1. 卷积原理
- Conv1d代表一维卷积,Conv2d代表二维卷积,Conv3d代表三维卷积。
- kernel_size在训练过程中不断调整,定义为3就是3 * 3的卷积核,实际我们在训练神经网络过程中其实就是对kernel_size不断调整。
torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0,
dilation=1, groups=1, bias=True, padding_mode='zeros', device=None, dtype=None)
常用参数:
in_channels (int) – 输入图像中的通道数
out_channels (int) – 由卷积产生的通道数
padding (int, tuple or str, optional) – 添加到输入四个边的Padding Default: 0
padding_mode (str, optional) – , , or . Default:
'zeros'
'reflect'
'replicate'
'circular'
'zeros'
dilation (int or tuple, optional) – Spacing between kernel elements. Default: 1
groups (int, optional) – Number of blocked connections from input channels to output channels. Default: 1
bias (bool, optional) – If , adds a learnable bias to the output. Default:
True
True
可以根据输入的参数获得输出的情况,如下图所示。
2. 搭建卷积层 并处理图片
import torch
from torch import nn
import torchvision
from torch.nn import Conv2d
from torch.utils.data import DataLoader
#搭建卷积层
dataset = torchvision.datasets.CIFAR10("./dataset",train=False,transform=torchvision.transforms.ToTensor(),download=True)
dataloader = DataLoader(dataset, batch_size=64)
class Tudui(nn.Module):
def __init__(self):
super(Tudui, self).__init__()
self.conv1 = Conv2d(in_channels=3,out_channels=6,kernel_size=3,stride=1,padding=0) # 彩色图像输入为3层,我们想让它的输出为6层,选3 * 3 的卷积
def forward(self,x):
x = self.conv1(x)
return x
#卷积层处理图片
tudui = Tudui()
writer = SummaryWriter("logs")
step = 0
for data in dataloader:
imgs, targets = data
output = tudui(imgs)
print(imgs.shape) # 输入为3通道32×32的64张图片
print(output.shape) # 输出为6通道30×30的64张图片
writer.add_images("input", imgs, step)
output = torch.reshape(output,(-1,3,30,30)) #-1为占位符,让计算机自行计算后填入
# 把原来6个通道拉为3个通道,为了保证所有维度总数不变,其余的分量分到第一个维度中
writer.add_images("output", output, step)
step = step + 1