Pytorch实战100例-第8天：C3模块实现

本文为🔗365天深度学习训练营 内部限免文章
参考本文所写记录性文章，请在文章开头注明以下内容，复制粘贴即可

• 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
• 🍦 参考文章：Pytorch实战 | P8周：C3模块实现
• 🍖 原作者：K同学啊|接辅导、项目定制

⏲往期文章：

• 深度学习实战训练 | 第8周：猫狗识别
• 深度学习实战训练 | 第7周：咖啡豆识别
• 深度学习实战训练 | 第6周：好莱坞明星识别
• 深度学习实战训练 | 第5周：运动鞋品牌识别

☕ Pytorch实战

• Pytorch实战 | 第P1周：实现mnist手写数字识别
• Pytorch实战 | 第P2周：彩色图片识别
• Pytorch实战 | 第P3周：天气识别

• 难度：新手入门⭐
• 语言：Python3、Pytorch

🍺要求：

1. 本地读取并加载数据。
2. 测试集accuracy到达93%

🍻拔高：

1. 测试集accuracy到达95%
2. 调用模型识别一张本地图片

🏡 我的环境：

• 语言环境：Python3.8
• 编译器：jupyter notebook
• 深度学习环境：Pytorch

一、 前期准备

1. 设置GPU

如果设备上支持GPU就使用GPU,否则使用CPU

import torch
import torch.nn as nn
import torchvision.transforms as transforms
import torchvision
from torchvision import transforms, datasets

import os,PIL,pathlib

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

device

device(type='cpu')

2. 导入数据

import os,PIL,random,pathlib

data_dir = '/home/liangjie/test/Modelwhale/deep learning/p8/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']

total_datadir = '/deep learning/p8/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

Dataset ImageFolder
    Number of datapoints: 1125
    Root location: /home/liangjie/test/Modelwhale/deep learning/p8/weather_photos
    StandardTransform
Transform: Compose(
               Resize(size=[224, 224], interpolation=bilinear, max_size=None, antialias=None)
               ToTensor()
               Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
           )

3. 划分数据集

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_dataset, test_dataset

(<torch.utils.data.dataset.Subset at 0x7f67affefca0>,
 <torch.utils.data.dataset.Subset at 0x7f67b0093040>)

train_size,test_size

(900, 225)

batch_size = 32

train_dl = torch.utils.data.DataLoader(train_dataset,
                                           batch_size=batch_size,
                                           shuffle=True,
                                           num_workers=1)
test_dl = torch.utils.data.DataLoader(test_dataset,
                                          batch_size=batch_size,
                                          shuffle=True,
                                          num_workers=1)

for X, y in test_dl:
    print("Shape of X [N, C, H, W]: ", X.shape)
    print("Shape of y: ", y.shape, y.dtype)
    break

Shape of X [N, C, H, W]:  torch.Size([32, 3, 224, 224])
Shape of y:  torch.Size([32]) torch.int64

二、搭建包含C3模块的模型

对于一般的CNN网络来说，都是由特征提取网络和分类网络构成，其中特征提取网络用于提取图片的特征，分类网络用于将图片进行分类。

⭐1. torch.nn.Conv2d()详解

函数原型：

torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode=‘zeros’, device=None, dtype=None)

关键参数说明：

• in_channels ( int ) – 输入图像中的通道数
• out_channels ( int ) – 卷积产生的通道数
• kernel_size ( int or tuple ) – 卷积核的大小
• stride ( int or tuple , optional ) – 卷积的步幅。默认值：1
• padding ( int , tuple或str , optional ) – 添加到输入的所有四个边的填充。默认值：0
• padding_mode (字符串,可选) – ‘zeros’, ‘reflect’, ‘replicate’或’circular’. 默认：‘zeros’

⭐2. torch.nn.Linear()详解

函数原型：

torch.nn.Linear(in_features, out_features, bias=True, device=None, dtype=None)

关键参数说明：

• in_features：每个输入样本的大小
• out_features：每个输出样本的大小

⭐3. torch.nn.MaxPool2d()详解

函数原型：

torch.nn.MaxPool2d(kernel_size, stride=None, padding=0, dilation=1, return_indices=False, ceil_mode=False)

关键参数说明：

• kernel_size：最大的窗口大小
• stride：窗口的步幅，默认值为kernel_size
• padding：填充值，默认为0
• dilation：控制窗口中元素步幅的参数

def autopad(k, p=None):  # kernel, padding
    # Pad to 'same'
    if p is None:
        p = k // 2 if isinstance(k, int) else [x // 2 for x in k]  # auto-pad
    return p


class Conv(nn.Module):
    def __init__(self, in_channel, out_channel, k=1, s=1, p=None, g=1, act=True):
        super(Conv, self).__init__()
        self.conv = nn.Conv2d(in_channel, out_channel, kernel_size=k, stride=s, padding = autopad(k, p), groups=g, bias=False)
        self.bn = nn.BatchNorm2d(out_channel)
        self.act = nn.SiLU() if act is True else (act if isinstance(act, nn.Module) else nn.Identity())
    def forward(self, x):
        return self.act(self.bn(self.conv(x)))


class Bottleneck(nn.Module):
    def __init__(self, c1, c2, shortcut=True, g=1, e=0.5):
        super(Bottleneck, self).__init__()
        self.c_ = int(c2 * e) 
        self.conv1 =  Conv(c1, self.c_)
        self.conv2 =  Conv(self.c_, c2)
        self.add = shortcut and c1 == c2

    def forward(self, x):
        out = self.conv1(x)
        out = self.conv2(out)
        return out + x if self.add else out

