pytorch学习（二）、网络的构建和卷积层

提示：文章写完后，目录可以自动生成，如何生成可参考右边的帮助文档

前言

如果对卷积层和卷积不是很了解的可以去看博主之前的文章：
卷积到底卷了个啥？五千字详细介绍数学卷积、图像卷积和卷积神经网络的卷积

一、nn.Module（网络的构建）

#导包
import torch.nn as nn
import torch.nn.functional as F
# 创类
class Model(nn.Module):
    def __init__(self):
        super(Model,self).__init__()#初始化子类
        self.conv1 = nn.Conv2d(1,20,5)
        self.conv2 = nn.Conv2d(20,20,5)
	# 向前传播
    def forward(self,x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

二、torch.nn.functional.conv2d（卷积运算）

• 2维卷积

• conv2d(input: Tensor,
  weight: Tensor,
  bias: Tensor | None = None,
  stride: int | Size | list[int] | tuple[int, …] = 1,
  padding: str = “valid”,
  dilation: int | Size | list[int] | tuple[int, …] = 1,
  groups: int = 1) -> Tensor)

#导包
import torch.nn.functional as F
import torch
#设置输入矩阵
input = torch.tensor([[1,2,0,3,1],
                      [0,1,2,3,1],
                      [1,2,1,0,0],
                      [5,2,3,1,1],
                      [2,1,0,1,1]])
#设置卷积核大小
kernel = torch.tensor([[1,2,1],
                       [0,1,0],
                       [2,1,0]])
#转换成函数参数要求的形状
input = torch.reshape(input,(1,1,5,5))
kernel = torch.reshape(kernel,(1,1,3,3))
#调用函数
output = F.conv2d(input,kernel,stride=1)
print(output)
'''
tensor([[[[10, 12, 12],
          [18, 16, 16],
          [13,  9,  3]]]])
'''

三、神经网络–卷积层（一层卷积网络的构建）

• 主要对2维卷积进行演示
• Conv2d(self,
  in_channels: int,
  out_channels: int,
  kernel_size: Union[int, tuple[int, int]],
  stride: Union[int, tuple[int, int]] = 1,
  padding: Union[str, int, tuple[int, int]] = 0,
  dilation: Union[int, tuple[int, int]] = 1,
  groups: int = 1,
  bias: bool = True,
  padding_mode: str = ‘zeros’,
  device: Any = None,
  dtype: Any = None) -> None)
• in_channels：输入通道数
• out_channels：输出通道数
• kernel_size：卷积核大小
• stride：步子（一次移动距离为多少）
• padding：间距

import torchvision
import torch
from torch import nn
from torch.nn import Conv2d
from torch.utils.data import DataLoader
# 从官网下载数据
dataset = torchvision.datasets.CIFAR10('./dataset/CIFAR10',transform=torchvision.transforms.ToTensor(),download=True)
# 加载数据
dataloader = DataLoader(dataset,batch_size=64)
# 构建网络
class Model(nn.Module):
    def __init__(self):
        super(Model, self).__init__()
        self.conv1 = Conv2d(in_channels=3,out_channels=6,kernel_size=3,stride=1,padding=0)
# 向前传播
    def forward(self,x):
        x = self.conv1(x)
        return x
# 实例化网络
model = Model()
#对数据进行卷积
for data in dataloader:
    img,target = data
    output = model(img)
    print(output.shape)