利⽤PyTorch实现VGG16教程我就废话不多说了，⼤家还是直接看代码吧~
import torch
as nn
functional as F
class VGG16(nn.Module):
def __init__(self):
super(VGG16, self).__init__()
# 3 * 224 * 224
self.maxpool1 = nn.MaxPool2d((2, 2), padding=(1, 1)) # pooling 64 * 112 * 112
self.maxpool2 = nn.MaxPool2d((2, 2), padding=(1, 1)) # pooling 128 * 56 * 56
self.maxpool3 = nn.MaxPool2d((2, 2), padding=(1, 1)) # pooling 256 * 28 * 28
self.maxpool4 = nn.MaxPool2d((2, 2), padding=(1, 1)) # pooling 512 * 14 * 14
self.maxpool5 = nn.MaxPool2d((2, 2), padding=(1, 1)) # pooling 512 * 7 * 7
# view
self.fc1 = nn.Linear(512 * 7 * 7, 4096)
self.fc2 = nn.Linear(4096, 4096)
self.fc3 = nn.Linear(4096, 1000)
# softmax 1 * 1 * 1000
def forward(self, x):
# x.size(0)即为batch_size
in_size = x.size(0)
out = v1_1(x) # 222
out = F.relu(out)
out = v1_2(out) # 222
out = F.relu(out)
out = self.maxpool1(out) # 112
out = v2_1(out) # 110
out = F.relu(out)
out = v2_2(out) # 110
out = F.relu(out)
out = self.maxpool2(out) # 56
out = v3_1(out) # 54
out = F.relu(out)
out = v3_2(out) # 54
out = F.relu(out)
out = v3_3(out) # 54
out = F.relu(out)
out = self.maxpool3(out) # 28
out = v4_1(out) # 26
out = F.relu(out)
out = v4_2(out) # 26
out = F.relu(out)
out = v4_3(out) # 26
out = F.relu(out)
out = self.maxpool4(out) # 14
out = v5_1(out) # 12
out = F.relu(out)
out = v5_2(out) # 12
out = F.relu(out)
out = v5_3(out) # 12
out = F.relu(out)
out = self.maxpool5(out) # 7
# 展平
out = out.view(in_size, -1)
out = self.fc1(out)
out = F.relu(out)
out = self.fc2(out)
out = F.relu(out)
out = self.fc3(out)
out = F.log_softmax(out, dim=1)
return out
补充知识：Pytorch实现VGG（GPU版）
看代码吧~
import torch
from torch import nn
from torch import optim
from PIL import Image
import numpy as np
print(torch.cuda.is_available())
device = torch.device('cuda:0')
path="/content/drive/My Drive/Colab Notebooks/data/dog_vs_cat/"
train_pty((2000,224,224,3),dtype="float32")
train_pty((2000,),dtype="int")
train_pty((2000,3,224,224),dtype="float32")
for i in range(1000):
file_path=path+"cat."+str(i)+".jpg"
image=Image.open(file_path)
resized_image = size((224, 224), Image.ANTIALIAS)
img=np.array(resized_image)
train_X[i,:,:,:]=img
train_Y[i]=0
for i in range(1000):
file_path=path+"dog."+str(i)+".jpg"
image = Image.open(file_path)
resized_image = size((224, 224), Image.ANTIALIAS)
img = np.array(resized_image)
train_X[i+1000, :, :, :] = img
train_Y[i+1000] = 1
train_X /= 255
index = np.arange(2000)
np.random.shuffle(index)
train_X = train_X[index, :, :, :]
train_Y = train_Y[index]
for i in range(3):
train_XX[:,i,:,:]=train_X[:,:,:,i]
# 创建⽹络
class Net(nn.Module):
def __init__(self):
super(Net, self).__init__()
nn.Conv2d(in_channels=3, out_channels=64, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=64, out_channels=64, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.BatchNorm2d(num_features=64, eps=1e-05, momentum=0.1, affine=True),
nn.MaxPool2d(kernel_size=2,stride=2)
)
nn.Conv2d(in_channels=64,out_channels=128,kernel_size=3,stride=1,padding=1),
nn.ReLU(),
nn.Conv2d(in_channels=128, out_channels=128, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.BatchNorm2d(128,eps=1e-5,momentum=0.1,affine=True),
nn.MaxPool2d(kernel_size=2,stride=2)
)
nn.Conv2d(in_channels=128, out_channels=256, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=256, out_channels=256, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.BatchNorm2d(256,eps=1e-5, momentum=0.1, affine=True),
nn.MaxPool2d(kernel_size=2, stride=2)
)
nn.Conv2d(in_channels=256, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.BatchNorm2d(512, eps=1e-5, momentum=0.1, affine=True),
nn.MaxPool2d(kernel_size=2, stride=2)
)
nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.Conv2d(in_channels=512, out_channels=512, kernel_size=3, stride=1, padding=1),  nn.ReLU(),
nn.BatchNorm2d(512, eps=1e-5, momentum=0.1, affine=True),
nn.MaxPool2d(kernel_size=2, stride=2)
numpy官方教程)
self.dense1 = nn.Sequential(
nn.Linear(7*7*512,4096),
nn.ReLU(),
nn.Linear(4096,4096),
nn.ReLU(),
nn.Linear(4096,2)
)
def forward(self, x):
v1(x)
v2(x)
v3(x)
v4(x)
v5(x)
x=x.view(-1,7*7*512)
x=self.dense1(x)
return x
batch_size=16
net = Net().to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(net.parameters(), lr=0.0005)
train_loss = []
for epoch in range(10):
for i in range(2000//batch_size):
x=train_XX[i*batch_size:i*batch_size+batch_size]
y=train_Y[i*batch_size:i*batch_size+batch_size]
x = torch.from_numpy(x)  #(batch_size,input_feature_shape)
y = torch.from_numpy(y)  #(batch_size,label_onehot_shape)
x = x.cuda()
y = y.long().cuda()
out = net(x)
loss = criterion(out, y)  # 计算两者的误差
<_grad()    # 清空上⼀步的残余更新参数值
loss.backward()    # 误差反向传播, 计算参数更新值
optimizer.step()    # 将参数更新值施加到 net 的 parameters 上
train_loss.append(loss.item())
print(epoch, i*batch_size, np.mean(train_loss))
train_loss=[]
total_correct = 0
for i in range(2000):
x = train_XX[i].reshape(1,3,224,224)
y = train_Y[i]
x = torch.from_numpy(x)
x = x.cuda()
out = net(x).cpu()
out = out.detach().numpy()
pred=np.argmax(out)
if pred==y:
total_correct += 1
print(total_correct)
acc = total_correct / 2000.0
print('test acc:', acc)
pty_cache()
将上⾯代码中batch_size改为32，训练次数改为100轮，得到如下准确率
过拟合了~
以上这篇利⽤PyTorch实现VGG16教程就是⼩编分享给⼤家的全部内容了，希望能给⼤家⼀个参考，也希望⼤家多多⽀持。

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。