今天重点学习怎么在自定义的图像数据集上训练模型，以及如何调用模型识别一张本地图片
- 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
- 🍖 原作者：K同学啊
一. 检查device是否可用
# 检查GPU
import torch
import torch.nn as nn
import torch.nn.functional as F
import torchvision
from torchvision import transforms, datasets
import os,PIL,pathlib,random
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
print(device)
# cuda:0
二. 数据集处理
本任务中的自定义数据集以文件夹名称的形式存储标签，先把文件夹的路径处理成可读取的Path对象，然后再从中截取字符串。使用glob()方法获取data_dir路径下的所有文件路径，并以列表形式存储在data_paths中。
data_dir = './weather_photos/'
data_dir = pathlib.Path(data_dir)
data_paths = list(data_dir.glob('*'))
classeNames = [str(path).split("\\")[1] for path in data_paths]
classeNames
# ['cloudy', 'rain', 'shine', 'sunrise', 'test']
使用PIL.Image包的open方法预览图片，plt.subplots()构建子图
import matplotlib.pyplot as plt
from PIL import Image
# 指定图像文件夹路径
image_folder = './data/cloudy/'
# 获取文件夹中的所有图像文件
image_files = [f for f in os.listdir(image_folder) if f.endswith((".jpg", ".png", ".jpeg"))]
# 创建Matplotlib图像
fig, axes = plt.subplots(3, 8, figsize=(16, 6))
# 使用列表推导式加载和显示图像
for ax, img_file in zip(axes.flat, image_files):
img_path = os.path.join(image_folder, img_file)
img = Image.open(img_path)
ax.imshow(img)
ax.axis('off')
# 显示图像
plt.tight_layout()
plt.show()
使用torchvision.transforms()进行图像归一化，归一化到统一尺寸[224, 224]。
torchvision.datasets.ImageFolder假设所有的文件按文件夹保存，每个文件夹下存储同一个类别的图片，文件夹名为类名。
total_datadir = './weather_photos/'
# 关于transforms.Compose的更多介绍可以参考：https://blog.csdn.net/qq_38251616/article/details/124878863
train_transforms = transforms.Compose([
transforms.Resize([224, 224]), # 将输入图片resize成统一尺寸
transforms.ToTensor(), # 将PIL Image或numpy.ndarray转换为tensor，并归一化到[0,1]之间
transforms.Normalize( # 标准化处理-->转换为标准正太分布（高斯分布），使模型更容易收敛
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]) # 其中 mean=[0.485,0.456,0.406]与std=[0.229,0.224,0.225] 从数据集中随机抽样计算得到的。
])
total_data = datasets.ImageFolder(total_datadir,transform=train_transforms)
total_data
三. 构建dataloader
划分数据集, torch.utils.data.random_split()根据预先定义好的数据量划分训练集和测试集。
train_size = int(0.8 * len(total_data))
test_size = len(total_data) - train_size
train_dataset, test_dataset = torch.utils.data.random_split(total_data, [train_size, test_size])
train_size, test_size
# (901, 226)
构建dataloader并检查tensor形状
from torch.utils.data import DataLoader, Dataset
batch_size = 32
train_dl = DataLoader(train_dataset,batch_size=batch_size,shuffle=True,num_workers=1)
test_dl = DataLoader(test_dataset,batch_size=batch_size,shuffle=True,num_workers=1)
for X, y in train_dl:
print(X.shape, y.shape)
break
# torch.Size([32, 3, 224, 224]) torch.Size([32])
四. 直接套用之前的模型和训练代码
这次模型代码也是CNN，所以我选择在上一次的模型上进行修改。
class CNN(nn.Module):
def __init__(self, num_class) -> None:
super().__init__()
self.num_class = num_class
self.conv1 = nn.Conv2d(3, 32, kernel_size=3, dilation=1)
self.pool1 = nn.MaxPool2d(2)
self.conv2 = nn.Conv2d(32, 64, kernel_size=3, dilation=1)
self.pool2 = nn.MaxPool2d(2)
self.dropout = nn.Dropout(0.1)
self.MLP = nn.Linear(186624, self.num_class)
def forward(self, data):
data = self.pool1(F.relu(self.conv1(data)))
data = self.pool2(F.relu(self.conv2(data)))
b = data.shape[0]
data = data.view(b, -1)
print(data.shape)
data = self.MLP(data)
data = self.dropout(data)
return data
net = CNN(num_class=4)
net(X_demo).shape
训练函数
