小白学习PyTorch教程十四迁移学习：微调ResNet实现男人和女人图像分类

Posted 2021-08-30 刘润森！

@Author：Runsen

上次微调了Alexnet，这次微调ResNet实现男人和女人图像分类。

ResNet是 Residual Networks 的缩写，是一种经典的神经网络，用作许多计算机视觉任务。

• ResNet论文参见此处：

https://arxiv.org/abs/1512.03385

该模型是 2015 年 ImageNet 挑战赛的获胜者。 ResNet 的根本性突破是它使我们能够成功训练 150 层以上的极深神经网络。

下面是resnet18的整个网络结构：

Resnet 18 是在 ImageNet 数据集上预训练的图像分类模型。

这次使用Resnet 18 实现分类性别数据集，

该性别分类数据集共有58,658 张图像。（train：47,009 / val：11,649）

• Dataset: Kaggle Gender Classification Dataset

加载数据集

设置图像目录路径并初始化 PyTorch 数据加载器。和之前一样的模板套路

import torch
import torch.nn as nn
import torch.optim as optim

import torchvision
from torchvision import datasets, models, transforms

import numpy as np
import matplotlib.pyplot as plt

import time
import os


device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") # device object


transforms_train = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.RandomHorizontalFlip(), # data augmentation
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]) # normalization
])

transforms_val = transforms.Compose([
    transforms.Resize((224, 224)),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

data_dir = './gender_classification_dataset'
train_datasets = datasets.ImageFolder(os.path.join(data_dir, 'Training'), transforms_train)
val_datasets = datasets.ImageFolder(os.path.join(data_dir, 'Validation'), transforms_val)

train_dataloader = torch.utils.data.DataLoader(train_datasets, batch_size=16, shuffle=True, num_workers=4)
val_dataloader = torch.utils.data.DataLoader(val_datasets, batch_size=16, shuffle=True, num_workers=4)

print('Train dataset size:', len(train_datasets))
print('Validation dataset size:', len(val_datasets))

class_names = train_datasets.classes
print('Class names:', class_names)

plt.rcParams['figure.figsize'] = [12, 8]
plt.rcParams['figure.dpi'] = 60
plt.rcParams.update({'font.size': 20})


def imshow(input, title):
    # torch.Tensor => numpy
    input = input.numpy().transpose((1, 2, 0))
    # undo image normalization
    mean = np.array([0.485, 0.456, 0.406])
    std = np.array([0.229, 0.224, 0.225])
    input = std * input + mean
    input = np.clip(input, 0, 1)
    # display images
    plt.imshow(input)
    plt.title(title)
    plt.show()


# load a batch of train image
iterator = iter(train_dataloader)

# visualize a batch of train image
inputs, classes = next(iterator)
out = torchvision.utils.make_grid(inputs[:4])
imshow(out, title=[class_names[x] for x in classes[:4]])

定义模型

我们使用迁移学习方法，只需要修改最后的输出即可。

model = models.resnet18(pretrained=True)
num_features = model.fc.in_features
model.fc = nn.Linear(num_features, 2) # binary classification (num_of_class == 2)
model = model.to(device)

criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9)

训练阶段

由于ResNet18网络非常复杂，深，这里只训练num_epochs = 3

num_epochs = 3
start_time = time.time()

for epoch in range(num_epochs):
    """ Training  """
    model.train()

    running_loss = 0.
    running_corrects = 0

    # load a batch data of images
    for i, (inputs, labels) in enumerate(train_dataloader):
        inputs = inputs.to(device)
        labels = labels.to(device)

        optimizer.zero_grad()
        outputs = model(inputs)
        _, preds = torch.max(outputs, 1)
        loss = criterion(outputs, labels)

        # get loss value and update the network weights
        loss.backward()
        optimizer.step()

        running_loss += loss.item() * inputs.size(0)
        running_corrects += torch.sum(preds == labels.data)

    epoch_loss = running_loss / len(train_datasets)
    epoch_acc = running_corrects / len(train_datasets) * 100.
    print('[Train #{}] Loss: {:.4f} Acc: {:.4f}% Time: {:.4f}s'.format(epoch, epoch_loss, epoch_acc, time.time() - start_time))

    """ Validation"""
    model.eval()

    with torch.no_grad():
        running_loss = 0.
        running_corrects = 0

        for inputs, labels in val_dataloader:
            inputs = inputs.to(device)
            labels = labels.to(device)

            outputs = model(inputs)
            _, preds = torch.max(outputs, 1)
            loss = criterion(outputs, labels)

            running_loss += loss.item() * inputs.size(0)
            running_corrects += torch.sum(preds == labels.data)

        epoch_loss = running_loss / len(val_datasets)
        epoch_acc = running_corrects / len(val_datasets) * 100.
        print('[Validation #{}] Loss: {:.4f} Acc: {:.4f}% Time: {:.4f}s'.format(epoch, epoch_loss, epoch_acc, time.time() - start_time))

保存训练好的模型文件

save_path = 'face_gender_classification_transfer_learning_with_ResNet18.pth'
torch.save(model.state_dict(), save_path)

训练好的模型文件加载

model = models.resnet18(pretrained=True)
num_features = model.fc.in_features
model.fc = nn.Linear(num_features, 2) 
model.load_state_dict(torch.load(save_path))
model.to(device)

model.eval()
start_time = time.time()

with torch.no_grad():
    running_loss = 0.
    running_corrects = 0

    for i, (inputs, labels) in enumerate(val_dataloader):
        inputs = inputs.to(device)
        labels = labels.to(device)

        outputs = model(inputs)
        _, preds = torch.max(outputs, 1)
        loss = criterion(outputs, labels)

        running_loss += loss.item() * inputs.size(0)
        running_corrects += torch.sum(preds == labels.data)

        if i == 0:
            print('[Prediction Result Examples]')
            images = torchvision.utils.make_grid(inputs[:4])
            imshow(images.cpu(), title=[class_names[x] for x in labels[:4]])
            images = torchvision.utils.make_grid(inputs[4:8])
            imshow(images.cpu(), title=[class_names[x] for x in labels[4:8]])

    epoch_loss = running_loss / len(val_datasets)
    epoch_acc = running_corrects / len(val_datasets) * 100.
    print('[Validation #{}] Loss: {:.4f} Acc: {:.4f}% Time: {:.4f}s'.format(epoch, epoch_loss, epoch_acc, time.time() - start_time))

在最后的测试结果中，ACC达到了97，但是模型太复杂，运行太慢了，在项目中往往不可取。