Pytorch--手写数字识别
Pytorch--手写数字识别
-
- 实现步骤
- 1、加载数据集
- 2、创建网路
- 3、模型训练
- 4、测试
- 全局代码
实现步骤
1、加载图片（7万张）
2、建立模型（三层非线性层）
3、训练
4、测试
可视化图片、曲线的代码
###绘制曲线
def plot_curve(data):
fig = plt.figure()
plt.plot(range(len(data)), data, color='blue')
plt.legend(['value'], loc='upper right')
plt.xlabel('step')
plt.ylabel('value')
plt.show()
###写数字
def plot_image(img, label, name):
fig = plt.figure()
for i in range(6):
plt.subplot(2, 3, i + 1)
plt.tight_layout()
plt.imshow(img[i][0]*0.3081+0.1307, cmap='gray', interpolation='none')
plt.title("{}: {}".format(name, label[i].item()))
plt.xticks([])
plt.yticks([])
plt.show()
### 编码工具
def one_hot(label, depth=10):
out = torch.zeros(label.size(0), depth)
idx = torch.LongTensor(label).view(-1, 1)
out.scatter_(dim=1, index=idx, value=1)
return out
1、加载数据集
batch_size = 512
#train表示下载的是训练集还是测试集
#download表示是下载
#torchvision.transforms.Normalize((0.1307,), (0.3081,))表示将数据都放在0附近
#batch_size表示一次下载几条
#shuffle表示加载的时候做一个随机的打散
train_loader = torch.utils.data.DataLoader(
torchvision.datasets.MNIST('mnist_data', train=True, download=True,
transform=torchvision.transforms.Compose([
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize(
(0.1307,), (0.3081,))
])),
batch_size=batch_size, shuffle=True)
test_loader = torch.utils.data.DataLoader(
torchvision.datasets.MNIST('mnist_data/', train=False, download=True,
transform=torchvision.transforms.Compose([
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize(
(0.1307,), (0.3081,))
])),
batch_size=batch_size, shuffle=False)
x,y=next(iter(train_loader))
看看我们的数据
plot_image(x,y,"image_sample")
2、创建网路
#创建网络
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
##第一层xw+b
self.fc1=nn.Linear(28*28,256)
self.fc2=nn.Linear(256,64)
self.fc3=nn.Linear(64,10)#十分类
def forward(self,x):
# x:[b,1,28,28]
#h1=wx+b
#relu(h1)
x=F.relu(self.fc1(x))
#h2=relu(h1w2+b2)
x=F.relu(self.fc2(x))
#h3=softmax(h2w3+b3)
x = self.fc3(x)
return x
3、模型训练
#训练
net=Net()
#优化器
optimizer=optim.SGD(net.parameters(),lr=0.01,momentum=0.9)
train_loss=[]
for epoch in range(3):#迭代三编
for batch_idx,(x,y) in enumerate(train_loader):
# print(x.shape,y.shape)
#torch.Size([512, 1, 28, 28]) torch.Size([512])
#打平数据[b,1,28,28]=>[b,feature]
x=x.view(x.size(0),28*28)
#=>[b,10]
out=net(x)
y_onehot=one_hot(y)
#loss=mes(out,y_onehot)
loss=F.mse_loss(out,y_onehot)##计算均方差
#清零梯度
optimizer.zero_grad()
#计算梯度
loss.backward()
#w'=w-lr*grad
optimizer.step()
train_loss.append(loss.item())
if batch_idx%10==0:
print(epoch,batch_idx,loss.item())
绘制曲线
plot_curve(train_loss)
4、测试
#测试
total_correct=0
for x,y in test_loader:
x=x.view(x.size(0),28*28)
out=net(x)
#out:[b,10]
#返回值最大的索引
pred=out.argmax(dim=1)
#预测对的总个数
correct=pred.eq(y).sum().float().item()
total_correct+=correct
total_num=len(test_loader.dataset)
#计算准确度
acc=total_correct/total_num
print(acc)
x,y=next(iter(test_loader))
out=net(x.view(x.size(0),28*28))
pred=out.argmax(dim=1)
测试绘图
plot_image(x,pred,"test")
