Help us understand the problem. What is going on with this article?

Tensorflow API: tf.reverse

More than 3 years have passed since last update.

https://www.tensorflow.org/api_docs/python/tf/reverse

  • 入力のテンソルを反転させる
  • TensorFlow 1.0より機能が変更されている。
reverse(
    tensor,
    axis,
    name=None
)
  • tensor: 入力テンソル
  • axis: 反転させる次元を、integerのリストで指定
  • name: 出力テンソルの名前

使用例

  • 実行環境: TensorFlow 1.2.0
import tensorflow as tf

# 3次元のテンソルを定義
x = tf.constant([[[1, 2, 3], [4, 5, 6], [7, 8, 9]],
                [[11, 12, 13], [14, 15, 16], [17, 18, 19]]])

sess = tf.Session()
x_reverse_0 = tf.reverse(x, [0])
x_reverse_1 = tf.reverse(x, [1])
x_reverse_2 = tf.reverse(x, [2])
x_reverse_0_1 = tf.reverse(x, [0, 1])
x_reverse_0_1_2 = tf.reverse(x, [0, 1, 2])

# 結果の確認
print(str(sess.run(x)) + '\n')

print("tf.reverse(x, [0]):")
print(str(sess.run(x_reverse_0)) + '\n')

print("tf.reverse(x, [1]):")
print(str(sess.run(x_reverse_1)) + '\n')

print("tf.reverse(x, [2]):")
print(str(sess.run(x_reverse_2)) + '\n')

print("tf.reverse(x, [0, 1]):")
print(str(sess.run(x_reverse_0_1)) + '\n')

print("tf.reverse(x, [0, 1, 2]):")
print(str(sess.run(x_reverse_0_1_2)) + '\n')
出力
[[[ 1  2  3]
  [ 4  5  6]
  [ 7  8  9]]

 [[11 12 13]
  [14 15 16]
  [17 18 19]]]

tf.reverse(x, [0]):
[[[11 12 13]
  [14 15 16]
  [17 18 19]]

 [[ 1  2  3]
  [ 4  5  6]
  [ 7  8  9]]]

tf.reverse(x, [1]):
[[[ 7  8  9]
  [ 4  5  6]
  [ 1  2  3]]

 [[17 18 19]
  [14 15 16]
  [11 12 13]]]

tf.reverse(x, [2]):
[[[ 3  2  1]
  [ 6  5  4]
  [ 9  8  7]]

 [[13 12 11]
  [16 15 14]
  [19 18 17]]]

tf.reverse(x, [0, 1]):
[[[17 18 19]
  [14 15 16]
  [11 12 13]]

 [[ 7  8  9]
  [ 4  5  6]
  [ 1  2  3]]]

tf.reverse(x, [0, 1, 2]):
[[[19 18 17]
  [16 15 14]
  [13 12 11]]

 [[ 9  8  7]
  [ 6  5  4]
  [ 3  2  1]]]
Why not register and get more from Qiita?
  1. We will deliver articles that match you
    By following users and tags, you can catch up information on technical fields that you are interested in as a whole
  2. you can read useful information later efficiently
    By "stocking" the articles you like, you can search right away