图1. 输入参数
in_channels:输入通道个数。例如彩色图片有3个通道
out_channels:输出通道个数。out_channals=2时,会生成2个卷积核与输入图像进行卷积。
kernel_size:卷积核大小。
padding_mode:padding填充模式。
groups:一般设置成1
bias:偏置。一般设置为True
图2. 计算图像长和宽的方法
import torch
import torchvision
from torch.nn import Conv2d
from torch.utils.data import DataLoader
from torch import nn
from torch.utils.tensorboard import SummaryWriter
# 准备数据集
dataset = torchvision.datasets.CIFAR10("../pytorch_learn/dataset2", train=False, transform=torchvision.transforms.ToTensor(), download=True)
# 下载数据
dataloader = DataLoader(dataset, batch_size=64, shuffle=True)
# 搭建简单神经网络
class conv(nn.Module):
def __init__(self):
super(conv, self).__init__()
self.cov1 = Conv2d(in_channels=3, out_channels=6, kernel_size=3, stride=1, padding=0)
def forward(self, x):
x = self.cov1(x)
return x
writer = SummaryWriter("logs")
conv = conv()
# 将每张图像放到神经网络中
step = 0
for data in dataloader:
img, target = data
output = conv(img)
# torch.Size([64, 3, 32, 32])
writer.add_images("input", img, step)
# torch.Size([64, 6, 30, 30])
# 由于输出是6通道的,而输入是3通道的。所以会报错,需要进行变换
output = torch.reshape(output, (-1, 3, 30, 30))
writer.add_images("output", output, step)
step += 1
writer.close()