深度学习实验--第P4周：猴痘病识别

原创已于 2025-07-03 08:21:36 修改 · 265 阅读

5 ·

CC 4.0 BY-SA版权

文章标签：

#深度学习 #人工智能 #pytorch

于 2025-06-23 16:37:50 首次发布

AI学习专栏收录该内容

6 篇文章

订阅专栏

🍨 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
我的环境：

语言环境：Python3.12
编译器：PyCharm
深度学习环境：
torch==1.12.1+cu113
torchvision==0.13.1+cu113

一、实验

1、目的：

        训练过程中保存效果最好的模型参数。
        加载最佳模型参数识别本地的一张图片。
        调整网络结构，添加池化层2，使用动态学习率，观察差异。

2、总结：

训练好的模型通过torch.save函数进行保存；

保存后模型可通过load加载模型，加载完成后可使用模型识别功能；目前还只是掌握了模型的基本使用方法。

调整网络结构，添加池化2后，起始训练结果比未池化2的高3%左右，但20次最后准确率却下降了，再调整动态学习率后，最后准确率达到最高92%。

对比调整网络结构，未添加池化2时，准确率如下：

Epoch: 1, Train_acc:57.6%, Train_loss:0.704, Test_acc:63.9%，Test_loss:0.643
Epoch: 2, Train_acc:68.9%, Train_loss:0.580, Test_acc:70.6%，Test_loss:0.561
Epoch: 3, Train_acc:74.3%, Train_loss:0.525, Test_acc:74.4%，Test_loss:0.535
Epoch: 4, Train_acc:77.2%, Train_loss:0.484, Test_acc:73.9%，Test_loss:0.517
Epoch: 5, Train_acc:79.8%, Train_loss:0.452, Test_acc:74.6%，Test_loss:0.516
Epoch: 6, Train_acc:83.4%, Train_loss:0.420, Test_acc:76.2%，Test_loss:0.477
Epoch: 7, Train_acc:84.6%, Train_loss:0.401, Test_acc:77.2%，Test_loss:0.492
Epoch: 8, Train_acc:85.6%, Train_loss:0.376, Test_acc:78.3%，Test_loss:0.453
Epoch: 9, Train_acc:86.7%, Train_loss:0.363, Test_acc:79.5%，Test_loss:0.442
Epoch:10, Train_acc:87.6%, Train_loss:0.350, Test_acc:74.8%，Test_loss:0.521
Epoch:11, Train_acc:88.1%, Train_loss:0.336, Test_acc:80.4%，Test_loss:0.419
Epoch:12, Train_acc:90.3%, Train_loss:0.314, Test_acc:82.3%，Test_loss:0.404
Epoch:13, Train_acc:90.8%, Train_loss:0.295, Test_acc:82.3%，Test_loss:0.413
Epoch:14, Train_acc:91.4%, Train_loss:0.285, Test_acc:82.5%，Test_loss:0.405
Epoch:15, Train_acc:90.5%, Train_loss:0.280, Test_acc:82.8%，Test_loss:0.381
Epoch:16, Train_acc:91.3%, Train_loss:0.273, Test_acc:81.6%，Test_loss:0.384
Epoch:17, Train_acc:92.6%, Train_loss:0.262, Test_acc:83.9%，Test_loss:0.368
Epoch:18, Train_acc:92.7%, Train_loss:0.251, Test_acc:82.8%，Test_loss:0.380
Epoch:19, Train_acc:93.1%, Train_loss:0.243, Test_acc:82.5%，Test_loss:0.380
Epoch:20, Train_acc:92.4%, Train_loss:0.238, Test_acc:86.2%，Test_loss:0.365

