pytorch VGG11識別cifar10數(shù)據(jù)集(訓(xùn)練+預(yù)測單張輸入圖片操作)

更新時間：2020年06月24日 15:11:41 作者：ZJE_ANDY

這篇文章主要介紹了pytorch VGG11識別cifar10數(shù)據(jù)集(訓(xùn)練+預(yù)測單張輸入圖片操作)，具有很好的參考價值，希望對大家有所幫助。一起跟隨小編過來看看吧

首先這是VGG的結(jié)構(gòu)圖，VGG11則是紅色框里的結(jié)構(gòu)，共分五個block，如紅框中的VGG11第一個block就是一個conv3-64卷積層：

一，寫VGG代碼時，首先定義一個 vgg_block(n,in,out)方法，用來構(gòu)建VGG中每個block中的卷積核和池化層：

n是這個block中卷積層的數(shù)目，in是輸入的通道數(shù)，out是輸出的通道數(shù)

有了block以后，我們還需要一個方法把形成的block疊在一起，我們定義這個方法叫vgg_stack：

def vgg_stack(num_convs, channels): # vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))


 net = []
 for n, c in zip(num_convs, channels):
  in_c = c[0]
  out_c = c[1]
  net.append(vgg_block(n, in_c, out_c))
 return nn.Sequential(*net)

右邊的注釋

vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))

里，(1, 1, 2, 2, 2)表示五個block里，各自的卷積層數(shù)目，((3, 64), (64, 128), (128, 256), (256, 512), (512, 512))表示每個block中的卷積層的類型，如(3,64)表示這個卷積層輸入通道數(shù)是3，輸出通道數(shù)是64。vgg_stack方法返回的就是完整的vgg11模型了。

接著定義一個vgg類，包含vgg_stack方法：

#vgg類
class vgg(nn.Module):
 def __init__(self):
  super(vgg, self).__init__()
  self.feature = vgg_net
  self.fc = nn.Sequential(
   nn.Linear(512, 100),
   nn.ReLU(True),
   nn.Linear(100, 10)
  )
 
 def forward(self, x):
  x = self.feature(x)
  x = x.view(x.shape[0], -1)
  x = self.fc(x)
  return x

最后：

net = vgg() #就能獲取到vgg網(wǎng)絡(luò)

那么構(gòu)建vgg網(wǎng)絡(luò)完整的pytorch代碼是：

def vgg_block(num_convs, in_channels, out_channels):
 net = [nn.Conv2d(in_channels, out_channels, kernel_size=3, padding=1), nn.ReLU(True)]
 
 for i in range(num_convs - 1): # 定義后面的許多層
  net.append(nn.Conv2d(out_channels, out_channels, kernel_size=3, padding=1))
  net.append(nn.ReLU(True))
 
 net.append(nn.MaxPool2d(2, 2)) # 定義池化層
 return nn.Sequential(*net)
 
# 下面我們定義一個函數(shù)對這個 vgg block 進(jìn)行堆疊
def vgg_stack(num_convs, channels): # vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))
 net = []
 for n, c in zip(num_convs, channels):
  in_c = c[0]
  out_c = c[1]
  net.append(vgg_block(n, in_c, out_c))
 return nn.Sequential(*net)
 
#確定vgg的類型，是vgg11 還是vgg16還是vgg19
vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))
#vgg類
class vgg(nn.Module):
 def __init__(self):
  super(vgg, self).__init__()
  self.feature = vgg_net
  self.fc = nn.Sequential(
   nn.Linear(512, 100),
   nn.ReLU(True),
   nn.Linear(100, 10)
  )
 def forward(self, x):
  x = self.feature(x)
  x = x.view(x.shape[0], -1)
  x = self.fc(x)
  return x
 
#獲取vgg網(wǎng)絡(luò)
net = vgg()

基于VGG11的cifar10訓(xùn)練代碼：

import sys
import numpy as np
import torch
from torch import nn
from torch.autograd import Variable
from torchvision.datasets import CIFAR10
import torchvision.transforms as transforms
 
def vgg_block(num_convs, in_channels, out_channels):
 net = [nn.Conv2d(in_channels, out_channels, kernel_size=3, padding=1), nn.ReLU(True)]
 
 for i in range(num_convs - 1): # 定義后面的許多層
  net.append(nn.Conv2d(out_channels, out_channels, kernel_size=3, padding=1))
  net.append(nn.ReLU(True))
 
 net.append(nn.MaxPool2d(2, 2)) # 定義池化層
 return nn.Sequential(*net)
 
# 下面我們定義一個函數(shù)對這個 vgg block 進(jìn)行堆疊
def vgg_stack(num_convs, channels): # vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))
 net = []
 for n, c in zip(num_convs, channels):
  in_c = c[0]
  out_c = c[1]
  net.append(vgg_block(n, in_c, out_c))
 return nn.Sequential(*net)
 
#vgg類
class vgg(nn.Module):
 def __init__(self):
  super(vgg, self).__init__()
  self.feature = vgg_net
  self.fc = nn.Sequential(
   nn.Linear(512, 100),
   nn.ReLU(True),
   nn.Linear(100, 10)
  )
 def forward(self, x):
  x = self.feature(x)
  x = x.view(x.shape[0], -1)
  x = self.fc(x)
  return x
 
