李沐笔记（使用块的网络VGG）

最新推荐文章于 2024-06-29 19:11:31 发布

qq_42310607

最新推荐文章于 2024-06-29 19:11:31 发布

阅读量345

点赞数 1

文章标签：网络

本文链接：https://blog.csdn.net/qq_42310607/article/details/121705317

版权

本文介绍了如何使用PyTorch实现VGG模型，包括vgg_block函数的定义、vgg函数中卷积层和全连接层的设计，以及如何调整网络结构以适应不同规模的数据。通过Fashion-MNIST数据集进行训练，展示了VGG-11和轻量化版本的搭建过程。

摘要由CSDN通过智能技术生成

import torch
from d2l import torch as d2l
from torch import nn


# 卷积层个数 输入通道个数 输出通道个数
def vgg_block(num_convs,in_channels,out_channels):
    layars=[]
    for _ in range(num_convs):
        layars.append(nn.Conv2d(in_channels,out_channels,kernel_size=3,padding=1))
        layars.append(nn.ReLU())
        in_channels = out_channels
    layars.append(nn.MaxPool2d(kernel_size=2,stride=2))
    return nn.Sequential(*layars)


conv_arch = ((1,64),(1,128),(2,256),(2,512),(2,512))

def vgg(conv_arch):
    conv_blks=[]
    in_channels = 1
    # 卷积层部分
    for (num_convs, out_channels) in conv_arch:
        conv_blks.append(vgg_block(num_convs, in_channels, out_channels))
        in_channels = out_channels

    return nn.Sequential(
        *conv_blks, nn.Flatten(),
        # 全连接层部分
        nn.Linear(out_channels * 7 * 7, 4096), nn.ReLU(), nn.Dropout(0.5),
        nn.Linear(4096, 4096), nn.ReLU(), nn.Dropout(0.5),
        nn.Linear(4096, 10))


net = vgg(conv_arch)

X = torch.randn(size=(1, 1, 224, 224))
for blk in net:
    X = blk(X)
    print(blk.__class__.__name__,'output shape:\t',X.shape)

# 由于VGG-11比AlexNet计算量更大，因此构建一个通道数较少的网络
ratio = 4
small_conv_arch = [(pair[0], pair[1] // ratio) for pair in conv_arch]
net = vgg(small_conv_arch)

lr, num_epochs, batch_size = 0.05, 10, 128
train_iter, test_iter = d2l.load_data_fashion_mnist(batch_size, resize=224)
d2l.train_ch6(net, train_iter, test_iter, num_epochs, lr, d2l.try_gpu())