添加池化2后：

Epoch: 1, Train_acc:60.7%, Train_loss:0.686, Test_acc:56.2%，Test_loss:0.775
Epoch: 2, Train_acc:68.4%, Train_loss:0.601, Test_acc:70.4%，Test_loss:0.601
Epoch: 3, Train_acc:73.7%, Train_loss:0.546, Test_acc:71.8%，Test_loss:0.547
Epoch: 4, Train_acc:75.7%, Train_loss:0.504, Test_acc:73.9%，Test_loss:0.528
Epoch: 5, Train_acc:77.5%, Train_loss:0.473, Test_acc:74.6%，Test_loss:0.527
Epoch: 6, Train_acc:79.0%, Train_loss:0.452, Test_acc:75.1%，Test_loss:0.487
Epoch: 7, Train_acc:81.8%, Train_loss:0.426, Test_acc:76.0%，Test_loss:0.482
Epoch: 8, Train_acc:83.9%, Train_loss:0.401, Test_acc:76.5%，Test_loss:0.492
Epoch: 9, Train_acc:84.1%, Train_loss:0.386, Test_acc:79.7%，Test_loss:0.466
Epoch:10, Train_acc:85.9%, Train_loss:0.374, Test_acc:80.4%，Test_loss:0.449
Epoch:11, Train_acc:86.1%, Train_loss:0.361, Test_acc:81.1%，Test_loss:0.434
Epoch:12, Train_acc:87.8%, Train_loss:0.340, Test_acc:80.9%，Test_loss:0.436
Epoch:13, Train_acc:87.9%, Train_loss:0.334, Test_acc:83.4%，Test_loss:0.431
Epoch:14, Train_acc:88.6%, Train_loss:0.319, Test_acc:82.3%，Test_loss:0.412
Epoch:15, Train_acc:89.0%, Train_loss:0.310, Test_acc:81.1%，Test_loss:0.416
Epoch:16, Train_acc:90.4%, Train_loss:0.300, Test_acc:81.6%，Test_loss:0.409
Epoch:17, Train_acc:89.8%, Train_loss:0.294, Test_acc:79.7%，Test_loss:0.415
Epoch:18, Train_acc:90.5%, Train_loss:0.286, Test_acc:80.2%，Test_loss:0.405
Epoch:19, Train_acc:91.2%, Train_loss:0.272, Test_acc:80.7%，Test_loss:0.407
Epoch:20, Train_acc:91.5%, Train_loss:0.271, Test_acc:80.7%，Test_loss:0.401

调用动态学习率后：

Epoch: 1, Train_acc:60.7%, Train_loss:0.672, Test_acc:67.6%, Test_loss:0.614, Lr:1.00E-04
Epoch: 2, Train_acc:69.9%, Train_loss:0.602, Test_acc:71.1%, Test_loss:0.577, Lr:1.00E-04
Epoch: 3, Train_acc:73.7%, Train_loss:0.536, Test_acc:73.7%, Test_loss:0.558, Lr:9.20E-05
Epoch: 4, Train_acc:76.6%, Train_loss:0.489, Test_acc:75.8%, Test_loss:0.516, Lr:9.20E-05
Epoch: 5, Train_acc:80.0%, Train_loss:0.456, Test_acc:76.9%, Test_loss:0.492, Lr:8.46E-05
Epoch: 6, Train_acc:81.7%, Train_loss:0.424, Test_acc:77.9%, Test_loss:0.471, Lr:8.46E-05
Epoch: 7, Train_acc:83.8%, Train_loss:0.409, Test_acc:76.0%, Test_loss:0.472, Lr:7.79E-05
Epoch: 8, Train_acc:84.9%, Train_loss:0.391, Test_acc:75.8%, Test_loss:0.509, Lr:7.79E-05
Epoch: 9, Train_acc:86.3%, Train_loss:0.376, Test_acc:79.3%, Test_loss:0.458, Lr:7.16E-05
Epoch:10, Train_acc:87.4%, Train_loss:0.352, Test_acc:80.4%, Test_loss:0.440, Lr:7.16E-05
Epoch:11, Train_acc:86.9%, Train_loss:0.350, Test_acc:80.9%, Test_loss:0.430, Lr:6.59E-05
Epoch:12, Train_acc:88.0%, Train_loss:0.334, Test_acc:81.4%, Test_loss:0.422, Lr:6.59E-05
Epoch:13, Train_acc:88.3%, Train_loss:0.330, Test_acc:82.8%, Test_loss:0.431, Lr:6.06E-05
Epoch:14, Train_acc:90.2%, Train_loss:0.322, Test_acc:83.0%, Test_loss:0.418, Lr:6.06E-05
Epoch:15, Train_acc:89.7%, Train_loss:0.311, Test_acc:80.4%, Test_loss:0.414, Lr:5.58E-05
Epoch:16, Train_acc:90.9%, Train_loss:0.303, Test_acc:83.0%, Test_loss:0.404, Lr:5.58E-05
Epoch:17, Train_acc:90.7%, Train_loss:0.295, Test_acc:83.2%, Test_loss:0.400, Lr:5.13E-05
Epoch:18, Train_acc:90.0%, Train_loss:0.290, Test_acc:83.4%, Test_loss:0.401, Lr:5.13E-05
Epoch:19, Train_acc:91.5%, Train_loss:0.282, Test_acc:84.1%, Test_loss:0.404, Lr:4.72E-05
Epoch:20, Train_acc:92.2%, Train_loss:0.274, Test_acc:84.6%, Test_loss:0.399, Lr:4.72E-05

