评价此页

PixelUnshuffle#

class torch.nn.PixelUnshuffle(downscale_factor)[source]#

反转 PixelShuffle 操作。

反转 PixelShuffle 操作,将形状为 (,C,H×r,W×r)(*, C, H \times r, W \times r) 的张量重新排列,使其变为形状为 (,C×r2,H,W)(*, C \times r^2, H, W) 的张量,其中 r 是降采样因子。

更多详情请参见论文:Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network by Shi et al. (2016)。

参数

downscale_factor (int) – 用于减小空间分辨率的因子

形状
  • 输入:(,Cin,Hin,Win)(*, C_{in}, H_{in}, W_{in}),其中*是零个或多个批次维度

  • 输出:(,Cout,Hout,Wout)(*, C_{out}, H_{out}, W_{out}),其中

Cout=Cin×downscale_factor2C_{out} = C_{in} \times \text{downscale\_factor}^2
Hout=Hin÷downscale_factorH_{out} = H_{in} \div \text{downscale\_factor}
Wout=Win÷downscale_factorW_{out} = W_{in} \div \text{downscale\_factor}

示例

>>> pixel_unshuffle = nn.PixelUnshuffle(3)
>>> input = torch.randn(1, 1, 12, 12)
>>> output = pixel_unshuffle(input)
>>> print(output.size())
torch.Size([1, 9, 4, 4])