P3周：天气识别

千筱夜

已于 2023-09-29 19:15:59 修改

阅读量94

点赞数

文章标签：深度学习神经网络

于 2023-09-22 20:05:18 首次发布

本文链接：https://blog.csdn.net/geo436872/article/details/133183811

版权

🍨 本文为🔗365天深度学习训练营中的学习记录博客
🍖 原作者：K同学啊 | 接辅导、项目定制
🚀 文章来源：K同学的学习圈子

环境配置：
Python 3.11.4

Pytorch 2.0.1

torchvision 0.15.2+cpu

本次天气图像数据集为K同学提供，若需获取，请联系K同学。

一、前期准备

1.导入所需包并设置GPU


import torch
import torch.nn as nn
import matplotlib.pyplot as plt
import torch.nn.functional as F
import torchvision.transforms as transforms
from torchvision import datasets

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
device

查看GPU资源是否可获取，并打印可获取资源。

2.导入所需数据

将下载好的数据集进行整理，将weather_photos放于data文件夹下方便读取。该数据集为K同学提供，内含1125张图片，内分为四类：cloudy, rain, shine, sunrise；其图片数量分别为300，215，253，357。

在这里插入图片描述


import os, PIL, random, pathlib
data_dir = './data/weather_photos'
data_dir = pathlib.Path(data_dir)
data_paths = list(data_dir.glob('*'))
classNames = [str(path).split("\\")[1] for path in data_paths] #使用字符串分割函数以获取类名
print(classNames)

3.划分数据集


total_datadir = './data/weather_photos'
train_transforms = transforms.Compose([transforms.Resize([224,224]),
                                      transforms.ToTensor(),
                                      transforms.Normalize(
                                          mean=[0.485,0.456,0.406],
                                          std = [0.229, 0.224, 0.225])])
total_data = datasets.ImageFolder(total_datadir, transform=train_transforms)
print(total_data)
train_size = int(0.8*len(total_data))
test_size = len(total_data) - train_size
train_dataset, test_dataset = torch.utils.data.random_split(total_data, [train_size, test_size])


Dataset ImageFolder
    Number of datapoints: 1125
    Root location: ./data/weather_photos
    StandardTransform
Transform: Compose(
               Resize(size=[224, 224], interpolation=bilinear, max_size=None, antialias=warn)
               ToTensor()
               Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
           )

将图片重调尺寸，进行批量归一化，将训练集与测试集比例划分为0.8:0.2。


batch_size = 32
train_dl = torch.utils.data.DataLoader(train_dataset,
                                       batch_size=batch_size,
                                       shuffle=True,
                                       num_workers=0)
test_dl = torch.utils.data.DataLoader(test_dataset,
                                      batch_size=batch_size,
                                      shuffle = True,
                                      num_workers=0)

设置batch_size大小，并使用DataLoader封装。

二、模型构建

1.搭建网络模型

在这里插入图片描述
此图为K同学原创，引用仅为直观展示网络模型。


class Model(nn.Module):
    def __init__(self):
        super(Model, self).__init__()
        self.conv1 = nn.Conv2d(3, 12, kernel_size=5, stride=1, padding=0)
        self.bn1 = nn.BatchNorm2d(12)
        self.conv2 = nn.Conv2d(12, 12, kernel_size=5, stride=1, padding=0)
        self.bn2 = nn.BatchNorm2d(12)
        self.pool2 = nn.MaxPool2d(2, 2)
        self.conv3 = nn.Conv2d(12, 24, kernel_size=5, stride=1, padding=0)
        self.bn3 = nn.BatchNorm2d(24)
        self.conv4 = nn.Conv2d(24, 24, kernel_size=5, stride=1, padding=0)
        self.bn4 = nn.BatchNorm2d(24)
        self.pool4 = nn.MaxPool2d(2, 2)
        self.fc1 = nn.Linear(24 * 50 * 50, len(classNames))

    def forward(self, x):
        x = F.relu(self.bn1(self.conv1(x)))
        x = F.relu(self.bn2(self.conv2(x)))
        x = self.pool2(x)
        x = F.relu(self.bn3(self.conv3(x)))
        x = F.relu(self.bn4(self.conv4(x)))
        x = self.pool4(x)
        x = x.view(-1, 24 * 50 * 50)
        x = self.fc1(x)
        return x

使用summary打印出模型框架，计算参数量。


from torchsummary import summary
model = Model().to(device)
summary(model, input_size=(3, 224, 224))


