Question

我试图在tensorflow 1.0中编写一个简单的数据生成器来解决图像分类问题。我有一个图像路径列表和相应的标签，如2个列表：路径和标签。

我正在使用以下代码来获取数据对象和迭代器。

dataset = (
    tf.data.Dataset.from_tensor_slices((paths, labels))
    .shuffle(buffer_size = len(paths))
    .map(parse_fn, num_parallel_calls = 4)
    .batch(32)
    .prefetch(1)
)
train_iter = dataset.make_initializable_iterator()
train_next = train_iter.get_next()

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    sess.run(train_iter)
    x, y = sess.run(train_next)

但是我遇到以下错误：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-14-139e601c664d> in <module>()
     25 with tf.Session() as sess:
     26     sess.run(tf.global_variables_initializer())
---> 27     sess.run(train_iter)
     28     x, y = sess.run(train_next)
     29     print(x.shape, y.shape)

/home/surya/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc in run(self, fetches, feed_dict, options, run_metadata)
    927     try:
    928       result = self._run(None, fetches, feed_dict, options_ptr,
--> 929                          run_metadata_ptr)
    930       if run_metadata:
    931         proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)

/home/surya/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc in _run(self, handle, fetches, feed_dict, options, run_metadata)
   1135     # Create a fetch handler to take care of the structure of fetches.
   1136     fetch_handler = _FetchHandler(
-> 1137         self._graph, fetches, feed_dict_tensor, feed_handles=feed_handles)
   1138 
   1139     # Run request and get response.

/home/surya/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc in __init__(self, graph, fetches, feeds, feed_handles)
    469     """
    470     with graph.as_default():
--> 471       self._fetch_mapper = _FetchMapper.for_fetch(fetches)
    472     self._fetches = []
    473     self._targets = []

/home/surya/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc in for_fetch(fetch)
    269         if isinstance(fetch, tensor_type):
    270           fetches, contraction_fn = fetch_fn(fetch)
--> 271           return _ElementFetchMapper(fetches, contraction_fn)
    272     # Did not find anything.
    273     raise TypeError('Fetch argument %r has invalid type %r' % (fetch,

/home/surya/anaconda2/lib/python2.7/site-packages/tensorflow/python/client/session.pyc in __init__(self, fetches, contraction_fn)
    302         raise TypeError('Fetch argument %r has invalid type %r, '
    303                         'must be a string or Tensor. (%s)' %
--> 304                         (fetch, type(fetch), str(e)))
    305       except ValueError as e:
    306         raise ValueError('Fetch argument %r cannot be interpreted as a '

TypeError: Fetch argument <tensorflow.python.data.ops.iterator_ops.Iterator object at 0x7fe5326ebf90> has invalid type <class 'tensorflow.python.data.ops.iterator_ops.Iterator'>, must be a string or Tensor. (Can not convert a Iterator into a Tensor or Operation.)

如果将迭代器更改为

，则不会收到此错误

data_iter = dataset.make_one_shot_iterator()

为什么会出现此错误，以及如何解决？谢谢！

Answer 1

只需更改行：

sess.run(train_iter)

收件人：

sess.run(train_iter.initializer)

这是因为您要执行迭代器的初始化程序，而不是迭代器本身。

data.make_initializable_iterator（）引发错误：TypeFetch参数必须为字符串或张量

1 个答案: