1. 数据增广¶
2. 总结¶
1. 数据增广¶
In [1]:
%matplotlib inline
import torch
import torchvision
from torch import nn
from d2l import torch as d2l
d2l.set_figsize()
img = d2l.Image.open('01_Data/02_cat.jpg') # 读取图片
d2l.plt.imshow(img) # 显示图片
Out[1]:
<matplotlib.image.AxesImage at 0x1d31b4730f0>
In [2]:
def apply(img, aug, num_rows=2, num_cols=4, scale=1.5): # 传入aug图片增广方法
Y = [aug(img) for _ in range(num_rows * num_cols)] # 用aug方法对图片作用八次
d2l.show_images(Y, num_rows, num_cols, scale=scale) # 生成结果用num_cols行，num_cols列展示
apply(img, torchvision.transforms.RandomHorizontalFlip()) # 水平随机翻转
In [3]:
apply(img, torchvision.transforms.RandomVerticalFlip()) # 上下随机翻转
In [4]:
# 随机剪裁，剪裁后的大小为(200,200)
# (0.1,1)使得随即剪裁原始图片的10%到100%区域里的大小，ratio=(0.5,2)使得高宽比为2:1，下面是显示时显示的1:1
shape_aug = torchvision.transforms.RandomResizedCrop((200,200),scale=(0.1,1),ratio=(0.5,2))
apply(img,shape_aug)
In [5]:
# 随即更改图像的亮度
apply(img,torchvision.transforms.ColorJitter(brightness=0.5,contrast=0,saturation=0,hue=0))
In [6]:
# 随即改变色调
apply(img,torchvision.transforms.ColorJitter(brightness=0,contrast=0,saturation=0,hue=0.5))
In [7]:
# 随机更改图像的亮度（brightness）、对比度（constrast）、饱和度（saturation）和色调（hue）
color_aug = torchvision.transforms.ColorJitter(brightness=0.5,contrast=0.5,saturation=0.5,hue=0.5)
apply(img,color_aug)
In [8]:
# 结合多种图像增广方法
# 先随即水平翻转，再做颜色增广，再做形状增广
augs = torchvision.transforms.Compose([torchvision.transforms.RandomHorizontalFlip(),color_aug,shape_aug])
apply(img,augs)
In [9]:
# 下载图片，并显示部分图片
all_images = torchvision.datasets.CIFAR10(train=True, root='01_Data/03_CIFAR10', download=True)
d2l.show_images([all_images[i][0] for i in range(32)], 4, 8, scale=0.8)
Files already downloaded and verified
Out[9]:
array([<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>,
<AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>, <AxesSubplot:>],
dtype=object)
In [10]:
# 只使用最简单的随机左右翻转
train_augs = torchvision.transforms.Compose([
torchvision.transforms.RandomHorizontalFlip(),
torchvision.transforms.ToTensor()])
test_augs = torchvision.transforms.Compose([
torchvision.transforms.ToTensor()])
In [11]:
# 定义一个辅助函数，以便于读取图像和应用图像增广
def load_cifar10(is_train, augs, batch_size):
dataset = torchvision.datasets.CIFAR10(root='01_Data/03_CIFAR10',train=is_train,
transform=augs, download=True)
dataloader = torch.utils.data.DataLoader(dataset,batch_size=batch_size,shuffle=is_train,
num_workers = 0)
return dataloader
In [12]:
# 定义一个函数，使用多GPU模式进行训练和评估
def train_batch_ch13(net, X, y, loss, trainer, devices):
if isinstance(X, list):
X = [x.to(devices[0]) for x in X] # 如果X是一个list，则把数据一个接一个都挪到devices[0]上
else:
X = X.to(devices[0]) # 如果X不是一个list，则把X挪到devices[0]上
y = y.to(devices[0])
net.train()
trainer.zero_grad()
pred = net(X)
l = loss(pred, y)
l.sum().backward()
trainer.step()
train_loss_sum = l.sum()
train_acc_sum = d2l.accuracy(pred, y)
return train_loss_sum, train_acc_sum
def train_ch13(net, train_iter, test_iter, loss, trainer, num_epochs, devices=d2l.try_all_gpus()):
timer, num_batches = d2l.Timer(), len(train_iter)
animator = d2l.Animator(xlabel='epoch',xlim=[1,num_epochs],ylim=[0,1],
legend=['train loss', 'train acc', 'test acc'])
# nn.DataParallel使用多GPU
net = nn.DataParallel(net, device_ids=devices).to(devices[0])
for epoch in range(num_epochs):
metric = d2l.Accumulator(4)
for i, (features, labels) in enumerate(train_iter):
timer.start()
l, acc = train_batch_ch13(net,features,labels,loss,trainer,devices)
metric.add(l,acc,labels.shape[0],labels.numel())
timer.stop()
if (i + 1) % (num_batches // 5) == 0 or i == num_batches -1:
animator.add(
epoch + (i + 1) / num_batches,
(metric[0] / metric[2], metric[1] / metric[3], None))
test_acc = d2l.evaluate_accuracy_gpu(net,test_iter)
animator.add(epoch+1,(None,None,test_acc))
print(f'loss {metric[0] / metric[2]:.3f}, train acc'
f' {metric[1] / metric[3]:.3f}, test acc {test_acc:.3f}')
print(f' {metric[2] * num_epochs / timer.sum():.1f} examples/sec on '
f' {str(devices)}')
In [13]:
# 定义train_with_data_aug函数，使用图像增广来训练模型
batch_size, devices, net = 256, d2l.try_all_gpus(), d2l.resnet18(10,3)
def init_weights(m):
if type(m) in [nn.Linear, nn.Conv2d]:
nn.init.xavier_uniform_(m.weight)
net.apply(init_weights)
def train_with_data_aug(train_augs, test_augs, net, lr=0.001):
train_iter = load_cifar10(True, train_augs, batch_size)
test_iter = load_cifar10(False, test_augs, batch_size)
loss = nn.CrossEntropyLoss(reduction="none")
# Adam优化器算是一个比较平滑的SGD，它对学习率调参不是很敏感
trainer = torch.optim.Adam(net.parameters(),lr=lr)
train_ch13(net, train_iter, test_iter, loss, trainer, 10, devices)
train_with_data_aug(train_augs, test_augs, net)
loss 0.166, train acc 0.942, test acc 0.832 1013.2 examples/sec on [device(type='cuda', index=0)]
