https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html
在本教程中学习如何使用迁移学习训练用于图像分类的CNN网络。
实际上，很少有人从头开始训练整个卷积网络（随机初始化），因为拥有足够大的数据集的情况相对较少。相反，通常在非常大的数据集（例如 ImageNet，包含 1000 个类别的 120 万张图像）上预训练 ConvNet，然后使用 ConvNet 作为初始化或固定特征提取器来完成自己的任务。
通常来说有两种主要的迁移学习场景：
微调 convnet：即finetune convnet，使用预训练网络初始化网络，而不是随机初始化，就像使用在 imagenet 1000 数据集上训练的网络一样。其余的训练看起来像往常一样。
ConvNet 作为固定特征提取器：冻结除最终全连接层之外的所有网络的权重。最后一个完全连接的层被替换为一个具有随机权重的新层，并且只训练这一层。【只有这层的参数会在反向传播时更新】
一、首先导入相关的包
# License: BSD
# Author: Sasank Chilamkurthy
from __future__ import print_function, division
import torch
import torch.nn as nn
import torch.optim as optim
from torch.optim import lr_scheduler
import torch.backends.cudnn as cudnn
import numpy as np
import torchvision
from torchvision import datasets, models, transforms
import matplotlib.pyplot as plt
import time
import os
import copy
cudnn.benchmark = True
plt.ion() # interactive mode二、加载数据
使用 torchvision 和 torch.utils.data 包来加载数据，今天要解决的问题是训练一个模型来对蚂蚁和蜜蜂进行分类。有大约 120 张针对蚂蚁和蜜蜂的训练图像。每个类别有 75 个验证图像。通常，如果从头开始训练，对于泛化来说这是一个非常小的数据集（即很难泛化）。由于我们使用的是迁移学习，因此应该能够很好地进行泛化。This dataset is a very small subset of imagenet。
Download the data from here and extract it to the current directory.
# Data augmentation and normalization for training
# Just normalization for validation
data_transforms = {
'train': transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
'val': transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
}
data_dir = 'data/hymenoptera_data'
image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x),
data_transforms[x])
for x in ['train', 'val']}
dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=4,
shuffle=True, num_workers=4)
for x in ['train', 'val']}
dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}
class_names = image_datasets['train'].classes
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
我print了image_datasets，可以看到image_datasets是个结构体，包含训练数据和测试数据的样本数、路径、transform操作。
{'train': Dataset ImageFolder
Number of datapoints: 244
Root location: data/hymenoptera_data\train
StandardTransform
Transform: Compose(
RandomResizedCrop(size=(224, 224), scale=(0.08, 1.0), ratio=(0.75, 1.3333), interpolation=bilinear), antialias=None)
RandomHorizontalFlip(p=0.5)
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
), 'val': Dataset ImageFolder
Number of datapoints: 153
Root location: data/hymenoptera_data\val
StandardTransform
Transform: Compose(
Resize(size=256, interpolation=bilinear, max_size=None, antialias=None)
CenterCrop(size=(224, 224))
ToTensor()
Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
)}
print dataloaders：
{'train': <torch.utils.data.dataloader.DataLoader object at 0x000001940A9E99D0>, 'val': <torch.utils.data.dataloader.DataLoader object at 0x000001940A9E99A0>}
三、可视化数据
def imshow(inp, title=None): # 输出为inp
"""Imshow for Tensor."""
