基于经典网络架构训练图像分类模型
总体框架
数据预处理部分: - 数据增强:torchvision中transforms模块自带功能,比较实用 - 数据预处理:torchvision中transforms也帮我们实现好了,直接调用即可 - DataLoader模块直接读取batch数据
网络模块设置: - 加载预训练模型,torchvision中有很多经典网络架构,调用起来十分方便,并且可以用人家训练好的权重参数来继续训练,也就是所谓的迁移学习 - 需要注意的是别人训练好的任务跟咱们的可不是完全一样,需要把最后的head层改一改,一般也就是最后的全连接层,改成咱们自己的任务 - 训练时可以全部重头训练,也可以只训练最后咱们任务的层,因为前几层都是做特征提取的,本质任务目标是一致的
网络模型保存与测试 - 模型保存的时候可以带有选择性,例如在验证集中如果当前效果好则保存 - 读取模型进行实际测试
0.导入包
import os
import matplotlib.pyplot as plt
%matplotlib inline
import numpy as np
import torch
from torch import nn
import torch.optim as optim
import torchvision
#pip install torchvision
from torchvision import transforms, models, datasets
#https://pytorch.org/docs/stable/torchvision/index.html
import imageio
import time
import warnings
warnings.filterwarnings("ignore")
import random
import sys
import copy
import json
from PIL import Image
1. 数据读取与预处理操作
data_dir = './flower_data/'
train_dir = data_dir + '/train'
valid_dir = data_dir + '/valid'
2.制作好数据源
- data_transforms中指定了所有图像预处理操作 - ImageFolder假设所有的文件按文件夹保存好,每个文件夹下面存贮同一类别的图片,文件夹的名字为分类的名字
data_transforms = {
'train':
transforms.Compose([
transforms.Resize([96, 96]),
transforms.RandomRotation(45),#随机旋转,-45到45度之间随机选
transforms.CenterCrop(64),#从中心开始裁剪
transforms.RandomHorizontalFlip(p=0.5),#随机水平翻转 选择一个概率概率
transforms.RandomVerticalFlip(p=0.5),#随机垂直翻转
transforms.ColorJitter(brightness=0.2, contrast=0.1, saturation=0.1, hue=0.1),#参数1为亮度,参数2为对比度,参数3为饱和度,参数4为色相
transforms.RandomGrayscale(p=0.025),#概率转换成灰度率,3通道就是R=G=B
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])#均值,标准差
]),
'valid':
transforms.Compose([
transforms.Resize([64, 64]),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
]),
}
3.准备数据集和加载器
batch_size = 128
image_datasets = {x: datasets.ImageFolder(os.path.join(data_dir, x), data_transforms[x]) for x in ['train', 'valid']}
dataloaders = {x: torch.utils.data.DataLoader(image_datasets[x], batch_size=batch_size, shuffle=True) for x in ['train', 'valid']}
dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'valid']}
class_names = image_datasets['train'].classes
4.读取标签对应的实际名字
with open('cat_to_name.json', 'r') as f:
cat_to_name = json.load(f)
5. 加载models中提供的模型
并且直接用训练的好权重当做初始化参数
model_name = 'resnet' #可选的比较多 ['resnet', 'alexnet', 'vgg', 'squeezenet', 'densenet', 'inception']
#是否用人家训练好的特征来做
feature_extract = True #都用人家特征,咱先不更新
# 是否用GPU训练
train_on_gpu = torch.cuda.is_available()
if not train_on_gpu:
print('CUDA is not available. Training on CPU ...')
else:
print('CUDA is available! Training on GPU ...')
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
模型参数要不要更新
有时候用人家模型,就一直用了,更不更新咱们可以自己定
def set_parameter_requires_grad(model, feature_extracting):
if feature_extracting:
for param in model.parameters():
param.requires_grad = False
model_ft = models.resnet18()#18层的能快点,条件好点的也可以选152
6.把模型输出层改成自己的
def initialize_model(model_name, num_classes, feature_extract, use_pretrained=True):
model_ft = models.resnet18(pretrained=use_pretrained)
set_parameter_requires_grad(model_ft, feature_extract)
num_ftrs = model_ft.fc.in_features
model_ft.fc = nn.Linear(num_ftrs, 102)#类别数自己根据自己任务来
input_size = 64#输入大小根据自己配置来
return model_ft, input_size
7.设置哪些层需要训练
model_ft, input_size = initialize_model(model_name, 102, feature_extract, use_pretrained=True)
#GPU还是CPU计算
model_ft = model_ft.to(device)
# 模型保存,名字自己起
filename='checkpoint.pth'
# 是否训练所有层
params_to_update = model_ft.parameters()
print("Params to learn:")
if feature_extract:
params_to_update = []
for name,param in model_ft.named_parameters():
if param.requires_grad == True:
params_to_update.append(param)
print("\t",name)
else:
for name,param in model_ft.named_parameters():
if param.requires_grad == True:
print("\t",name)
8.优化器设置
optimizer_ft = optim.Adam(params_to_update, lr=1e-2)#要训练啥参数,你来定
scheduler = optim.lr_scheduler.StepLR(optimizer_ft, step_size=10, gamma=0.1)#学习率每7个epoch衰减成原来的1/10
criterion = nn.CrossEntropyLoss()
9.训练模块
def train_model(model, dataloaders, criterion, optimizer, num_epochs=25,filename='best.pt'):
#咱们要算时间的
since = time.time()
#也要记录最好的那一次
best_acc = 0
#模型也得放到你的CPU或者GPU
model.to(device)
#训练过程中打印一堆损失和指标
val_acc_history = []
train_acc_history = []
train_losses = []
valid_losses = []
#学习率
LRs = [optimizer.param_groups[0]['lr']]