----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1         [-1, 12, 220, 220]             912
       BatchNorm2d-2         [-1, 12, 220, 220]              24
            Conv2d-3         [-1, 12, 216, 216]           3,612
       BatchNorm2d-4         [-1, 12, 216, 216]              24
         MaxPool2d-5         [-1, 12, 108, 108]               0
            Conv2d-6         [-1, 24, 104, 104]           7,224
       BatchNorm2d-7         [-1, 24, 104, 104]              48
            Conv2d-8         [-1, 24, 100, 100]          14,424
       BatchNorm2d-9         [-1, 24, 100, 100]              48
        MaxPool2d-10           [-1, 24, 50, 50]               0
           Linear-11                    [-1, 4]         240,004
================================================================
Total params: 266,320
Trainable params: 266,320
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.57
Forward/backward pass size (MB): 26.55
Params size (MB): 1.02
Estimated Total Size (MB): 28.14
----------------------------------------------------------------

2.设置超参数


loss_fn = nn.CrossEntropyLoss()
learn_rate = 0.0001
opt = torch.optim.SGD(model.parameters(), lr=learn_rate)

引入交叉熵损失函数，设置学习率，优化函数，防止过拟合。

3.测试与训练


def train(dataloader, model, loss_fn, opt):
    size = len(dataloader.dataset)
    num_batches = len(dataloader)
    train_acc, train_loss = 0, 0

    for X, y in dataloader:
        X, y = X.to(device), y.to(device)
        pred = model(X)
        loss = loss_fn(pred, y)

        opt.zero_grad()
        loss.backward()
        opt.step()

        train_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
        train_loss += loss.item()

    train_acc /= size
    train_loss /= num_batches
    return train_acc, train_loss

当不进行训练时，停止梯度更新，节省计算内存消耗。


def test(dataloader, model, loss_fn):
    size = len(dataloader.dataset)
    num_batches = len(dataloader)
    test_acc, test_loss = 0, 0
    with torch.no_grad():
        for X, y in dataloader:
            X, y = X.to(device), y.to(device)
            pred = model(X)
            loss = loss_fn(pred, y)

            test_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
            test_loss += loss.item()

    test_acc /= size
    test_loss /= num_batches
    return test_acc, test_loss

4.主函数

将epochs设为50，打出每轮训练与测试的精度与损失。


epochs = 50
train_loss, train_acc, test_loss, test_acc = [], [], [], []
for epoch in range(epochs):
    model.train()
    epoch_train_acc, epoch_train_loss = train(train_dl, model, loss_fn, opt)
    model.eval()
    epoch_test_acc, epoch_test_loss = test(test_dl, model, loss_fn)
    train_acc.append(epoch_train_acc)
    train_loss.append(epoch_train_loss)
    test_acc.append(epoch_test_acc)
    test_loss.append(epoch_test_loss)
    template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}%, Test_acc:{:.1f}%, Test_loss:{:.3f}%,')
    print(template.format(epoch+1, epoch_train_acc*100, epoch_train_loss, epoch_test_acc*100, epoch_test_loss))
print('Done')


