numpy  as np
input_x = [
    [
        [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 ("input_x shape:",np.shape(input_x))

result = tf.transpose(input_x, perm=[2, 0, 1])
with tf.Session() as sess:
    print("input_x",sess.run(result))
input_y=np.zeros((4,2,3),dtype=int)
for i in range(4):
    for j in range(2):
        for k in range(3):
            input_y[i][j][k]=input_x[j][k][i]
print("input_y",input_y)

input_x shape: (2, 3, 4)

input_x [[[ 1  5  9]
  [13 17 21]]

 [[ 2  6 10]
  [14 18 22]]

 [[ 3  7 11]
  [15 19 23]]

 [[ 4  8 12]
  [16 20 24]]]

input_y [[[ 1  5  9]
  [13 17 21]]

 [[ 2  6 10]
  [14 18 22]]

 [[ 3  7 11]
  [15 19 23]]

 [[ 4  8 12]
  [16 20 24]]]