GlobalAveragePooling3D
类keras.layers.GlobalAveragePooling3D(data_format=None, keepdims=False, **kwargs)
3D 数据的全局平均池化操作。
参数
"channels_last"
或 "channels_first"
。输入维度顺序。"channels_last"
对应形状为 (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels)
的输入,而 "channels_first"
对应形状为 (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3)
的输入。默认为 Keras 配置文件 ~/.keras/keras.json
中找到的 image_data_format
值。如果从未设置过,则默认为 "channels_last"
。keepdims
为 False
(默认值),则张量的秩将针对空间维度降低。如果 keepdims
为 True
,则保留空间维度,长度为 1。行为与 tf.reduce_mean
或 np.mean
相同。输入形状
data_format='channels_last'
: 5D 张量,形状为:(batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels)
data_format='channels_first'
: 5D 张量,形状为:(batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)
输出形状
keepdims=False
: 2D 张量,形状为 (batch_size, channels)
。keepdims=True
: - 如果 data_format="channels_last"
: 5D 张量,形状为 (batch_size, 1, 1, 1, channels)
- 如果 data_format="channels_first"
: 5D 张量,形状为 (batch_size, channels, 1, 1, 1)
示例
>>> x = np.random.rand(2, 4, 5, 4, 3)
>>> y = keras.layers.GlobalAveragePooling3D()(x)
>>> y.shape
(2, 3)