Epoch: 1, Train_acc:53.0%, Train_loss:1.093%, Test_acc:69.3%, Test_loss:1.062%,
Epoch: 2, Train_acc:79.1%, Train_loss:0.680%, Test_acc:79.6%, Test_loss:0.675%,
Epoch: 3, Train_acc:83.6%, Train_loss:0.573%, Test_acc:84.9%, Test_loss:0.611%,
Epoch: 4, Train_acc:85.2%, Train_loss:0.521%, Test_acc:88.4%, Test_loss:0.485%,
Epoch: 5, Train_acc:88.2%, Train_loss:0.430%, Test_acc:85.8%, Test_loss:0.505%,
Epoch: 6, Train_acc:88.1%, Train_loss:0.395%, Test_acc:84.4%, Test_loss:0.403%,
Epoch: 7, Train_acc:89.8%, Train_loss:0.372%, Test_acc:83.1%, Test_loss:0.458%,
Epoch: 8, Train_acc:90.8%, Train_loss:0.379%, Test_acc:88.0%, Test_loss:0.376%,
Epoch: 9, Train_acc:92.1%, Train_loss:0.323%, Test_acc:87.1%, Test_loss:0.422%,
Epoch:10, Train_acc:90.6%, Train_loss:0.310%, Test_acc:87.1%, Test_loss:0.378%,
Epoch:11, Train_acc:91.6%, Train_loss:0.279%, Test_acc:89.8%, Test_loss:0.336%,
Epoch:12, Train_acc:92.1%, Train_loss:0.266%, Test_acc:90.2%, Test_loss:0.327%,
Epoch:13, Train_acc:92.9%, Train_loss:0.294%, Test_acc:85.8%, Test_loss:0.429%,
Epoch:14, Train_acc:92.7%, Train_loss:0.263%, Test_acc:91.6%, Test_loss:0.314%,
Epoch:15, Train_acc:95.0%, Train_loss:0.222%, Test_acc:93.8%, Test_loss:0.311%,
Epoch:16, Train_acc:93.8%, Train_loss:0.233%, Test_acc:88.9%, Test_loss:0.357%,
Epoch:17, Train_acc:94.7%, Train_loss:0.201%, Test_acc:92.4%, Test_loss:0.286%,
Epoch:18, Train_acc:94.7%, Train_loss:0.199%, Test_acc:88.9%, Test_loss:0.311%,
Epoch:19, Train_acc:93.6%, Train_loss:0.198%, Test_acc:92.0%, Test_loss:0.341%,
Epoch:20, Train_acc:93.3%, Train_loss:0.210%, Test_acc:89.8%, Test_loss:0.324%,
Epoch:21, Train_acc:95.7%, Train_loss:0.184%, Test_acc:87.6%, Test_loss:0.322%,
Epoch:22, Train_acc:95.0%, Train_loss:0.201%, Test_acc:89.8%, Test_loss:0.316%,
Epoch:23, Train_acc:94.9%, Train_loss:0.196%, Test_acc:91.6%, Test_loss:0.338%,
Epoch:24, Train_acc:95.3%, Train_loss:0.184%, Test_acc:92.4%, Test_loss:0.281%,
Epoch:25, Train_acc:96.1%, Train_loss:0.174%, Test_acc:91.6%, Test_loss:0.277%,
Epoch:26, Train_acc:95.2%, Train_loss:0.186%, Test_acc:92.4%, Test_loss:0.264%,
Epoch:27, Train_acc:95.9%, Train_loss:0.169%, Test_acc:92.9%, Test_loss:0.281%,
Epoch:28, Train_acc:97.0%, Train_loss:0.146%, Test_acc:87.6%, Test_loss:0.490%,
Epoch:29, Train_acc:97.2%, Train_loss:0.141%, Test_acc:90.7%, Test_loss:0.284%,
Epoch:30, Train_acc:96.9%, Train_loss:0.125%, Test_acc:91.6%, Test_loss:0.260%,
Epoch:31, Train_acc:97.0%, Train_loss:0.129%, Test_acc:92.4%, Test_loss:0.278%,
Epoch:32, Train_acc:95.9%, Train_loss:0.172%, Test_acc:80.0%, Test_loss:0.460%,
Epoch:33, Train_acc:96.3%, Train_loss:0.160%, Test_acc:91.1%, Test_loss:0.268%,
Epoch:34, Train_acc:96.4%, Train_loss:0.142%, Test_acc:90.2%, Test_loss:0.261%,
Epoch:35, Train_acc:97.3%, Train_loss:0.128%, Test_acc:92.9%, Test_loss:0.252%,
Epoch:36, Train_acc:95.9%, Train_loss:0.159%, Test_acc:88.9%, Test_loss:0.353%,
Epoch:37, Train_acc:97.3%, Train_loss:0.117%, Test_acc:92.4%, Test_loss:0.345%,
Epoch:38, Train_acc:98.3%, Train_loss:0.104%, Test_acc:91.1%, Test_loss:0.251%,
Epoch:39, Train_acc:98.1%, Train_loss:0.115%, Test_acc:85.3%, Test_loss:0.350%,
Epoch:40, Train_acc:98.1%, Train_loss:0.112%, Test_acc:89.3%, Test_loss:0.261%,
Epoch:41, Train_acc:97.8%, Train_loss:0.099%, Test_acc:90.2%, Test_loss:0.303%,
Epoch:42, Train_acc:97.2%, Train_loss:0.124%, Test_acc:88.9%, Test_loss:0.400%,
Epoch:43, Train_acc:98.0%, Train_loss:0.102%, Test_acc:90.7%, Test_loss:0.240%,
Epoch:44, Train_acc:98.0%, Train_loss:0.099%, Test_acc:92.4%, Test_loss:0.253%,
Epoch:45, Train_acc:97.9%, Train_loss:0.091%, Test_acc:92.0%, Test_loss:0.257%,
Epoch:46, Train_acc:98.2%, Train_loss:0.109%, Test_acc:88.0%, Test_loss:0.589%,
Epoch:47, Train_acc:98.0%, Train_loss:0.094%, Test_acc:91.6%, Test_loss:0.246%,
Epoch:48, Train_acc:97.9%, Train_loss:0.098%, Test_acc:91.1%, Test_loss:0.258%,
Epoch:49, Train_acc:98.4%, Train_loss:0.097%, Test_acc:92.4%, Test_loss:0.292%,
Epoch:50, Train_acc:98.1%, Train_loss:0.091%, Test_acc:92.9%, Test_loss:0.229%,
Done

5.可视化


import warnings
warnings.filterwarnings("ignore")               #忽略警告信息
plt.rcParams['font.sans-serif']    = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False      # 用来正常显示负号
plt.rcParams['figure.dpi']         = 100        #分辨率

epochs_range = range(epochs)

plt.figure(figsize=(12, 3))
plt.subplot(1, 2, 1)

plt.plot(epochs_range, train_acc, label='Training Accuracy')
plt.plot(epochs_range, test_acc, label='Test Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Training Loss')
plt.plot(epochs_range, test_loss, label='Test Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()