错误：从tensorflow.examples.tutorials.mnist导入input_data

Question

我的环境如下：

* Windows 7, 64 bit
* Anaconda Navigator 1.8.7
* python 3.6.5
* tensorflow 1.8.0

在Python中，我输入：

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

我遇到了如下错误：

>>> from tensorflow.examples.tutorials.mnist import input_data
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\examples\tutorials\mnist\__init__.py", line 21, in <module>
    from tensorflow.examples.tutorials.mnist import input_data
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\examples\tutorials\mnist\input_data.py", line 30, in <module>
    from tensorflow.contrib.learn.python.learn.datasets.mnist import read_data_sets
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\__init__.py", line 34, in <module>
    from tensorflow.contrib import data
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\data\__init__.py", line 67, in <module>
    from tensorflow.contrib.data.python.ops.error_ops import ignore_errors
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\data\python\ops\error_ops.py", line 20, in <module>
    from tensorflow.contrib.data.python.ops import contrib_op_loader  # pylint: disable=unused-import
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\data\python\ops\contrib_op_loader.py", line 24, in <module>
    resource_loader.get_path_to_datafile("../../_dataset_ops.so"))
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\util\loader.py", line 56, in load_op_library
    ret = load_library.load_op_library(path)
  File "E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\python\framework\load_library.py", line 56, in load_op_library
    lib_handle = py_tf.TF_LoadLibrary(library_filename)
tensorflow.python.framework.errors_impl.NotFoundError: E:\Anaconda3\envs\opencv\lib\site-packages\tensorflow\contrib\data\python\ops\..\..\_dataset_ops.so not found
>>>

它还会弹出一个窗口，显示以下内容：

The procedure entry point ?addcleanup@arenaimpl@internal@protobuf@google@@QEAAXPEAXP6AX0@Z@Z could not be located in the dynamic link library _pywarp_tensorflow_internal.pyd

请帮忙，非常感谢。

最热情的问候， Suryadi

- sliawatimena

---

请查看：https://stackoverflow.com/questions/50666681/how-to-load-mnist-via-tensorflow-including-download/50669146#50669146 - Vijay Mariappan

我在你提供的链接上运行代码，出现了错误：---> 16 train_image.shape = (10, 784) NameError: name 'train_image' is not defined - sliawatimena

代码中没有train_image这个变量，它只是一个注释。请确保正确复制代码，我刚试过，它可以正常工作。 - Vijay Mariappan

Answer 1

tensorflow.examples.tutorials.mnist 和 tf.contrib.learn.datasets.load_dataset('mnist') 都会抛出废弃警告。你可以通过keras datasets进行加载：

(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data().

您可以在此处查看如何加载MNIST并将其用于训练：如何通过TensorFlow加载MNIST（包括下载）？。

- V.M

Answer 2

你还需要将图像数据转换为浮点表示。

 mnist_train, mnist_test = tf.keras.datasets.mnist.load_data()
 train_data = np.float16(mnist_train[0])  # Returns np.array
 train_labels = np.asarray(mnist_train[1], dtype=np.int32)
 eval_data = np.float16(mnist_test[0])  # Returns np.array
 eval_labels = np.asarray(mnist_test[1], dtype=np.int32)

- Rabin Neslo