#最好的那次模型,后续会变的,先初始化
best_model_wts = copy.deepcopy(model.state_dict())
#一个个epoch来遍历
for epoch in range(num_epochs):
print('Epoch {}/{}'.format(epoch, num_epochs - 1))
print('-' * 10)
# 训练和验证
for phase in ['train', 'valid']:
if phase == 'train':
model.train() # 训练
else:
model.eval() # 验证
running_loss = 0.0
running_corrects = 0
# 把数据都取个遍
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)#放到你的CPU或GPU
labels = labels.to(device)
# 清零
optimizer.zero_grad()
# 只有训练的时候计算和更新梯度
outputs = model(inputs)
loss = criterion(outputs, labels)
_, preds = torch.max(outputs, 1)
# 训练阶段更新权重
if phase == 'train':
loss.backward()
optimizer.step()
# 计算损失
running_loss += loss.item() * inputs.size(0)#0表示batch那个维度
running_corrects += torch.sum(preds == labels.data)#预测结果最大的和真实值是否一致
epoch_loss = running_loss / len(dataloaders[phase].dataset)#算平均
epoch_acc = running_corrects.double() / len(dataloaders[phase].dataset)
time_elapsed = time.time() - since#一个epoch我浪费了多少时间
print('Time elapsed {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
print('{} Loss: {:.4f} Acc: {:.4f}'.format(phase, epoch_loss, epoch_acc))
# 得到最好那次的模型
if phase == 'valid' and epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
state = {
'state_dict': model.state_dict(),#字典里key就是各层的名字,值就是训练好的权重
'best_acc': best_acc,
'optimizer' : optimizer.state_dict(),
}
torch.save(state, filename)
if phase == 'valid':
val_acc_history.append(epoch_acc)
valid_losses.append(epoch_loss)
#scheduler.step(epoch_loss)#学习率衰减
if phase == 'train':
train_acc_history.append(epoch_acc)
train_losses.append(epoch_loss)
print('Optimizer learning rate : {:.7f}'.format(optimizer.param_groups[0]['lr']))
LRs.append(optimizer.param_groups[0]['lr'])
print()
scheduler.step()#学习率衰减
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
print('Best val Acc: {:4f}'.format(best_acc))
# 训练完后用最好的一次当做模型最终的结果,等着一会测试
model.load_state_dict(best_model_wts)
return model, val_acc_history, train_acc_history, valid_losses, train_losses, LRs
开始训练!
- 我们现在只训练了输出层
model_ft, val_acc_history, train_acc_history, valid_losses, train_losses, LRs = train_model(model_ft, dataloaders, criterion, optimizer_ft, num_epochs=20)
### 再继续训练所有层
for param in model_ft.parameters():
param.requires_grad = True
# 再继续训练所有的参数,学习率调小一点
optimizer = optim.Adam(model_ft.parameters(), lr=1e-3)
scheduler = optim.lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1)
# 损失函数
criterion = nn.CrossEntropyLoss()
# 加载之前训练好的权重参数
checkpoint = torch.load(filename)
best_acc = checkpoint['best_acc']
model_ft.load_state_dict(checkpoint['state_dict'])
model_ft, val_acc_history, train_acc_history, valid_losses, train_losses, LRs = train_model(model_ft, dataloaders, criterion, optimizer, num_epochs=10,)
### 加载训练好的模型
model_ft, input_size = initialize_model(model_name, 102, feature_extract, use_pretrained=True)
# GPU模式
model_ft = model_ft.to(device)
# 保存文件的名字
filename='best.pt'
# 加载模型
checkpoint = torch.load(filename)
best_acc = checkpoint['best_acc']
model_ft.load_state_dict(checkpoint['state_dict'])
测试数据预处理
- 测试数据处理方法需要跟训练时一直才可以
- crop操作的目的是保证输入的大小是一致的
- 标准化操作也是必须的,用跟训练数据相同的mean和std,但是需要注意一点训练数据是在0-1上进行标准化,所以测试数据也需要先归一化
- 最后一点,PyTorch中颜色通道是第一个维度,跟很多工具包都不一样,需要转换
# 得到一个batch的测试数据
dataiter = iter(dataloaders['valid'])
images, labels = dataiter.next()
model_ft.eval()
if train_on_gpu:
output = model_ft(images.cuda())
else:
output = model_ft(images)
output表示对一个batch中每一个数据得到其属于各个类别的可能性
### 得到概率最大的那个
_, preds_tensor = torch.max(output, 1)
preds = np.squeeze(preds_tensor.numpy()) if not train_on_gpu else np.squeeze(preds_tensor.cpu().numpy())
preds
展示预测结果
def im_convert(tensor):
""" 展示数据"""
image = tensor.to("cpu").clone().detach()
image = image.numpy().squeeze()
image = image.transpose(1,2,0)
image = image * np.array((0.229, 0.224, 0.225)) + np.array((0.485, 0.456, 0.406))
image = image.clip(0, 1)
return image
fig=plt.figure(figsize=(20, 20))
columns =4
rows = 2
for idx in range (columns*rows):
ax = fig.add_subplot(rows, columns, idx+1, xticks=[], yticks=[])
plt.imshow(im_convert(images[idx]))
ax.set_title("{} ({})".format(cat_to_name[str(preds[idx])], cat_to_name[str(labels[idx].item())]),
color=("green" if cat_to_name[str(preds[idx])]==cat_to_name[str(labels[idx].item())] else "red"))
plt.show()