P4猴痘病识别（小白入门）

于 2025-03-21 20:55:30 发布
阅读量1k 收藏 18
点赞数 25
CC 4.0 BY-SA版权
文章标签： pytorch 人工智能 python
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.youkuaiyun.com/AMHSA/article/details/146428787

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")
device

device(type=‘cuda’)

2.导入数据

import os,PIL,random,pathlib

data_dir = r"D:\z_temp\data\第4周"
data_dir = pathlib.Path(data_dir)

data_paths = list(data_dir.glob('*'))
classNames = [str(path).split("\\")[4] for path in data_paths]
classNames

total_datadir = r"D:\z_temp\data\第4周"

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)
total_data

Dataset ImageFolder
Number of datapoints: 2142
Root location: D:\z_temp\data\第4周
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])
)

3.划分数据集

import os,PIL,random,pathlib

data_dir = r"D:\z_temp\data\第4周"
data_dir = pathlib.Path(data_dir)

data_paths = list(data_dir.glob('*'))
classNames = [str(path).split("\\")[4] for path in data_paths]
classNames

total_datadir = r"D:\z_temp\data\第4周"

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)
total_data

Dataset ImageFolder
Number of datapoints: 2142
Root location: D:\z_temp\data\第4周
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])
)

4.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])
train_dataset,test_dataset

(<torch.utils.data.dataset.Subset at 0x24c0c33d6a0>,
<torch.utils.data.dataset.Subset at 0x24c0c33da30>)

batch_size = 32
train_dl = torch.utils.data.DataLoader(train_dataset,
                                      batch_size = batch_size,
                                      shuffle=True,
                                      num_workers = 1)
test_dl = torch.utils.data.DataLoader(test_dataset,
                                     batch_size = batch_size,
                                     shuffle=True,
                                     num_workers=1)

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

Shap of X [N,C,H,W]: torch.Size([32, 3, 224, 224])
Shap of y: torch.Size([32]) torch.int64

二、构建CNN

import torch.nn.functional as F

class Network_bn(nn.Module):
    def __init__(self):
        super(Network_bn,self).__init__()
        
        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.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.pool(x)                        
        x = F.relu(self.bn4(self.conv4(x)))     
        x = F.relu(self.bn5(self.conv5(x)))  
        x = self.pool(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)
model

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)
(fc1): Linear(in_features=60000, out_features=2, bias=True)
)

三、训练模型

1.设置超参数

loss_fn    = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-4 # 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     = 35
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:61.9%, Train_loss:0.666, Test_acc:64.1%，Test_loss:0.646
Epoch: 2, Train_acc:71.3%, Train_loss:0.578, Test_acc:68.3%，Test_loss:0.598
Epoch: 3, Train_acc:74.9%, Train_loss:0.521, Test_acc:69.7%，Test_loss:0.567
Epoch: 4, Train_acc:77.7%, Train_loss:0.487, Test_acc:71.6%，Test_loss:0.548
Epoch: 5, Train_acc:80.7%, Train_loss:0.449, Test_acc:70.2%，Test_loss:0.557
Epoch: 6, Train_acc:81.8%, Train_loss:0.432, Test_acc:75.8%，Test_loss:0.509
Epoch: 7, Train_acc:83.7%, Train_loss:0.398, Test_acc:72.0%，Test_loss:0.536
Epoch: 8, Train_acc:84.7%, Train_loss:0.390, Test_acc:72.0%，Test_loss:0.524
Epoch: 9, Train_acc:86.5%, Train_loss:0.368, Test_acc:76.7%，Test_loss:0.496
Epoch:10, Train_acc:87.5%, Train_loss:0.356, Test_acc:75.8%，Test_loss:0.496
Epoch:11, Train_acc:88.5%, Train_loss:0.337, Test_acc:79.0%，Test_loss:0.475
Epoch:12, Train_acc:88.1%, Train_loss:0.327, Test_acc:78.8%，Test_loss:0.462
Epoch:13, Train_acc:89.8%, Train_loss:0.315, Test_acc:78.3%，Test_loss:0.464
Epoch:14, Train_acc:89.8%, Train_loss:0.303, Test_acc:80.0%，Test_loss:0.488
Epoch:15, Train_acc:91.4%, Train_loss:0.287, Test_acc:80.0%，Test_loss:0.439
Epoch:16, Train_acc:90.9%, Train_loss:0.285, Test_acc:80.7%，Test_loss:0.442
Epoch:17, Train_acc:91.8%, Train_loss:0.272, Test_acc:79.3%，Test_loss:0.432
Epoch:18, Train_acc:92.2%, Train_loss:0.265, Test_acc:81.1%，Test_loss:0.419
Epoch:19, Train_acc:93.1%, Train_loss:0.253, Test_acc:81.4%，Test_loss:0.438
Epoch:20, Train_acc:92.5%, Train_loss:0.257, Test_acc:80.7%，Test_loss:0.405
Epoch:21, Train_acc:93.9%, Train_loss:0.239, Test_acc:82.8%，Test_loss:0.412
Epoch:22, Train_acc:93.5%, Train_loss:0.237, Test_acc:81.1%，Test_loss:0.408
Epoch:23, Train_acc:93.9%, Train_loss:0.231, Test_acc:83.2%，Test_loss:0.401
Epoch:24, Train_acc:93.6%, Train_loss:0.221, Test_acc:82.8%，Test_loss:0.403
Epoch:25, Train_acc:94.9%, Train_loss:0.215, Test_acc:82.1%，Test_loss:0.399
Epoch:26, Train_acc:94.8%, Train_loss:0.217, Test_acc:82.3%，Test_loss:0.402
Epoch:27, Train_acc:95.2%, Train_loss:0.210, Test_acc:83.7%，Test_loss:0.389
Epoch:28, Train_acc:94.7%, Train_loss:0.204, Test_acc:82.1%，Test_loss:0.400
Epoch:29, Train_acc:95.0%, Train_loss:0.202, Test_acc:82.3%，Test_loss:0.386
Epoch:30, Train_acc:95.2%, Train_loss:0.195, Test_acc:82.3%，Test_loss:0.388
Epoch:31, Train_acc:95.4%, Train_loss:0.190, Test_acc:82.1%，Test_loss:0.392
Epoch:32, Train_acc:94.9%, Train_loss:0.190, Test_acc:85.3%，Test_loss:0.383
Epoch:33, Train_acc:95.6%, Train_loss:0.185, Test_acc:81.6%，Test_loss:0.403
Epoch:34, Train_acc:95.9%, Train_loss:0.181, Test_acc:85.5%，Test_loss:0.370
Epoch:35, Train_acc:95.9%, Train_loss:0.179, Test_acc:83.7%，Test_loss:0.374
Done

