到目前为止,我们已经了解了如何定义NN,计算loss以及更新网络的权重。那我们接下来再来看一下数据的处理
当我们需要处理图像、文本、或音/视频文件时,通常可以找到一些python包来载入数据到numpy数组中。然后我们可以将这个数组再转化为torch.*Tensor
。
Pillow
,OpenCV
scipy
,librosa
NLTK
和SpaCy
特别地,对于视觉方面,可以使用torchvision
,其中包括了对Imagenet、CIFAR10、MNIST等常用数据集的数据加载器(data loaders),包括对图片数据变形的操作。即torchvision.datasets
和torch.utils.data.DataLoader
。
在这个教程中,我们将使用CIFAR10数据集。它有如下的分类:airplane、automobile、bird、 cat、 deer、 dog、 frog、horse、 ship、 truck 等。在CIFAR-10里面的图片数据大小是3x32x32,即三通道彩色图,图片大小是32x32像素。
接下来会逐步做如下操作,其实也就是训练分类器的步骤:
torchvision
加载CIFAR10里面的训练和测试数据集,并对数据进行标准化处理使用torchvision
可以很方便地加载CIFAR10
import torch
import torchvision
import torchvision.transforms as transforms
torchvision数据集加载完的输出为范围在[0,1]之间的PILImage图片。我们需要将其标准化为范围为[-1,1]之间的张量。下面的Normalize
,param第一项是mean序列,此处为3项,因为有三个channel。第二项是std序列,同样有3项。同时注意此处的转换是out-of-place,它不会改变原有输出。
transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
trainset = torchvision.datasets.CIFAR10(root='./data', train=True,
download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,
shuffle=True, num_workers=2)
testset = torchvision.datasets.CIFAR10(root='./data', train=False,
download=True, transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=4,
shuffle=False, num_workers=2)
classes = ('plane', 'car', 'bird', 'cat',
'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
输出:
Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data/cifar-10-python.tar.gz
Extracting ./data/cifar-10-python.tar.gz to ./data
Files already downloaded and verified
我们可以先通过下面的方法看一下训练的数据
import matplotlib.pyplot as plt
import numpy as np
# functions to show an image
def imshow(img):
# 让照片再回到[0,1]的范围中
img = img / 2 + 0.5 # unnormalize
npimg = img.numpy()
plt.imshow(np.transpose(npimg, (1, 2, 0)))
plt.show()
# get some random training images
dataiter = iter(trainloader)
images, labels = dataiter.next()
# show images
imshow(torchvision.utils.make_grid(images))
# print labels
print(' '.join('%5s' % classes[labels[j]] for j in range(4)))
输出:
deer truck frog dog
此处可以把上一篇中的NN拿来用,但需要把它换成3通道的图片。(之前是单通道)
import torch.nn as nn
import torch.nn.functional as F
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(3, 6, 5)
self.pool = nn.MaxPool2d(2, 2)
self.conv2 = nn.Conv2d(6, 16, 5)
self.fc1 = nn.Linear(16 * 5 * 5, 120)
self.fc2 = nn.Linear(120, 84)
self.fc3 = nn.Linear(84, 10)
def forward(self, x):
x = self.pool(F.relu(self.conv1(x)))
x = self.pool(F.relu(self.conv2(x)))
x = x.view(-1, 16 * 5 * 5)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = self.fc3(x)
return x
net = Net()
此处我们可以用Cross-Entropy作为损失函数和SGD with momentum作为优化器
import torch.optim as optim
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9)
然后就是关键的步骤了。现在我们需要把遍历数据迭代器,把数据输入到网络和优化函数中。
for epoch in range(2): # loop over the dataset multiple times
running_loss = 0.0
for i, data in enumerate(trainloader, 0):
# get the inputs; data is a list of [inputs, labels]
inputs, labels = data
# zero the parameter gradients
optimizer.zero_grad()
# forward + backward + optimize
outputs = net(inputs)
loss = criterion(outputs, labels)
loss.backward()
optimizer.step()
# print statistics
running_loss += loss.item()
if i % 2000 == 1999: # print every 2000 mini-batches
print('[%d, %5d] loss: %.3f' %
(epoch + 1, i + 1, running_loss / 2000))
running_loss = 0.0
print('Finished Training')
然后需要快速保存这个模型
PATH = './cifar_net.pth'
torch.save(net.state_dict(), PATH)
关于保存的其他操作
我们在训练集上训练了两遍网络,现在我们需要检查网络是否学到了东西。
我们将通过预测神经网络输出的标签来检查这个问题,并和正确样本(ground-truth)对比。如果预测是正确的,我们将样本添加到正确预测的列表中。
dataiter = iter(testloader)
images, labels = dataiter.next()
# print images
imshow(torchvision.utils.make_grid(images))
print('GroundTruth: ', ' '.join('%5s' % classes[labels[j]] for j in range(4)))
# output:
# GroundTruth: cat ship ship plane
然后,我们需要重新载入我们之前保存的模型。并查看输入在模型的处理下,输出会是神马
net = Net()
net.load_state_dict(torch.load(PATH))
outputs = net(images)
输出是10个类别的概率值。一个类的值越高,代表网络就越认为这个图像属于该类。我们得到最高量值的下标/索引;
_, predicted = torch.max(outputs, 1)
print('Predicted: ', ' '.join('%5s' % classes[predicted[j]] for j in range(4)))
# output:
# Predicted: dog ship ship plane
简单的测试结束,我们再看看网络在整个数据集上效果怎么样:
correct = 0
total = 0
with torch.no_grad():
for data in testloader:
images, labels = data
outputs = net(images)
_, predicted = torch.max(outputs.data, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
print('Accuracy of the network on the 10000 test images: %d %%' % (
100 * correct / total))
# output:
# Accuracy of the network on the 10000 test images: 54 %
从结果来看,这显然比随机(10%)的效果好多了。同时,我们也可以看哪些class的识别效果比较好,哪些的不是很好:
class_correct = list(0. for i in range(10))
class_total = list(0. for i in range(10))
with torch.no_grad():
for data in testloader:
images, labels = data
outputs = net(images)
_, predicted = torch.max(outputs, 1)
c = (predicted == labels).squeeze()
for i in range(4):
label = labels[i]
class_correct[label] += c[i].item()
class_total[label] += 1
for i in range(10):
print('Accuracy of %5s : %2d %%' % (
classes[i], 100 * class_correct[i] / class_total[i]))
# output:
# Accuracy of plane : 75 %
# Accuracy of car : 85 %
# Accuracy of bird : 32 %
# Accuracy of cat : 30 %
# Accuracy of deer : 31 %
# Accuracy of dog : 56 %
# Accuracy of frog : 67 %
# Accuracy of horse : 68 %
# Accuracy of ship : 40 %
# Accuracy of truck : 58 %
就像我们把tensor发到GPU,我们也可以把网络发到GPU。
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
# Assuming that we are on a CUDA machine, this should print a CUDA device:
print(device)
# cuda:0
然后这些方法会遍历所有模块,将它们的参数和缓冲区转换为CUDA张量
net.to(device)
同时,要记得把输入和目标在每一步都送入GPU
inputs, labels = inputs.to(device), labels.to(device)
关于GPU加速的其他操作