GoogLeNet模型详解

news2025/1/18 6:58:29

模型介绍

GoogLeNet是谷歌工程师设计的深度神经网络结构，于2014年在ImageNet比赛中取得了冠军。它的设计特点在于既有深度，又在横向上拥有“宽度”，并采用了一种名为Inception的核心子网络结构。这个网络名字中的“GoogLeNet”是对LeNet的致敬，LeNet是早期由Yann LeCun设计的卷积神经网络。

论文地址：https://arxiv.org/pdf/1409.4842.pdf

模型结构

左图(a)就是Inception模块的设计思想，它的设计是通过并行使用多个卷积和池化操作来提取特征，然后将这四个操作的输出进行通道维度的拼接，这样获得的图像是经过不同核大小的卷积提取的特征，有利于捕捉不同尺度的特征。

1x1卷积核用于学习通道间的线性关系和降维。
3x3卷积核用于捕捉局部特征和空间相关性。
5x5卷积核用于学习更大范围的特征。
最大池化用于捕捉局部的空间不变性。

不过，Inception的计算复杂度相对较高，尤其是包含大尺寸卷积核的部分。并行使用了多个卷积和池化操作，模块内的参数数量较大，这可能导致过拟合，尤其是在数据量有限的情况下。

为了减小参数量，又使用像右图(b)所示的那样，在每个3x3和5x5的卷积层之前，增加1x1的卷积层来控制输出通道数，在最大池化层后面增加1x1卷积层减小输出通道数。

代码实现

该网络的实现比较简单，没有加辅助分类器。

import torch
import torch.nn as nn

__all__ = ["GoogLeNet"]

class Inception(nn.Module):
    def __init__(self, in_channels, ch1x1, ch3x3red, ch3x3, ch5x5red, ch5x5, pool_proj):
        super(Inception, self).__init__()
        self.branch1 = self.BasicConv(in_channels, ch1x1, kernel_size=1)
        self.branch2 = nn.Sequential(
            self.BasicConv(in_channels, ch3x3red, kernel_size=1),
            self.BasicConv(ch3x3red, ch3x3, kernel_size=3, padding=1)
        )
        self.branch3 = nn.Sequential(
            self.BasicConv(in_channels, ch5x5red, kernel_size=1),
            self.BasicConv(ch5x5red, ch5x5, kernel_size=3, padding=1)
        )
        self.branch4 = nn.Sequential(
            nn.MaxPool2d(kernel_size=3, stride=1, padding=1, ceil_mode=True),
            self.BasicConv(in_channels, pool_proj, kernel_size=1)
        )

    def BasicConv(self, in_channels, out_channels, **kwargs):
        return nn.Sequential(nn.Conv2d(in_channels, out_channels, bias=False, **kwargs),
                             nn.BatchNorm2d(out_channels, eps=0.001),
                             nn.ReLU(inplace=True),)

    def forward(self, x):
        branch1 = self.branch1(x)
        branch2 = self.branch2(x)
        branch3 = self.branch3(x)
        branch4 = self.branch4(x)

        outputs = [branch1, branch2, branch3, branch4]
        return torch.cat(outputs, dim=1)

class GoogLeNet(nn.Module):
    def __init__(self, num_classes=1000, init_weights=False):
        super(GoogLeNet, self).__init__()
        # 可参考 https://arxiv.org/pdf/1409.4842.pdf 第七页结构图和第六页Table 1
        self.conv1 = self.BasicConv(3, 64, kernel_size=7, stride=2, padding=3)
        self.maxpool1 = nn.MaxPool2d(3, stride=2, ceil_mode=True)
        self.conv2 = self.BasicConv(64, 64, kernel_size=1, stride=1)
        self.conv3 = self.BasicConv(64, 192, kernel_size=3, padding=1)
        self.maxpool2 = nn.MaxPool2d(3, stride=2, ceil_mode=True)

