pytorch-study/09.ipynb

In [4]:

import torch
import torch.nn as nn

In [8]:

input_feat = torch.tensor([[4, 1, 7, 5], [4, 4, 2, 5], [7, 7, 2, 4], [1, 0, 2, 4]], dtype=torch.float32).unsqueeze(0).unsqueeze(0)
print(input_feat)
print(input_feat.shape)

tensor([[[[4., 1., 7., 5.],
          [4., 4., 2., 5.],
          [7., 7., 2., 4.],
          [1., 0., 2., 4.]]]])
torch.Size([1, 1, 4, 4])

In [16]:

conv2d = nn.Conv2d(1, 1, (2, 2), stride=1, padding='same', bias=False)
# 卷积核要有四个维度：输出通道数，输入通道数，卷积核高度，卷积核宽度
kernels = torch.tensor([[[[1, 0], [2, 1]]]], dtype=torch.float32)
conv2d.weight = nn.Parameter(kernels, requires_grad=False)  # 设置卷积核
# 默认情况随机初始化参数
print(conv2d.weight)
print(conv2d.bias)

Parameter containing:
tensor([[[[1., 0.],
          [2., 1.]]]])
None

In [17]:

output = conv2d(input_feat)
output

Out[17]:

tensor([[[[16., 11., 16., 15.],
          [25., 20., 10., 13.],
          [ 9.,  9., 10., 12.],
          [ 1.,  0.,  2.,  4.]]]])