毫无疑问test_acc最高85.5%这个结果准确率不够高，但是train_acc=95.9%这个结果表明网络结构可能对训练集有一定的过拟合。得重新调整网络结构

四、结果可视化

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()

请添加图片描述

五、指定图片预测

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}')

predict_one_image(image_path=str(pathlib.Path(data_dir)/"Monkeypox"/"M01_01_00.jpg"),
                  model=model,
                  transform=train_transforms,
                  classes=classes)

预测结果是：Monkeypox

六、优化conv1\2 55~conv1\2\3 33

import torch.nn.functional as F

class Network_bn(nn.Module):
    def __init__(self):
        super(Network_bn,self).__init__()
        
        self.conv1 = nn.Conv2d(in_channels=3,
                               out_channels=32,
                               kernel_size=3,
                               stride = 1,
                               padding=0)#224-3+1=222
        self.bn1 = nn.BatchNorm2d(32)
        self.conv2 = nn.Conv2d(in_channels=32,
                               out_channels=32,
                               kernel_size=3,
                               stride = 1,
                               padding=0)#222-3+1=220
        self.bn2 = nn.BatchNorm2d(32)
        self.conv3 = nn.Conv2d(in_channels=32,
                               out_channels=32,
                               kernel_size=3,
                               stride = 1,
                               padding=0)#220-3+1=218
        self.bn3 = nn.BatchNorm2d(32)
        self.pool = nn.MaxPool2d(2,2)#109
        self.conv5 = nn.Conv2d(in_channels=32,
                              out_channels = 64,
                              kernel_size =5,
                              stride = 1,
                              padding = 0)#105
        self.bn5 = nn.BatchNorm2d(64)
        self.conv6 = nn.Conv2d(in_channels = 64,
                              out_channels = 64,
                              kernel_size=5,
                              stride = 1,
                              padding = 0)#101
        self.bn6 = nn.BatchNorm2d(64)
        #pool2 50
        self.fc1 = nn.Linear(64*50*50,64)
        self.fc2 = nn.Linear(64,len(classNames))
        