全局代码
#
import torch
from matplotlib import pyplot as plt
from torch import nn
from torch.nn import functional as F
from torch import optim
import torchvision
###绘制曲线
def plot_curve(data):
fig = plt.figure()
plt.plot(range(len(data)), data, color='blue')
plt.legend(['value'], loc='upper right')
plt.xlabel('step')
plt.ylabel('value')
plt.show()
###写数字
def plot_image(img, label, name):
fig = plt.figure()
for i in range(6):
plt.subplot(2, 3, i + 1)
plt.tight_layout()
plt.imshow(img[i][0]*0.3081+0.1307, cmap='gray', interpolation='none')
plt.title("{}: {}".format(name, label[i].item()))
plt.xticks([])
plt.yticks([])
plt.show()
### 编码工具
def one_hot(label, depth=10):
out = torch.zeros(label.size(0), depth)
idx = torch.LongTensor(label).view(-1, 1)
out.scatter_(dim=1, index=idx, value=1)
return out
##第一步、加载数据集
batch_size = 512
#train表示下载的是训练集还是测试集
#download表示是下载
#torchvision.transforms.Normalize((0.1307,), (0.3081,))表示将数据都放在0附近
#batch_size表示一次下载几条
#shuffle表示加载的时候做一个随机的打散
train_loader = torch.utils.data.DataLoader(
torchvision.datasets.MNIST('mnist_data', train=True, download=True,
transform=torchvision.transforms.Compose([
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize(
(0.1307,), (0.3081,))
])),
batch_size=batch_size, shuffle=True)
test_loader = torch.utils.data.DataLoader(
torchvision.datasets.MNIST('mnist_data/', train=False, download=True,
transform=torchvision.transforms.Compose([
torchvision.transforms.ToTensor(),
torchvision.transforms.Normalize(
(0.1307,), (0.3081,))
])),
batch_size=batch_size, shuffle=False)
x,y=next(iter(train_loader))
#创建网络
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
##第一层xw+b
self.fc1=nn.Linear(28*28,256)
self.fc2=nn.Linear(256,64)
self.fc3=nn.Linear(64,10)#十分类
def forward(self,x):
# x:[b,1,28,28]
#h1=wx+b
#relu(h1)
x=F.relu(self.fc1(x))
#h2=relu(h1w2+b2)
x=F.relu(self.fc2(x))
#h3=softmax(h2w3+b3)
x = self.fc3(x)
return x
#训练
net=Net()
#优化器
optimizer=optim.SGD(net.parameters(),lr=0.01,momentum=0.9)
train_loss=[]
for epoch in range(3):#迭代三编
for batch_idx,(x,y) in enumerate(train_loader):
# print(x.shape,y.shape)
#torch.Size([512, 1, 28, 28]) torch.Size([512])
#打平数据[b,1,28,28]=>[b,feature]
x=x.view(x.size(0),28*28)
#=>[b,10]
out=net(x)
y_onehot=one_hot(y)
#loss=mes(out,y_onehot)
loss=F.mse_loss(out,y_onehot)##计算均方差
#清零梯度
optimizer.zero_grad()
#计算梯度
loss.backward()
#w'=w-lr*grad
optimizer.step()
train_loss.append(loss.item())
if batch_idx%10==0:
print(epoch,batch_idx,loss.item())
# 得到w1，b1，w2，b2，w3，b3
plot_curve(train_loss)
#测试
total_correct=0
for x,y in test_loader:
x=x.view(x.size(0),28*28)
out=net(x)
#out:[b,10]
#返回值最大的索引
pred=out.argmax(dim=1)
#预测对的总个数
correct=pred.eq(y).sum().float().item()
total_correct+=correct
total_num=len(test_loader.dataset)
#计算准确度
acc=total_correct/total_num
print(acc)
x,y=next(iter(test_loader))
out=net(x.view(x.size(0),28*28))
pred=out.argmax(dim=1)
plot_image(x,pred,"test")