3、结果：

未池化2：

添加池化2后：

使用动态学习率后：

最终代码结果：

D:\Programs\Python\Python39\python.exe D:\PycharmProjects\pythonProject\P4\main.py 
cpu
['Monkeypox', 'Others']
Dataset ImageFolder
    Number of datapoints: 2142
    Root location: ./data/
    StandardTransform
Transform: Compose(
               Resize(size=[224, 224], interpolation=bilinear, max_size=None, antialias=True)
               ToTensor()
               Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
           )
{'Monkeypox': 0, 'Others': 1}
<torch.utils.data.dataset.Subset object at 0x00000249FAFC23A0>
<torch.utils.data.dataset.Subset object at 0x00000249FAFC2340>
1713
429
Shape of X [N, C, H, W]:  torch.Size([32, 3, 224, 224])
Shape of y:  torch.Size([32]) torch.int64
Using cpu device
Network_bn(
  (conv1): Conv2d(3, 12, kernel_size=(5, 5), stride=(1, 1))
  (bn1): BatchNorm2d(12, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv2): Conv2d(12, 12, kernel_size=(5, 5), stride=(1, 1))
  (bn2): BatchNorm2d(12, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (pool): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  (conv4): Conv2d(12, 24, kernel_size=(5, 5), stride=(1, 1))
  (bn4): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (conv5): Conv2d(24, 24, kernel_size=(5, 5), stride=(1, 1))
  (bn5): BatchNorm2d(24, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  (fc1): Linear(in_features=60000, out_features=2, bias=True)
)
Epoch: 1, Train_acc:60.7%, Train_loss:0.672, Test_acc:67.6%, Test_loss:0.614, Lr:1.00E-04
Epoch: 2, Train_acc:69.9%, Train_loss:0.602, Test_acc:71.1%, Test_loss:0.577, Lr:1.00E-04
Epoch: 3, Train_acc:73.7%, Train_loss:0.536, Test_acc:73.7%, Test_loss:0.558, Lr:9.20E-05
Epoch: 4, Train_acc:76.6%, Train_loss:0.489, Test_acc:75.8%, Test_loss:0.516, Lr:9.20E-05
Epoch: 5, Train_acc:80.0%, Train_loss:0.456, Test_acc:76.9%, Test_loss:0.492, Lr:8.46E-05
Epoch: 6, Train_acc:81.7%, Train_loss:0.424, Test_acc:77.9%, Test_loss:0.471, Lr:8.46E-05
Epoch: 7, Train_acc:83.8%, Train_loss:0.409, Test_acc:76.0%, Test_loss:0.472, Lr:7.79E-05
Epoch: 8, Train_acc:84.9%, Train_loss:0.391, Test_acc:75.8%, Test_loss:0.509, Lr:7.79E-05
Epoch: 9, Train_acc:86.3%, Train_loss:0.376, Test_acc:79.3%, Test_loss:0.458, Lr:7.16E-05
Epoch:10, Train_acc:87.4%, Train_loss:0.352, Test_acc:80.4%, Test_loss:0.440, Lr:7.16E-05
Epoch:11, Train_acc:86.9%, Train_loss:0.350, Test_acc:80.9%, Test_loss:0.430, Lr:6.59E-05
Epoch:12, Train_acc:88.0%, Train_loss:0.334, Test_acc:81.4%, Test_loss:0.422, Lr:6.59E-05
Epoch:13, Train_acc:88.3%, Train_loss:0.330, Test_acc:82.8%, Test_loss:0.431, Lr:6.06E-05
Epoch:14, Train_acc:90.2%, Train_loss:0.322, Test_acc:83.0%, Test_loss:0.418, Lr:6.06E-05
Epoch:15, Train_acc:89.7%, Train_loss:0.311, Test_acc:80.4%, Test_loss:0.414, Lr:5.58E-05
Epoch:16, Train_acc:90.9%, Train_loss:0.303, Test_acc:83.0%, Test_loss:0.404, Lr:5.58E-05
Epoch:17, Train_acc:90.7%, Train_loss:0.295, Test_acc:83.2%, Test_loss:0.400, Lr:5.13E-05
Epoch:18, Train_acc:90.0%, Train_loss:0.290, Test_acc:83.4%, Test_loss:0.401, Lr:5.13E-05
Epoch:19, Train_acc:91.5%, Train_loss:0.282, Test_acc:84.1%, Test_loss:0.404, Lr:4.72E-05
Epoch:20, Train_acc:92.2%, Train_loss:0.274, Test_acc:84.6%, Test_loss:0.399, Lr:4.72E-05
Done
预测结果是：Monkeypox