def train(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset) # 训练集的大小，一共60000张图片
num_batches = len(dataloader) # 批次数目，1875（60000/32）
train_loss, train_acc = 0, 0 # 初始化训练损失和正确率
for X, y in dataloader: # 获取图片及其标签
X, y = X.to(device), y.to(device)
# 计算预测误差
pred = model(X) # 网络输出
loss = loss_fn(pred, y) # 计算网络输出和真实值之间的差距，targets为真实值，计算二者差值即为损失
# 反向传播
optimizer.zero_grad() # grad属性归零
loss.backward() # 反向传播
optimizer.step() # 每一步自动更新
# 记录acc与loss
train_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
train_loss += loss.item()
train_acc /= size
train_loss /= num_batches
return train_acc, train_loss
测试函数
def test (dataloader, model, loss_fn):
size = len(dataloader.dataset) # 测试集的大小，一共10000张图片
num_batches = len(dataloader) # 批次数目，313（10000/32=312.5，向上取整）
test_loss, test_acc = 0, 0
# 当不进行训练时，停止梯度更新，节省计算内存消耗
with torch.no_grad():
for imgs, target in dataloader:
imgs, target = imgs.to(device), target.to(device)
# 计算loss
target_pred = model(imgs)
loss = loss_fn(target_pred, target)
test_loss += loss.item()
test_acc += (target_pred.argmax(1) == target).type(torch.float).sum().item()
test_acc /= size
test_loss /= num_batches
return test_acc, test_loss
主函数
epochs = 20
train_loss = []
train_acc = []
test_loss = []
test_acc = []
model = CNN(num_class=4).to(device)
loss_fn = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-4 # 学习率
opt = torch.optim.SGD(model.parameters(),lr=learn_rate)
for epoch in range(epochs):
model.train()
epoch_train_acc, epoch_train_loss = train(train_dl, model, loss_fn, opt)
model.eval()
epoch_test_acc, epoch_test_loss = test(test_dl, model, loss_fn)
train_acc.append(epoch_train_acc)
train_loss.append(epoch_train_loss)
test_acc.append(epoch_test_acc)
test_loss.append(epoch_test_loss)
template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%，Test_loss:{:.3f}')
print(template.format(epoch+1, epoch_train_acc*100, epoch_train_loss, epoch_test_acc*100, epoch_test_loss))
print('Done')
运行结果
五. 模型调优
由结果可知，模型并未出现过拟合，初步分析是模型最后的MLP参数量过大，导致精度不高。其次卷积层太少，特征提取不充分。 解决方案：
- 在卷积层后面加一个batch_normalization, 需要对一批数据做标准化
- 多添加了一个卷积层
- 加了超参数调度器随训练动态调整学习率，
scheduler = torch.optim.lr_scheduler.StepLR(opt, step_size=5, gamma=0.5)
修改后的模型代码如下
class CNN(nn.Module):
def __init__(self, num_class) -> None:
super().__init__()
self.num_class = num_class
self.conv1 = nn.Conv2d(in_channels=3, out_channels=32, kernel_size=3, dilation=1)
self.bn1 = nn.BatchNorm2d(32)
self.pool1 = nn.MaxPool2d(2)
self.conv2 = nn.Conv2d(in_channels=32, out_channels=64, kernel_size=3, dilation=1)
self.bn2 = nn.BatchNorm2d(64)
self.pool2 = nn.MaxPool2d(2)
self.conv3 = nn.Conv2d(in_channels=64, out_channels=128, kernel_size=3, dilation=1)
self.bn3 = nn.BatchNorm2d(128)
self.pool3 = nn.MaxPool2d(2)
self.conv4 = nn.Conv2d(in_channels=128, out_channels=256, kernel_size=3, dilation=1)
self.bn4 = nn.BatchNorm2d(256)
self.pool4 = nn.MaxPool2d(2)
self.dropout = nn.Dropout(0.1)
self.MLP = nn.Linear(256 * 12 * 12, self.num_class)
def forward(self, data):
data = self.pool1(F.relu(self.bn1(self.conv1(data))))
data = self.pool2(F.relu(self.bn2(self.conv2(data))))
data = self.pool3(F.relu(self.bn3(self.conv3(data))))
data = self.pool4(F.relu(self.bn4(self.conv4(data))))
data = data.view(data.size(0), -1)
data = self.MLP(data)
data = self.dropout(data)
return data
运行结果：
