Pytorch实战：用经典网络实现猫狗大战

最新推荐文章于 2024-07-07 16:20:37 发布

BlackdogC

最新推荐文章于 2024-07-07 16:20:37 发布

阅读量1.7k

点赞数 3

文章标签： python pytorch 深度学习计算机视觉数据库

本文链接：https://blog.csdn.net/m0_63567560/article/details/125913787

版权

该博客介绍了如何使用PyTorch构建一个简单的InceptionA网络进行猫狗图像分类任务。数据集来源于AI研习社，网络结构包含卷积层和InceptionA模块，用于特征提取。训练和验证模型的流程清晰，包括数据预处理、模型构建、损失函数定义、优化器设置以及训练和验证过程。最后，展示了训练过程中损失函数的变化以及验证集上的精度。

摘要由CSDN通过智能技术生成

一：数据集准备：

数据集来源于研习社：猫狗大战--经典图像分类题 - AI算法竞赛-AI研习社 (yanxishe.com)

二：读取数据：

class Cat_and_Dog_Dataset(Dataset):
    def __init__(self, filepath):
        self.images = []
        self.labels = []
        self.transform = transform
        for filename in tqdm(os.listdir(filepath)):
            image = Image.open(filepath + filename)
            image = image.resize((224,224))         #裁剪图片
            image = self.transform(image)           #转化为Tensor格式
            self.images.append(image)
            if filename.split('_')[0] == 'cat':     #添加标签
                self.labels.append(0)
            elif filename.split('_')[0] == 'dog':
                self.labels.append(1)
        self.labels = torch.LongTensor(self.labels) #标签转化位Tensor格式
        print(self.labels)
        
    
    def __getitem__(self, index):           #构造迭代器
        return self.images[index], self.labels[index]

    def __len__(self):                      #迭代器长度
        images = np.array(self.images)
        len = images.shape[0]
        return len

train_data = Cat_and_Dog_Dataset('cat_dog/train/')              #加载训练集
train_loader = DataLoader(dataset=train_data, batch_size=64, shuffle=True)  


val_data = Cat_and_Dog_Dataset('cat_dog/val/')                  #加载验证集
val_loader = DataLoader(dataset=val_data, batch_size=64, shuffle=True)

二：搭建网络：

class InceptionA(torch.nn.Module):                              #构造Inception层
    def __init__(self,in_channels):
        super(InceptionA,self).__init__()
        self.branch1x1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)

        self.branch5x5_1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)
        self.branch5x5_2 = torch.nn.Conv2d(16,24,kernel_size = 5,padding = 2)

        self.branch3x3_1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)
        self.branch3x3_2 = torch.nn.Conv2d(16,24,kernel_size = 3,padding = 1)
        self.branch3x3_3 = torch.nn.Conv2d(24,24,kernel_size = 3,padding = 1)

        self.branch_pool = torch.nn.Conv2d(in_channels,24,kernel_size = 1)


    def forward(self,x):
        branch1x1 = self.branch1x1(x)

        branch5x5 = self.branch5x5_1(x)
        branch5x5 = self.branch5x5_2(branch5x5)

        branch3x3 = self.branch3x3_1(x)
        branch3x3 = self.branch3x3_2(branch3x3)
        branch3x3 = self.branch3x3_3(branch3x3)

        branch_pool = F.avg_pool2d(x,kernel_size = 3,stride = 1,padding = 1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1,branch3x3,branch5x5,branch_pool]
        return torch.cat(outputs,dim = 1)

class Net(torch.nn.Module):                             #构造网络
    def __init__(self):
        super(Net,self).__init__()
        self.conv1 = torch.nn.Conv2d(3,10,kernel_size = 5)
        self.conv2 = torch.nn.Conv2d(88,20,kernel_size = 5)

        self.incep1 = InceptionA(in_channels = 10)
        self.incep2 = InceptionA(in_channels = 20)

        self.mp = torch.nn.MaxPool2d(2)

        self.fc = torch.nn.Linear(247192,2)

    def forward(self,x):
        in_size = x.size(0)
        x = self.mp(F.relu((self.conv1(x))))
        x = self.incep1(x)
        x = self.mp(F.relu((self.conv2(x))))
        x = self.incep2(x)
        x = x.view(in_size,-1)
        x = self.fc(x)


        return x

三：构建损失函数和优化器：

criterion = torch.nn.CrossEntropyLoss()         #构造损失函数
optimizer = optim.SGD(model.parameters(),lr = 0.001)        #构造优化器

四：训练模型：

def train(epoch):
    running_loss = 0.0                              #训练模型
    for batch_idx,data in enumerate(train_loader,0):
        inputs,targets = data
        inputs, targets = inputs.to(device), targets.to(device)
        optimizer.zero_grad()                       #梯度清零

        outputs = model(inputs)                     #前向传播
        loss = criterion(outputs,targets)           #计算损失
        loss.backward()                             #逆向传播
        optimizer.step()                            #梯度递进
        running_loss += loss.item()

    print('train loss: %.3f' % (running_loss/batch_idx))

五：验证模型：

def val():                                          #验证模型精度
    correct = 0
    total = 0
    with torch.no_grad():                           #不需要梯度，减少计算量
        for data in val_loader:
            images, labels = data
            images, labels = images.to(device), labels.to(device)
            outputs = model(images)
            _, predicted = torch.max(outputs.data, dim=1)
            total += labels.size(0)
            correct += (predicted == labels).sum().item()
    print('accuracy on test set: %d %% ' % (100*correct/total))
    return correct/total

