Pytorch上采样函数 包括interpolate

2倍反卷积上采样：

        self.trans_conv4 = nn.ConvTranspose2d(64, 64, kernel_size=3, stride=2, padding=1, output_padding=1, bias=False)
        self.stage3d = RSU4(128,16,64)

torch.nn的上下采样，可以任意分辨率：

最近邻上下采样：

import torch
from torch import nn

if __name__ == '__main__':

    import os
    data=torch.rand(2,16,16,16)
    upsample = nn.Upsample(size=(20,20), mode='nearest', align_corners=None)  # , mode='bilinear', align_corners=False)

    bbb=upsample(data)

    print(bbb.size())

双线性上下采样：

import torch