大家好呀，我是小黑喵

核心：

1.1*1的kernel_size能夠有效減少參數(shù)數(shù)量
2.Inception block

Inception結(jié)構(gòu)

Inception Module

網(wǎng)絡(luò)結(jié)構(gòu)

Googlenet

忽略輔助分類：使用CIFAR10數(shù)據(jù)集

代碼：

###author:xiaoheimiao
import torch
import torchvision
import matplotlib.pyplot as plt
import torch.nn as nn
from torch.utils.data import DataLoader
from torchvision.transforms import transforms
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

#定義conv-bn-relu函數(shù)
def conv_relu(in_channel, out_channel, kernel, stride=1, padding=0):
    conv = nn.Sequential(
        nn.Conv2d(in_channel, out_channel, kernel, stride, padding),
        nn.BatchNorm2d(out_channel, eps=1e-3),
        nn.ReLU(True),
    )
    return conv

#定義incepion結(jié)構(gòu)，見inception圖
class inception(nn.Module):
    def __init__(self, in_channel, out1_1, out2_1, out2_3, out3_1, out3_5,
                 out4_1):
        super(inception, self).__init__()
        self.branch1 = conv_relu(in_channel, out1_1, 1)
        self.branch2 = nn.Sequential(
            conv_relu(in_channel, out2_1, 1),
            conv_relu(out2_1, out2_3, 3, padding=1))
        self.branch3 = nn.Sequential(
            conv_relu(in_channel, out3_1, 1),
            conv_relu(out3_1, out3_5, 5, padding=2))
        self.branch4 = nn.Sequential(
            nn.MaxPool2d(3, stride=1, padding=1),
            conv_relu(in_channel, out4_1, 1),
        )

    def forward(self, x):
        b1 = self.branch1(x)
        b2 = self.branch2(x)
        b3 = self.branch3(x)
        b4 = self.branch4(x)
        output = torch.cat([b1, b2, b3, b4], dim=1)
        return output

# 堆疊GOOGLENET，見上表所示結(jié)構(gòu)
class GOOGLENET(nn.Module):
    def __init__(self):
        super(GOOGLENET, self).__init__()
        self.features = nn.Sequential(
            conv_relu(3, 64, 7, 2, 3), nn.MaxPool2d(3, stride=2, padding=0),
            conv_relu(64, 64, 1), conv_relu(64, 192, 3, padding=1),
            nn.MaxPool2d(3, 2), inception(192, 64, 96, 128, 16, 32, 32),
            inception(256, 128, 128, 192, 32, 96, 64), nn.MaxPool2d(
                3, stride=2), inception(480, 192, 96, 208, 16, 48, 64),
            inception(512, 160, 112, 224, 24, 64, 64),
            inception(512, 128, 128, 256, 24, 64, 64),
            inception(512, 112, 144, 288, 32, 64, 64),
            inception(528, 256, 160, 320, 32, 128, 128), nn.MaxPool2d(3, 2),
            inception(832, 256, 160, 320, 32, 128, 128),
            inception(832, 384, 182, 384, 48, 128, 128), nn.AvgPool2d(2))
        self.classifier = nn.Sequential(
            nn.Linear(9216,1024),
            nn.Dropout2d(p=0.4),
            nn.Linear(1024, 10))

    def forward(self, x):
        x = self.features(x)
        x = x.view(x.size(0), -1)
        out = self.classifier(x)
        return out

#訓(xùn)練函數(shù)
def net_train():
    net.train()
    running_loss = 0.0
    for i, data in enumerate(trainloader, 0):
        # 將輸入傳入GPU
        inputs, labels = data
        inputs, labels = inputs.to(device), labels.to(device)

        # 將梯度置零
        optimizer.zero_grad()

        # 前向傳播-計算誤差-反向傳播-優(yōu)化
        outputs = net(inputs)
        loss = criterion(outputs, labels)
        loss.backward()
        optimizer.step()

        # 計算誤差并顯示
        running_loss += loss.item()
        if i % 60 == 0:    # print every 60 mini-batches
            print('[%d, %5d] loss: %.3f' %
                  (epoch + 1, i + 1, running_loss / 60))
            running_loss = 0.0

    print('Training Epoch Finished')

#測試函數(shù)
def net_test():
    correct = 0
    total = 0
    # 關(guān)閉梯度
    with torch.no_grad():
        for data in testloader:
            images, labels = data
            images, labels = images.to(device), labels.to(device)
            outputs = net(images)
            _, predicted = torch.max(outputs.data, 1)
            total += labels.size(0)
            correct += (predicted == labels).sum().item()

    print('Accuracy of the network on the 10000 test images: %d %%' % (100 * correct / total))
    return

#數(shù)據(jù)集函數(shù)
def net_dataloader(root, train_transform, test_transform):
    trainset = torchvision.datasets.CIFAR10(
        root, train=True, transform=train_transform, download=False)
    testset = torchvision.datasets.CIFAR10(
        root, train=False, transform=test_transform, download=False)
    trainloader = DataLoader(
        trainset, batch_size=60, shuffle=True, num_workers=2)
    testloader = DataLoader(
        testset, batch_size=8, shuffle=False, num_workers=2)
    print('Initializing Dataset...')
    return trainloader, testloader

#main
if __name__ == "__main__":
    # 創(chuàng)建實例并送入GPU
    net = GOOGLENET().to(device)
    # 選擇誤差
    criterion = nn.CrossEntropyLoss()
    # 選擇優(yōu)化器
    optimizer = torch.optim.Adam(net.parameters(), lr=1e-3)
    # 數(shù)據(jù)位置
    root = './pydata/data/'
    # 數(shù)據(jù)處理
    train_transform = transforms.Compose([
        transforms.RandomResizedCrop(224),
        transforms.RandomHorizontalFlip(),
        transforms.ToTensor(),
        transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
    ])
    test_transform = transforms.Compose([
        transforms.Resize(224),
        transforms.ToTensor(),
        transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
    ])
    # 創(chuàng)建數(shù)據(jù)loader
    trainloader, testloader = net_dataloader(root, train_transform,
                                             test_transform)
    # run
    n_epoch = 5#改變epoch
    for epoch in range(n_epoch):
        net_train()#每個epoch訓(xùn)練一次，測試一次
        net_test()

喵喵喵