脚本之家,脚本语言编程技术及教程分享平台!
分类导航

Python|VBS|Ruby|Lua|perl|VBA|Golang|PowerShell|Erlang|autoit|Dos|bat|

服务器之家 - 脚本之家 - Python - keras.layer.input()用法说明

keras.layer.input()用法说明

2020-06-17 11:00TinaO-O Python

这篇文章主要介绍了keras.layer.input()用法说明,具有很好的参考价值,希望对大家有所帮助。一起跟随小编过来看看吧

tenserflow建立网络由于先建立静态的graph,所以没有数据,用placeholder来占位好申请内存。

那么keras的layer类其实是一个方便的直接帮你建立深度网络中的layer的类。

该类继承了object,是个基础的类,后续的诸如input_layer类都会继承与layer

由于model.py中利用这个方法建立网络,所以仔细看一下:他的说明详尽而丰富。

input()这个方法是用来初始化一个keras tensor的,tensor说白了就是个数组。他强大到之通过输入和输出就能建立一个keras模型。shape或者batch shape 必须只能给一个。shape = [None,None,None],会创建一个?*?*?的三维数组。

下面还举了个例子,a,b,c都是keras的tensor, `model = Model(input=[a, b], output=c)`

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
def Input(shape=None, batch_shape=None,
     name=None, dtype=None, sparse=False,
     tensor=None):
  """`Input()` is used to instantiate a Keras tensor.
  A Keras tensor is a tensor object from the underlying backend
  (Theano, TensorFlow or CNTK), which we augment with certain
  attributes that allow us to build a Keras model
  just by knowing the inputs and outputs of the model.
  For instance, if a, b and c are Keras tensors,
  it becomes possible to do:
  `model = Model(input=[a, b], output=c)`
  The added Keras attributes are:
    `_keras_shape`: Integer shape tuple propagated
      via Keras-side shape inference.
    `_keras_history`: Last layer applied to the tensor.
      the entire layer graph is retrievable from that layer,
      recursively.
  # Arguments
    shape: A shape tuple (integer), not including the batch size.
      For instance, `shape=(32,)` indicates that the expected input
      will be batches of 32-dimensional vectors.
    batch_shape: A shape tuple (integer), including the batch size.
      For instance, `batch_shape=(10, 32)` indicates that
      the expected input will be batches of 10 32-dimensional vectors.
      `batch_shape=(None, 32)` indicates batches of an arbitrary number
      of 32-dimensional vectors.
    name: An optional name string for the layer.
      Should be unique in a model (do not reuse the same name twice).
      It will be autogenerated if it isn't provided.
    dtype: The data type expected by the input, as a string
      (`float32`, `float64`, `int32`...)
    sparse: A boolean specifying whether the placeholder
      to be created is sparse.
    tensor: Optional existing tensor to wrap into the `Input` layer.
      If set, the layer will not create a placeholder tensor.
  # Returns
    A tensor.
  # Example
  ```python
  # this is a logistic regression in Keras
  x = Input(shape=(32,))
  y = Dense(16, activation='softmax')(x)
  model = Model(x, y)
  ```
  """

tip:我们在model.py中用到了shape这个attribute,

?
1
2
3
4
input_image = KL.Input(
     shape=[None, None, config.IMAGE_SHAPE[2]], name="input_image")
   input_image_meta = KL.Input(shape=[config.IMAGE_META_SIZE],
                 name="input_image_meta")

阅读input()里面的句子逻辑:

可以发现,进入if语句的情况是batch_shape不为空,并且tensor为空,此时进入if,用assert判断如果shape不为空,那么久会有错误提示,告诉你要么输入shape 要么输入batch_shape, 还提示你shape不包含batch个数,就是一个batch包含多少张图片。

那么其实如果tensor不空的话,我们可以发现,也会弹出这个提示,但是作者没有写这种题型,感觉有点没有安全感。注意点好了

?
1
2
3
4
5
if not batch_shape and tensor is None:
  assert shape is not None, ('Please provide to Input either a `shape`'
                ' or a `batch_shape` argument. Note that '
                '`shape` does not include the batch '
                'dimension.')

如果单纯的按照规定输入shape,举个例子:只将shape输入为None,也就是说tensor的dimension我都不知道,但我知道这是个向量,你看着办吧。

input_gt_class_ids = KL.Input(
shape=[None], name="input_gt_class_ids", dtype=tf.int32)

就会调用Input()函数中的这个判断句式,注意因为shape是个List,所以shape is not None 会返回true。同时有没有输入batch_shape的话,就会用shape的参数去创造一个batch_shape.

if shape is not None and not batch_shape:
batch_shape = (None,) + tuple(shape)

比如如果输入:

?
1
2
3
4
shape = (None,)
batch_shape = (None,)+shape
batch_shape
#会得到(None, None)

可以发现,这里要求使用者至少指明你的数据维度,比如图片的话,是三维的,所以shape至少是[None,None,None],而且我认为shape = [None,1] 与shape = [None]是一样的都会创建一个不知道长度的向量。

以上这篇keras.layer.input()用法说明就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持服务器之家。

原文链接:https://blog.csdn.net/u013249853/article/details/88950943

延伸 · 阅读

精彩推荐