PixelShuffle¶

class PixelShuffle(upscale_factor, **kwargs)[源代码]¶

将形状为（, C x r^2, H, W）的向量中的元素重新排列为形状为（, C, H x r, W x r）的向量，其中 r 是一个上采样因子，其中*是零或 batch 的维度。

See the paper: Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network<https://arxiv.org/abs/1609.05158> for more details.

参数: upscale_factor (int) – factor to increase spatial resolution by.

Shape:

input: \((*, C_{in}, H_{in}, W_{in})\), where * is zero or more batch dimensions.
output: \((*, C_{out}, H_{out}, W_{out})\), where:

\[\]

begin{aligned}
C_{out} = C_{in} div text{upscale_factor}^2 \ H_{out} = H_{in} times text{upscale_factor} \ W_{out} = W_{in} times text{upscale_factor} \

end{aligned}

实际案例

>>> import numpy as np
>>> pixel_shuffle = M.PixelShuffle(3)
>>> input = mge.tensor(np.random.randn(1, 9, 4, 4))
>>> output = pixel_shuffle(input)
>>> output.numpy().shape
(1, 1, 12, 12)

Embedding

ConvRelu2d