inp = inp.numpy().transpose((1, 2, 0))
mean = np.array([0.485, 0.456, 0.406])
std = np.array([0.229, 0.224, 0.225])
inp = std * inp + mean
inp = np.clip(inp, 0, 1)
plt.imshow(inp)
if title is not None:
plt.title(title)
plt.pause(0.001) # pause a bit so that plots are updated
# Get a batch of training data 得到一个batch=4的train数据
inputs, classes = next(iter(dataloaders['train']))
# Make a grid from batch
out = torchvision.utils.make_grid(inputs)
imshow(out, title=[class_names[x] for x in classes])
四、训练模型
Now, let’s write a general function to train a model. Here, we will illustrate:
Scheduling the learning rate（学习率）
Saving the best model（保存最佳模型）
In the following, parameter scheduler is an LR scheduler object from torch.optim.lr_scheduler
即描述训练过程的函数。
def train_model(model, criterion, optimizer, scheduler, num_epochs=25):
since = time.time()
best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
for epoch in range(num_epochs):
print(f'Epoch {epoch}/{num_epochs - 1}')
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'val']:
if phase == 'train':
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)
labels = labels.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == 'train'):
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
loss = criterion(outputs, labels)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_corrects += torch.sum(preds == labels.data)
if phase == 'train':
scheduler.step()
epoch_loss = running_loss / dataset_sizes[phase]
epoch_acc = running_corrects.double() / dataset_sizes[phase]
print(f'{phase} Loss: {epoch_loss:.4f} Acc: {epoch_acc:.4f}')
# deep copy the model
if phase == 'val' and epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
print()
time_elapsed = time.time() - since
print(f'Training complete in {time_elapsed // 60:.0f}m {time_elapsed % 60:.0f}s')
print(f'Best val Acc: {best_acc:4f}')
# load best model weights
model.load_state_dict(best_model_wts)
return model五、可视化模型预测结果
def visualize_model(model, num_images=6):
was_training = model.training
model.eval()
images_so_far = 0
fig = plt.figure()
with torch.no_grad():
for i, (inputs, labels) in enumerate(dataloaders['val']):
inputs = inputs.to(device)
labels = labels.to(device)
outputs = model(inputs)
_, preds = torch.max(outputs, 1)
for j in range(inputs.size()[0]):
images_so_far += 1
ax = plt.subplot(num_images//2, 2, images_so_far)
ax.axis('off')
ax.set_title(f'predicted: {class_names[preds[j]]}')
imshow(inputs.cpu().data[j])
if images_so_far == num_images:
model.train(mode=was_training)
return
model.train(mode=was_training)六、Finetuning预训练网络
Load a pretrained model and reset final fully connected layer.微调ConvNet，加载预训练模型并重置最终全连接层，并进行训练，主要关注第五行model_ft.fc = nn.Linear(num_ftrs, 2)为将多分类网络改为二分类网络：
model_ft = models.resnet18(pretrained=True)
num_ftrs = model_ft.fc.in_features
# Here the size of each output sample is set to 2.
# Alternatively, it can be generalized to nn.Linear(num_ftrs, len(class_names)).
model_ft.fc = nn.Linear(num_ftrs, 2)
model_ft = model_ft.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that all parameters are being optimized
optimizer_ft = optim.SGD(model_ft.parameters(), lr=0.001, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)七、Train and evaluate
model_ft = train_model(model_ft, criterion, optimizer_ft, exp_lr_scheduler,
num_epochs=25)Epoch 22/24
----------
train Loss: 0.3835 Acc: 0.8320
val Loss: 0.2321 Acc: 0.8954
Epoch 23/24
----------
train Loss: 0.3731 Acc: 0.8320
val Loss: 0.2228 Acc: 0.8954
Epoch 24/24
----------
train Loss: 0.3357 Acc: 0.8689
val Loss: 0.2345 Acc: 0.9020
Training complete in 1m 26s
Best val Acc: 0.928105
visualize_model(model_ft)
ConvNet as fixed feature extractor
再将ConvNet作为固定特征提取器，在这里需要冻结除最后一层之外的所有网络。通过设置requires_grad=Falsebackward()来冻结参数，这样反向传播backward()的时候它们的梯度就不会被计算。
model_conv = torchvision.models.resnet18(pretrained=True)
for param in model_conv.parameters():
param.requires_grad = False
# Parameters of newly constructed modules have requires_grad=True by default
num_ftrs = model_conv.fc.in_features
model_conv.fc = nn.Linear(num_ftrs, 2)
model_conv = model_conv.to(device)
criterion = nn.CrossEntropyLoss()
# Observe that only parameters of final layer are being optimized as
# opposed to before.
optimizer_conv = optim.SGD(model_conv.fc.parameters(), lr=0.001, momentum=0.9)
# Decay LR by a factor of 0.1 every 7 epochs
exp_lr_scheduler = lr_scheduler.StepLR(optimizer_conv, step_size=7, gamma=0.1)