                            
    def forward(self, x):
        x = F.relu(self.bn1(self.conv1(x)))      
        x = F.relu(self.bn2(self.conv2(x)))
        x = F.relu(self.bn3(self.conv3(x)))
        x = self.pool(x)                        
        x = F.relu(self.bn5(self.conv5(x)))     
        x = F.relu(self.bn6(self.conv6(x)))  
        x = self.pool(x)                        
        x = x.view(x.size(0),-1)
        x = self.fc1(x)
        x = self.fc2(x)
        
        return x
                    
device = "cuda" if torch.cuda.is_available() else "cpu"
print("Using {} device".format(device))

model = Network_bn().to(device)
model

Epoch: 1, Train_acc:60.5%, Train_loss:0.698, Test_acc:62.5%，Test_loss:0.636
Epoch: 2, Train_acc:68.5%, Train_loss:0.607, Test_acc:72.5%，Test_loss:0.548
Epoch: 3, Train_acc:71.8%, Train_loss:0.553, Test_acc:72.0%，Test_loss:0.536
Epoch: 4, Train_acc:75.2%, Train_loss:0.516, Test_acc:74.1%，Test_loss:0.512
Epoch: 5, Train_acc:77.9%, Train_loss:0.481, Test_acc:72.3%，Test_loss:0.518
Epoch: 6, Train_acc:79.9%, Train_loss:0.449, Test_acc:77.4%，Test_loss:0.472
Epoch: 7, Train_acc:83.5%, Train_loss:0.422, Test_acc:76.5%，Test_loss:0.490
Epoch: 8, Train_acc:83.7%, Train_loss:0.401, Test_acc:80.0%，Test_loss:0.445
Epoch: 9, Train_acc:86.4%, Train_loss:0.379, Test_acc:76.7%，Test_loss:0.472
Epoch:10, Train_acc:86.7%, Train_loss:0.360, Test_acc:79.7%，Test_loss:0.427
Epoch:11, Train_acc:88.4%, Train_loss:0.342, Test_acc:80.2%，Test_loss:0.426
Epoch:12, Train_acc:89.8%, Train_loss:0.330, Test_acc:80.0%，Test_loss:0.431
Epoch:13, Train_acc:89.5%, Train_loss:0.321, Test_acc:78.6%，Test_loss:0.444
Epoch:14, Train_acc:90.0%, Train_loss:0.308, Test_acc:82.3%，Test_loss:0.403
Epoch:15, Train_acc:90.7%, Train_loss:0.292, Test_acc:82.1%，Test_loss:0.385
Epoch:16, Train_acc:91.4%, Train_loss:0.286, Test_acc:82.8%，Test_loss:0.376
Epoch:17, Train_acc:91.8%, Train_loss:0.270, Test_acc:82.8%，Test_loss:0.366
Epoch:18, Train_acc:92.7%, Train_loss:0.257, Test_acc:84.8%，Test_loss:0.368
Epoch:19, Train_acc:93.0%, Train_loss:0.253, Test_acc:82.1%，Test_loss:0.370
Epoch:20, Train_acc:93.4%, Train_loss:0.246, Test_acc:84.6%，Test_loss:0.358
Epoch:21, Train_acc:93.9%, Train_loss:0.240, Test_acc:82.3%，Test_loss:0.379
Epoch:22, Train_acc:93.6%, Train_loss:0.234, Test_acc:84.1%，Test_loss:0.361
Epoch:23, Train_acc:94.2%, Train_loss:0.224, Test_acc:84.1%，Test_loss:0.342
Epoch:24, Train_acc:94.0%, Train_loss:0.217, Test_acc:85.8%，Test_loss:0.333
Epoch:25, Train_acc:94.2%, Train_loss:0.213, Test_acc:86.2%，Test_loss:0.329
Epoch:26, Train_acc:94.1%, Train_loss:0.202, Test_acc:84.1%，Test_loss:0.322
Epoch:27, Train_acc:94.8%, Train_loss:0.200, Test_acc:83.2%，Test_loss:0.347
Epoch:28, Train_acc:95.4%, Train_loss:0.196, Test_acc:85.1%，Test_loss:0.323
Epoch:29, Train_acc:95.3%, Train_loss:0.188, Test_acc:85.1%，Test_loss:0.325
Epoch:30, Train_acc:95.9%, Train_loss:0.183, Test_acc:85.1%，Test_loss:0.315
Epoch:31, Train_acc:95.7%, Train_loss:0.185, Test_acc:87.4%，Test_loss:0.309
Epoch:32, Train_acc:96.0%, Train_loss:0.174, Test_acc:88.1%，Test_loss:0.305
Epoch:33, Train_acc:96.1%, Train_loss:0.171, Test_acc:87.2%，Test_loss:0.295
Epoch:34, Train_acc:96.1%, Train_loss:0.165, Test_acc:86.7%，Test_loss:0.298
Epoch:35, Train_acc:96.0%, Train_loss:0.164, Test_acc:87.4%，Test_loss:0.294
Done
请添加图片描述

test_acc 勉强能够达到88%

AMHSA
关注
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值