进程已结束，退出代码为 0

二、源代码

#一、 前期准备
#1. 设置GPU

import torch
import torch.nn as nn
import torchvision.transforms as transforms
import torchvision
from torchvision import transforms, datasets

import os,PIL,pathlib

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

print(device)

#2. 导入数据
import os,PIL,random,pathlib

data_dir = './data/'
data_dir = pathlib.Path(data_dir)

data_paths = list(data_dir.glob('*'))
classeNames = [str(path).split("\\")[1] for path in data_paths]
print(classeNames)

total_datadir = './data/'

# 关于transforms.Compose的更多介绍可以参考：https://blog.youkuaiyun.com/qq_38251616/article/details/124878863
train_transforms = transforms.Compose([
    transforms.Resize([224, 224]),  # 将输入图片resize成统一尺寸
    transforms.ToTensor(),          # 将PIL Image或numpy.ndarray转换为tensor，并归一化到[0,1]之间
    transforms.Normalize(           # 标准化处理-->转换为标准正太分布（高斯分布），使模型更容易收敛
        mean=[0.485, 0.456, 0.406],
        std=[0.229, 0.224, 0.225])  # 其中 mean=[0.485,0.456,0.406]与std=[0.229,0.224,0.225] 从数据集中随机抽样计算得到的。
])

total_data = datasets.ImageFolder(total_datadir,transform=train_transforms)
print(total_data)
print(total_data.class_to_idx)

#3. 划分数据集
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])
print(train_dataset)
print(test_dataset)
print(train_size)
print(test_size)

batch_size = 32

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

for X, y in test_dl:
    print("Shape of X [N, C, H, W]: ", X.shape)
    print("Shape of y: ", y.shape, y.dtype)
    break


#二、构建简单的CNN网络
import torch.nn.functional as F

class Network_bn(nn.Module):
    def __init__(self):
        super(Network_bn, self).__init__()
        """
        nn.Conv2d()函数：
        第一个参数（in_channels）是输入的channel数量
        第二个参数（out_channels）是输出的channel数量
        第三个参数（kernel_size）是卷积核大小
        第四个参数（stride）是步长，默认为1
        第五个参数（padding）是填充大小，默认为0
        """
        self.conv1 = nn.Conv2d(in_channels=3, out_channels=12, kernel_size=5, stride=1, padding=0)
        self.bn1 = nn.BatchNorm2d(12)
        self.conv2 = nn.Conv2d(in_channels=12, out_channels=12, kernel_size=5, stride=1, padding=0)
        self.bn2 = nn.BatchNorm2d(12)
        self.pool = nn.MaxPool2d(2,2)
        self.conv4 = nn.Conv2d(in_channels=12, out_channels=24, kernel_size=5, stride=1, padding=0)
        self.bn4 = nn.BatchNorm2d(24)
        self.conv5 = nn.Conv2d(in_channels=24, out_channels=24, kernel_size=5, stride=1, padding=0)
        self.bn5 = nn.BatchNorm2d(24)
        self.pool2 = nn.MaxPool2d(2, 2)
        self.fc1 = nn.Linear(24*50*50, len(classeNames))

    def forward(self, x):
        x = F.relu(self.bn1(self.conv1(x)))
        x = F.relu(self.bn2(self.conv2(x)))
        x = self.pool(x)
        x = F.relu(self.bn4(self.conv4(x)))
        x = F.relu(self.bn5(self.conv5(x)))
        x = self.pool2(x)
        x = x.view(-1, 24*50*50)
        x = self.fc1(x)

        return x

device = "cuda" if torch.cuda.is_available() else "cpu"
print("Using {} device".format(device))

model = Network_bn().to(device)
print(model)

