mindspore.ops.AdaptiveAvgPool3D

class mindspore.ops.AdaptiveAvgPool3D(output_size)[源代码]

三维自适应平均池化。

警告

这是一个实验性API,后续可能修改或删除。

更多参考详见 mindspore.ops.adaptive_avg_pool3d()

参数:
  • output_size (Union[int, tuple]) - 指定输出特征图的尺寸。可以为一个整数或者三元tuple。

输入:
  • x (Tensor) - AdaptiveAvgPool3D的输入,为五维或四维的Tensor。

输出:

Tensor,数据类型与 x 相同。

支持平台:

Ascend GPU CPU

样例:

>>> import mindspore
>>> import numpy as np
>>> from mindspore import nn, Tensor
>>> from mindspore.ops import AdaptiveAvgPool3D
>>> class AdaptiveAvgPool3DNet(nn.Cell):
...     def __init__(self, output_size):
...         super(AdaptiveAvgPool3DNet, self).__init__()
...         self.output_size_ = output_size
...         self.adaptive_avg_pool_3d = AdaptiveAvgPool3D(self.output_size_)
...     def construct(self, x_):
...         return self.adaptive_avg_pool_3d(x_)
...
>>> output_size=(1,1,1)
>>> input_x_val = np.zeros((1,1,2,2,2))
>>> input_x_val[:,:,0,:,:]  += 1
>>> input_x = Tensor(input_x_val, mindspore.float32)
>>> adaptive_avg_pool_3d = AdaptiveAvgPool3DNet(output_size)
>>> output = adaptive_avg_pool_3d(input_x)
>>> print(output)
[[[[[0.5]]]]]