使用Pytorch进行数字手写体识别

写在前面

本文基于Pytorch，采用CNN卷积神经网络实现手写数字识别，共采用了2个卷积层、1个池化层和2个线性层。

实验准备

首先需要先导入必要的依赖包

import torch
from torchvision import transforms
from PIL import Image
import matplotlib.pyplot as plt
from model import CNN

然后需要准备数据集并进行加载

# 1. 数据准备
transform = transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize((0.5,), (0.5,))
])

# 下载并加载训练集和测试集
train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
test_dataset = datasets.MNIST(root='./data', train=False, download=True, transform=transform)

train_loader = DataLoader(train_dataset, batch_size=64, shuffle=True)
test_loader = DataLoader(test_dataset, batch_size=1000, shuffle=False)

随后定义了CNN卷积模型，包含2个卷积层、1个最大池化层以及2个线性层

class CNN(nn.Module):
    def __init__(self):
        super(CNN, self).__init__()
        self.conv1 = nn.Conv2d(1, 32, kernel_size=3, padding=1)
        self.conv2 = nn.Conv2d(32, 64, kernel_size=3, padding=1)
        self.pool = nn.MaxPool2d(2, 2)
        self.dropout1 = nn.Dropout(0.25)
        self.dropout2 = nn.Dropout(0.5)
        self.fc1 = nn.Linear(64 * 7 * 7, 128)
        self.fc2 = nn.Linear(128, 10)

    def forward(self, x):
        x = self.pool(torch.relu(self.conv1(x)))
        x = self.pool(torch.relu(self.conv2(x)))
        x = self.dropout1(x)
        x = x.view(-1, 64 * 7 * 7)  # 展平
        x = torch.relu(self.fc1(x))
        x = self.dropout2(x)
        x = self.fc2(x)
        return x

初始化参数

# 2. 定义模型
model = CNN().to(device)

# 3. 定义损失函数和优化器
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=0.001)

训练与测试

训练函数

def train(model, device, train_loader, optimizer, criterion, epoch):
    model.train()
    train_loss = 0
    correct = 0
    for batch_idx, (data, target) in enumerate(train_loader):
        data, target = data.to(device), target.to(device)
        optimizer.zero_grad()
        output = model(data)
        loss = criterion(output, target)
        loss.backward()
        optimizer.step()

        train_loss += loss.item()
        pred = output.argmax(dim=1, keepdim=True)
        correct += pred.eq(target.view_as(pred)).sum().item()

    train_loss /= len(train_loader.dataset)
    accuracy = 100. * correct / len(train_loader.dataset)
    print(f'Train Epoch: {epoch} \tLoss: {train_loss:.6f} \tAccuracy: {accuracy:.2f}%')
    return train_loss, accuracy

测试函数

# 5. 测试函数
def test(model, device, test_loader, criterion):
    model.eval()
    test_loss = 0
    correct = 0
    with torch.no_grad():
        for data, target in test_loader:
            data, target = data.to(device), target.to(device)
            output = model(data)
            test_loss += criterion(output, target).item()
            pred = output.argmax(dim=1, keepdim=True)
            correct += pred.eq(target.view_as(pred)).sum().item()

    test_loss /= len(test_loader.dataset)
    accuracy = 100. * correct / len(test_loader.dataset)
    print(f'Test set: Average loss: {test_loss:.6f} \tAccuracy: {accuracy:.2f}%')
    return test_loss, accuracy

训练出最优模型

def main():
    epochs = 10
    best_accuracy = 0.0
    model_save_path = 'best_model.pth'

    for epoch in range(1, epochs + 1):
        train_loss, train_acc = train(model, device, train_loader, optimizer, criterion, epoch)
        test_loss, test_acc = test(model, device, test_loader, criterion)

        # 保存最佳模型
        if test_acc > best_accuracy:
            best_accuracy = test_acc
            torch.save({
                'epoch': epoch,
                'model_state_dict': model.state_dict(),
                'optimizer_state_dict': optimizer.state_dict(),
                'loss': test_loss,
                'accuracy': test_acc
            }, model_save_path)
            print(f"New best model saved with accuracy: {best_accuracy:.2f}%")

    print(f"Training complete. Best test accuracy: {best_accuracy:.2f}%")

实际预测

加载模型

def load_model(model_path):
    model = CNN()
    checkpoint = torch.load(model_path, map_location='cpu')  # 使用CPU加载
    model.load_state_dict(checkpoint['model_state_dict'])
    model.eval()  # 设置为评估模式
    return model

图像预处理

def preprocess_image(image_path):
    # 与训练时相同的转换
    transform = transforms.Compose([
        transforms.Grayscale(),  # 转换为灰度图
        transforms.Resize((28, 28)),  # MNIST是28x28
        transforms.ToTensor(),
        transforms.Normalize((0.1307,), (0.3081,))
    ])
    image = Image.open(image_path)
    image = transform(image).unsqueeze(0)  # 添加batch维度
    return image

预测函数

def predict(model, image_tensor):
    with torch.no_grad():
        output = model(image_tensor)
        _, predicted = torch.max(output.data, 1)
        probabilities = torch.softmax(output, dim=1)
    return predicted.item(), probabilities.squeeze().tolist()

可视化函数

def visualize_prediction(image_path, prediction, probabilities):
    image = Image.open(image_path)
    plt.imshow(image, cmap='gray')
    plt.title(f'Predicted: {prediction}')
    plt.axis('off')

    # 显示概率分布
    plt.figure()
    plt.bar(range(10), probabilities)
    plt.xticks(range(10))
    plt.xlabel('Digit')
    plt.ylabel('Probability')
    plt.title('Prediction Probabilities')
    plt.show()

主函数

def recognition(image_path):
    # 参数设置
    model_path = 'best_model.pth'

    # 加载模型
    model = load_model(model_path)
    print("Model loaded successfully.")

    # 预处理图像
    image_tensor = preprocess_image(image_path)

    # 进行预测
    prediction, probabilities = predict(model, image_tensor)
    print(f"Predicted digit: {prediction}")
    print("Probabilities for each digit (0-9):")
    for i, prob in enumerate(probabilities):
        print(f"{i}: {prob * 100:.2f}%")

    # 可视化结果
    visualize_prediction(image_path, prediction, probabilities)

最后，调用主函数检测手写数字体。以手写数字0为例

检测结果如下所示，展示了CNN模型对每一个数字的检测概率，据条形图可知数字0的概率最大。