源码：

import torch
import numpy as np
from torch.utils.data import DataLoader,Dataset
from torchvision import transforms
import torchvision
import torch.nn.functional as F
import torch.optim as optim
import os
from tqdm import tqdm
from PIL import Image
import matplotlib.pyplot as plt

batch_size = 64

transform = transforms.Compose([transforms.ToTensor(),
                                transforms.Normalize((0.1307,), (0.3081,))])#设置transform

class Cat_and_Dog_Dataset(Dataset):
    def __init__(self, filepath):
        self.images = []
        self.labels = []
        self.transform = transform
        for filename in tqdm(os.listdir(filepath)):
            image = Image.open(filepath + filename)
            image = image.resize((224,224))         #裁剪图片
            image = self.transform(image)           #转化为Tensor格式
            self.images.append(image)
            if filename.split('_')[0] == 'cat':     #添加标签
                self.labels.append(0)
            elif filename.split('_')[0] == 'dog':
                self.labels.append(1)
        self.labels = torch.LongTensor(self.labels) #标签转化位Tensor格式
        print(self.labels)
        
    
    def __getitem__(self, index):           #构造迭代器
        return self.images[index], self.labels[index]

    def __len__(self):                      #迭代器长度
        images = np.array(self.images)
        len = images.shape[0]
        return len

train_data = Cat_and_Dog_Dataset('cat_dog/train/')              #加载训练集
train_loader = DataLoader(dataset=train_data, batch_size=64, shuffle=True)  


val_data = Cat_and_Dog_Dataset('cat_dog/val/')                  #加载验证集
val_loader = DataLoader(dataset=val_data, batch_size=64, shuffle=True)



class InceptionA(torch.nn.Module):                              #构造Inception层
    def __init__(self,in_channels):
        super(InceptionA,self).__init__()
        self.branch1x1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)

        self.branch5x5_1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)
        self.branch5x5_2 = torch.nn.Conv2d(16,24,kernel_size = 5,padding = 2)

        self.branch3x3_1 = torch.nn.Conv2d(in_channels,16,kernel_size = 1)
        self.branch3x3_2 = torch.nn.Conv2d(16,24,kernel_size = 3,padding = 1)
        self.branch3x3_3 = torch.nn.Conv2d(24,24,kernel_size = 3,padding = 1)

        self.branch_pool = torch.nn.Conv2d(in_channels,24,kernel_size = 1)


    def forward(self,x):
        branch1x1 = self.branch1x1(x)

        branch5x5 = self.branch5x5_1(x)
        branch5x5 = self.branch5x5_2(branch5x5)

        branch3x3 = self.branch3x3_1(x)
        branch3x3 = self.branch3x3_2(branch3x3)
        branch3x3 = self.branch3x3_3(branch3x3)

        branch_pool = F.avg_pool2d(x,kernel_size = 3,stride = 1,padding = 1)
        branch_pool = self.branch_pool(branch_pool)

        outputs = [branch1x1,branch3x3,branch5x5,branch_pool]
        return torch.cat(outputs,dim = 1)

class Net(torch.nn.Module):                             #构造网络
    def __init__(self):
        super(Net,self).__init__()
        self.conv1 = torch.nn.Conv2d(3,10,kernel_size = 5)
        self.conv2 = torch.nn.Conv2d(88,20,kernel_size = 5)

        self.incep1 = InceptionA(in_channels = 10)
        self.incep2 = InceptionA(in_channels = 20)

        self.mp = torch.nn.MaxPool2d(2)

        self.fc = torch.nn.Linear(247192,2)

    def forward(self,x):
        in_size = x.size(0)
        x = self.mp(F.relu((self.conv1(x))))
        x = self.incep1(x)
        x = self.mp(F.relu((self.conv2(x))))
        x = self.incep2(x)
        x = x.view(in_size,-1)
        x = self.fc(x)


        return x


model = Net()

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)            #转换为cuda格式

criterion = torch.nn.CrossEntropyLoss()         #构造损失函数
optimizer = optim.SGD(model.parameters(),lr = 0.001)        #构造优化器

def train(epoch):
    running_loss = 0.0                              #训练模型
    for batch_idx,data in enumerate(train_loader,0):
        inputs,targets = data
        inputs, targets = inputs.to(device), targets.to(device)
        optimizer.zero_grad()                       #梯度清零

        outputs = model(inputs)                     #前向传播
        loss = criterion(outputs,targets)           #计算损失
        loss.backward()                             #逆向传播
        optimizer.step()                            #梯度递进
        running_loss += loss.item()

    print('train loss: %.3f' % (running_loss/batch_idx))


def val():                                          #验证模型精度
    correct = 0
    total = 0
    with torch.no_grad():                           #不需要梯度，减少计算量
        for data in val_loader:
            images, labels = data
            images, labels = images.to(device), labels.to(device)
            outputs = model(images)
            _, predicted = torch.max(outputs.data, dim=1)
            total += labels.size(0)
            correct += (predicted == labels).sum().item()
    print('accuracy on test set: %d %% ' % (100*correct/total))
    return correct/total


if __name__ == '__main__':
    accuracy_list = []
    epoch_list = []

    for epoch in range(10):
        train(epoch)
        acc = val()
        accuracy_list.append(acc)
        epoch_list.append(epoch)

    plt.plot(epoch_list,accuracy_list)
    plt.xlabel(epoch)
    plt.ylabel(accuracy_list)
    plt.show()