二维平均池化层
AveragePooling2D class
keras.layers.AveragePooling2D(
pool_size, strides=None, padding="valid", data_format=None, name=None, **kwargs
)
平均池化操作,用于2D空间数据.
通过在输入窗口（大小由pool_size定义）上取平均值,沿其空间维度（高度和宽度）对输入进行下采样,
每个输入通道对应一个窗口.窗口沿每个维度按strides移动.
当使用"valid"填充选项时,生成的输出在空间形状（行数或列数）为:
output_shape = math.floor((input_shape - pool_size) / strides) + 1
（当input_shape >= pool_size时）
当使用"same"填充选项时,生成的输出形状为:
output_shape = math.floor((input_shape - 1) / strides) + 1
参数:
pool_size: 整数或2个整数的元组,用于缩小比例
(dim1, dim2).如果只指定一个整数,则所有维度将使用相同长度的窗口.
strides: 整数或2个整数的元组,或None.步幅值.如果为None,
则默认为pool_size.如果只指定一个整数,则所有维度将使用相同的步幅大小.
padding: 字符串,可以是"valid"或"same"（不区分大小写）.
"valid"表示无填充."same"会导致在输入的左右或上下均匀填充,
使得输出具有与输入相同的高度/宽度维度.
data_format: 字符串,可以是"channels_last"或"channels_first".
输入中维度的顺序."channels_last"对应于形状为(batch, height, width, channels)的输入,
而"channels_first"对应于形状为(batch, channels, height, width)的输入.
它默认为在~/.keras/keras.json中的image_data_format值.
如果您从未设置过它,那么它将是"channels_last".
输入形状:
- 如果
data_format="channels_last": 4D张量,形状为(batch_size, height, width, channels). - 如果
data_format="channels_first": 4D张量,形状为(batch_size, channels, height, width).
输出形状:
- 如果
data_format="channels_last": 4D张量,形状为(batch_size, pooled_height, pooled_width, channels). - 如果
data_format="channels_first": 4D张量,形状为(batch_size, channels, pooled_height, pooled_width).
示例:
strides=(1, 1)和padding="valid":
>>> x = np.array([[1., 2., 3.],
... [4., 5., 6.],
... [7., 8., 9.]])
>>> x = np.reshape(x, [1, 3, 3, 1])
>>> avg_pool_2d = keras.layers.AveragePooling2D(pool_size=(2, 2),
... strides=(1, 1), padding="valid")
>>> avg_pool_2d(x)
strides=(2, 2)和padding="valid":
>>> x = np.array([[1., 2., 3., 4.],
... [5., 6., 7., 8.],
... [9., 10., 11., 12.]])
>>> x = np.reshape(x, [1, 3, 4, 1])
>>> avg_pool_2d = keras.layers.AveragePooling2D(pool_size=(2, 2),
... strides=(2, 2), padding="valid")
>>> avg_pool_2d(x)
stride=(1, 1)和padding="same":
>>> x = np.array([[1., 2., 3.],
... [4., 5., 6.],
... [7., 8., 9.]])
>>> x = np.reshape(x, [1, 3, 3, 1])
>>> avg_pool_2d = keras.layers.AveragePooling2D(pool_size=(2, 2),
... strides=(1, 1), padding="same")
>>> avg_pool_2d(x)