# 然后我們可以訓(xùn)練我們的模型看看在 cifar10 上的效果
def data_tf(x):
 x = np.array(x, dtype='float32') / 255
 x = (x - 0.5) / 0.5
 x = x.transpose((2, 0, 1)) ## 將 channel 放到第一維，只是 pytorch 要求的輸入方式
 x = torch.from_numpy(x)
 return x
 
transform = transforms.Compose([transforms.ToTensor(),
         transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)),
         ])
def get_acc(output, label):
 total = output.shape[0]
 _, pred_label = output.max(1)
 num_correct = (pred_label == label).sum().item()
 return num_correct / total
 
def train(net, train_data, valid_data, num_epochs, optimizer, criterion):
 if torch.cuda.is_available():
  net = net.cuda()
 for epoch in range(num_epochs):
  train_loss = 0
  train_acc = 0
  net = net.train()
  for im, label in train_data:
   if torch.cuda.is_available():
    im = Variable(im.cuda())
    label = Variable(label.cuda())
   else:
    im = Variable(im)
    label = Variable(label)
   # forward
   output = net(im)
   loss = criterion(output, label)
   # forward
   optimizer.zero_grad()
   loss.backward()
   optimizer.step()
 
   train_loss += loss.item()
   train_acc += get_acc(output, label)
 
  if valid_data is not None:
   valid_loss = 0
   valid_acc = 0
   net = net.eval()
   for im, label in valid_data:
    if torch.cuda.is_available():
     with torch.no_grad():
      im = Variable(im.cuda())
      label = Variable(label.cuda())
    else:
     with torch.no_grad():
      im = Variable(im)
      label = Variable(label)
    output = net(im)
    loss = criterion(output, label)
    valid_loss += loss.item()
    valid_acc += get_acc(output, label)
   epoch_str = (
     "Epoch %d. Train Loss: %f, Train Acc: %f, Valid Loss: %f, Valid Acc: %f, "
     % (epoch, train_loss / len(train_data),
      train_acc / len(train_data), valid_loss / len(valid_data),
      valid_acc / len(valid_data)))
  else:
   epoch_str = ("Epoch %d. Train Loss: %f, Train Acc: %f, " %
       (epoch, train_loss / len(train_data),
       train_acc / len(train_data)))
 
  # prev_time = cur_time
  print(epoch_str)
 
if __name__ == '__main__':
 # 作為實例，我們定義一個稍微簡單一點(diǎn)的 vgg11 結(jié)構(gòu)，其中有 8 個卷積層
 vgg_net = vgg_stack((1, 1, 2, 2, 2), ((3, 64), (64, 128), (128, 256), (256, 512), (512, 512)))
 print(vgg_net)
 
 train_set = CIFAR10('./data', train=True, transform=transform, download=True)
 train_data = torch.utils.data.DataLoader(train_set, batch_size=64, shuffle=True)
 test_set = CIFAR10('./data', train=False, transform=transform, download=True)
 test_data = torch.utils.data.DataLoader(test_set, batch_size=128, shuffle=False)
 
 net = vgg()
 optimizer = torch.optim.SGD(net.parameters(), lr=1e-1)
 criterion = nn.CrossEntropyLoss() #損失函數(shù)為交叉熵
 
 train(net, train_data, test_data, 50, optimizer, criterion)
 torch.save(net, 'vgg_model.pth')

結(jié)束后，會出現(xiàn)一個模型文件vgg_model.pth

二，然后網(wǎng)上找張圖片，把圖片縮成32x32，放到預(yù)測代碼中，即可有預(yù)測結(jié)果出現(xiàn)，預(yù)測代碼如下：

import torch
import cv2
import torch.nn.functional as F
from vgg2 import vgg ##重要，雖然顯示灰色(即在次代碼中沒用到)，但若沒有引入這個模型代碼，加載模型時會找不到模型
from torch.autograd import Variable
from torchvision import datasets, transforms
import numpy as np
 
classes = ('plane', 'car', 'bird', 'cat',
   'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
if __name__ == '__main__':
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 model = torch.load('vgg_model.pth') # 加載模型
 model = model.to(device)
 model.eval() # 把模型轉(zhuǎn)為test模式
 
 img = cv2.imread("horse.jpg") # 讀取要預(yù)測的圖片
 trans = transforms.Compose(
  [
   transforms.ToTensor(),
   transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))
  ])
 
 img = trans(img)
 img = img.to(device)
 img = img.unsqueeze(0) # 圖片擴(kuò)展多一維,因為輸入到保存的模型中是4維的[batch_size,通道,長，寬]，而普通圖片只有三維，[通道,長，寬]
 # 擴(kuò)展后，為[1，1，28，28]
 output = model(img)
 prob = F.softmax(output,dim=1) #prob是10個分類的概率
 print(prob)
 value, predicted = torch.max(output.data, 1)
 print(predicted.item())
 print(value)
 pred_class = classes[predicted.item()]
 print(pred_class)
 
 # prob = F.softmax(output, dim=1)
 # prob = Variable(prob)
 # prob = prob.cpu().numpy() # 用GPU的數(shù)據(jù)訓(xùn)練的模型保存的參數(shù)都是gpu形式的，要顯示則先要轉(zhuǎn)回cpu，再轉(zhuǎn)回numpy模式
 # print(prob) # prob是10個分類的概率
 # pred = np.argmax(prob) # 選出概率最大的一個
 # # print(pred)
 # # print(pred.item())
 # pred_class = classes[pred]
 # print(pred_class)

縮成32x32的圖片：