#三、 训练模型
#1. 设置超参数

loss_fn    = nn.CrossEntropyLoss() # 创建损失函数
#固定学习率
# learn_rate = 1e-4 # 学习率
# opt        = torch.optim.SGD(model.parameters(),lr=learn_rate)

#设置动态学习率
def adjust_learning_rate(optimizer, epoch, start_lr):
    # 每 2 个epoch衰减到原来的 0.92
    lr = start_lr * (0.92 ** (epoch // 2))
    for param_group in optimizer.param_groups:
        param_group['lr'] = lr


learn_rate = 1e-4 # 初始学习率
opt  = torch.optim.SGD(model.parameters(), lr=learn_rate)

#2. 编写训练函数

# 训练循环
def train(dataloader, model, loss_fn, optimizer):
    size = len(dataloader.dataset)  # 训练集的大小，一共60000张图片
    num_batches = len(dataloader)  # 批次数目，1875（60000/32）

    train_loss, train_acc = 0, 0  # 初始化训练损失和正确率

    for X, y in dataloader:  # 获取图片及其标签
        X, y = X.to(device), y.to(device)

        # 计算预测误差
        pred = model(X)  # 网络输出
        loss = loss_fn(pred, y)  # 计算网络输出和真实值之间的差距，targets为真实值，计算二者差值即为损失

        # 反向传播
        optimizer.zero_grad()  # grad属性归零
        loss.backward()  # 反向传播
        optimizer.step()  # 每一步自动更新

        # 记录acc与loss
        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

#3. 编写测试函数
def test(dataloader, model, loss_fn):
    size = len(dataloader.dataset)  # 测试集的大小，一共10000张图片
    num_batches = len(dataloader)  # 批次数目，313（10000/32=312.5，向上取整）
    test_loss, test_acc = 0, 0

    # 当不进行训练时，停止梯度更新，节省计算内存消耗
    with torch.no_grad():
        for imgs, target in dataloader:
            imgs, target = imgs.to(device), target.to(device)

            # 计算loss
            target_pred = model(imgs)
            loss = loss_fn(target_pred, target)

            test_loss += loss.item()
            test_acc += (target_pred.argmax(1) == target).type(torch.float).sum().item()

    test_acc /= size
    test_loss /= num_batches

    return test_acc, test_loss

#4. 正式训练
epochs = 20
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)
    # 更新学习率（使用自定义学习率时使用）
    adjust_learning_rate(opt, epoch, learn_rate)

    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)

    # 获取当前的学习率
    lr = opt.state_dict()['param_groups'][0]['lr']

    template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%, Test_loss:{:.3f}, Lr:{:.2E}')
    print(template.format(epoch+1, epoch_train_acc*100, epoch_train_loss,
                          epoch_test_acc*100, epoch_test_loss, lr))
print('Done')


#四、 结果可视化
#1. Loss与Accuracy图
import matplotlib.pyplot as plt
#隐藏警告
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

from datetime import datetime
current_time = datetime.now() # 获取当前时间

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.xlabel(current_time) # 打卡请带上时间戳，否则代码截图无效

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()

#2. 指定图片进行预测

from PIL import Image

classes = list(total_data.class_to_idx)


def predict_one_image(image_path, model, transform, classes):
    test_img = Image.open(image_path).convert('RGB')
    # plt.imshow(test_img)  # 展示预测的图片

    test_img = transform(test_img)
    img = test_img.to(device).unsqueeze(0)

    model.eval()
    output = model(img)

    _, pred = torch.max(output, 1)
    pred_class = classes[pred]
    print(f'预测结果是：{pred_class}')

# 预测训练集中的某张照片
predict_one_image(image_path='./data/Monkeypox/M01_01_00.jpg',
                  model=model,
                  transform=train_transforms,
                  classes=classes)

#五、保存并加载模型
# 模型保存
PATH = './model.pth'  # 保存的参数文件名
torch.save(model.state_dict(), PATH)

# 将参数加载到model当中
model.load_state_dict(torch.load(PATH, map_location=device))