        self.inception3a = Inception(192, 64, 96, 128, 16, 32, 32)
        self.inception3b = Inception(256, 128, 128, 192, 32, 96, 64)
        self.maxpool3 = nn.MaxPool2d(3, stride=2, ceil_mode=True)

        self.inception4a = Inception(480, 192, 96, 208, 16, 48, 64)
        self.inception4b = Inception(512, 160, 112, 224, 24, 64, 64)
        self.inception4c = Inception(512, 128, 128, 256, 24, 64, 64)
        self.inception4d = Inception(512, 112, 144, 288, 32, 64, 64)
        self.inception4e = Inception(528, 256, 160, 320, 32, 128, 128)
        self.maxpool4 = nn.MaxPool2d(2, stride=2, ceil_mode=True)

        self.inception5a = Inception(832, 256, 160, 320, 32, 128, 128)
        self.inception5b = Inception(832, 384, 192, 384, 48, 128, 128)

        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.dropout = nn.Dropout(0.2)
        self.fc = nn.Linear(1024, num_classes)

        if init_weights:
            self._initialize_weights()

    def _initialize_weights(self):
        for m in self.modules():
            if isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear):
                import scipy.stats as stats
                X = stats.truncnorm(-2, 2, scale=0.01)
                values = torch.as_tensor(X.rvs(m.weight.numel()), dtype=m.weight.dtype)
                values = values.view(m.weight.size())
                with torch.no_grad():
                    m.weight.copy_(values)
            elif isinstance(m, nn.BatchNorm2d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

    def BasicConv(self, in_channels, out_channels, **kwargs):
        return nn.Sequential(nn.Conv2d(in_channels, out_channels, bias=False, **kwargs),
                             nn.BatchNorm2d(out_channels, eps=0.001),
                             nn.ReLU(inplace=True),)

    def forward(self, x):
        # N x 3 x 224 x 224
        x = self.conv1(x)
        # N x 64 x 112 x 112
        x = self.maxpool1(x)
        # N x 64 x 56 x 56
        x = self.conv2(x)
        # N x 64 x 56 x 56
        x = self.conv3(x)
        # N x 192 x 56 x 56
        x = self.maxpool2(x)
        # N x 192 x 28 x 28
        x = self.inception3a(x)
        # N x 256 x 28 x 28
        x = self.inception3b(x)
        # N x 480 x 28 x 28
        x = self.maxpool3(x)
        # N x 480 x 14 x 14
        x = self.inception4a(x)
        # N x 512 x 14 x 14
        x = self.inception4b(x)
        # N x 512 x 14 x 14
        x = self.inception4c(x)
        # N x 512 x 14 x 14
        x = self.inception4d(x)
        # N x 528 x 14 x 14
        x = self.inception4e(x)
        # N x 832 x 14 x 14
        x = self.maxpool4(x)
        # N x 832 x 7 x 7
        x = self.inception5a(x)
        # N x 832 x 7 x 7
        x = self.inception5b(x)
        # N x 1024 x 7 x 7
        x = self.avgpool(x)
        # N x 1024 x 1 x 1
        x = torch.flatten(x, 1)
        # N x 1024
        x = self.dropout(x)
        x = self.fc(x)
        # N x 1000 (num_classes)
        return x


if __name__=="__main__":
    import torchsummary
    input = torch.ones(2, 3, 224, 224).cpu()
    net = GoogLeNet(num_classes=4)
    net = net.cpu()
    out = net(input)
    print(out)
    print(out.shape)
    torchsummary.summary(net, input_size=(3, 224, 224))
    # Total params: 5,604,004

通过使用不同大小的卷积核，模型实现了对不同尺度感受野的提取，这样能够捕获到局部与全局的信息，而且1，3，5这样的选择也方便进行对齐，在网络的后期，为了减少计算量，使用了 1x1 的卷积核来进行降维。这一设计在保留特征信息的同时，有效地降低了计算复杂度。