class C3(nn.Module):
    def __init__(self, c1, c2, n=1, shortcut=True, g=1, e=0.5):
        super(C3, self).__init__()
        self.c_ = int(c2 * e)
        self.conv1 = Conv(c1, self.c_)
        self.conv2 = Conv(c1, self.c_)
        self.conv3 = Conv(2*self.c_, c2)
        self.m = nn.Sequential(*(Bottleneck(self.c_, self.c_, shortcut, g, e=1.0) for _ in range(n)))
    def forward(self, x):
        out1 = self.conv1(x)
        out1 = self.m(out1)
        out2 = self.conv2(x)
        out = torch.cat((out1, out2), dim=1)
        return self.conv3(out)
class C3_net(nn.Module):
    def __init__(self):
        super(C3_net, self).__init__()
        self.Conv1 = Conv(3, 32, 3, 2) 
        self.C3_1 = C3(32, 64, 3, 2)
        self.classifier = nn.Sequential(
            nn.Linear(in_features=802816, out_features=100),
            nn.ReLU(),
            nn.Linear(in_features=100, out_features=4)
        )
    def forward(self, x):
        x = self.Conv1(x)
        x = self.C3_1(x)
        x = torch.flatten(x, start_dim=1)
        x = self.classifier(x)

        return x
    
torch.manual_seed(42)
model = C3_net()
model

C3_net(
  (Conv1): Conv(
    (conv): Conv2d(3, 32, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
    (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
    (act): SiLU()
  )
  (C3_1): C3(
    (conv1): Conv(
      (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (act): SiLU()
    )
    (conv2): Conv(
      (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (act): SiLU()
    )
    (conv3): Conv(
      (conv): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (act): SiLU()
    )
    (m): Sequential(
      (0): Bottleneck(
        (conv1): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (conv2): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
      )
      (1): Bottleneck(
        (conv1): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (conv2): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
      )
      (2): Bottleneck(
        (conv1): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
        (conv2): Conv(
          (conv): Conv2d(32, 32, kernel_size=(1, 1), stride=(1, 1), bias=False)
          (bn): BatchNorm2d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
          (act): SiLU()
        )
      )
    )
  )
  (classifier): Sequential(
    (0): Linear(in_features=802816, out_features=100, bias=True)
    (1): ReLU()
    (2): Linear(in_features=100, out_features=4, bias=True)
  )
)

import torchsummary as summary
model.to(device)
summary.summary(model, (3, 224, 224))

----------------------------------------------------------------
        Layer (type)               Output Shape         Param #
================================================================
            Conv2d-1         [-1, 32, 112, 112]             864
       BatchNorm2d-2         [-1, 32, 112, 112]              64
              SiLU-3         [-1, 32, 112, 112]               0
              Conv-4         [-1, 32, 112, 112]               0
            Conv2d-5         [-1, 32, 112, 112]           1,024
       BatchNorm2d-6         [-1, 32, 112, 112]              64
              SiLU-7         [-1, 32, 112, 112]               0
              Conv-8         [-1, 32, 112, 112]               0
            Conv2d-9         [-1, 32, 112, 112]           1,024
      BatchNorm2d-10         [-1, 32, 112, 112]              64
             SiLU-11         [-1, 32, 112, 112]               0
             Conv-12         [-1, 32, 112, 112]               0
           Conv2d-13         [-1, 32, 112, 112]           1,024
      BatchNorm2d-14         [-1, 32, 112, 112]              64
             SiLU-15         [-1, 32, 112, 112]               0
             Conv-16         [-1, 32, 112, 112]               0
       Bottleneck-17         [-1, 32, 112, 112]               0
           Conv2d-18         [-1, 32, 112, 112]           1,024
      BatchNorm2d-19         [-1, 32, 112, 112]              64
             SiLU-20         [-1, 32, 112, 112]               0
             Conv-21         [-1, 32, 112, 112]               0
           Conv2d-22         [-1, 32, 112, 112]           1,024
      BatchNorm2d-23         [-1, 32, 112, 112]              64
             SiLU-24         [-1, 32, 112, 112]               0
             Conv-25         [-1, 32, 112, 112]               0
       Bottleneck-26         [-1, 32, 112, 112]               0
           Conv2d-27         [-1, 32, 112, 112]           1,024
      BatchNorm2d-28         [-1, 32, 112, 112]              64
             SiLU-29         [-1, 32, 112, 112]               0
             Conv-30         [-1, 32, 112, 112]               0
           Conv2d-31         [-1, 32, 112, 112]           1,024
      BatchNorm2d-32         [-1, 32, 112, 112]              64
             SiLU-33         [-1, 32, 112, 112]               0
             Conv-34         [-1, 32, 112, 112]               0
       Bottleneck-35         [-1, 32, 112, 112]               0
           Conv2d-36         [-1, 32, 112, 112]           1,024
      BatchNorm2d-37         [-1, 32, 112, 112]              64
             SiLU-38         [-1, 32, 112, 112]               0
             Conv-39         [-1, 32, 112, 112]               0
           Conv2d-40         [-1, 64, 112, 112]           4,096
      BatchNorm2d-41         [-1, 64, 112, 112]             128
             SiLU-42         [-1, 64, 112, 112]               0
             Conv-43         [-1, 64, 112, 112]               0
               C3-44         [-1, 64, 112, 112]               0
           Linear-45                  [-1, 100]      80,281,700
             ReLU-46                  [-1, 100]               0
           Linear-47                    [-1, 4]             404
================================================================
Total params: 80,295,960
Trainable params: 80,295,960
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.57
Forward/backward pass size (MB): 150.06
Params size (MB): 306.30
Estimated Total Size (MB): 456.94
----------------------------------------------------------------

三、 训练模型

1. 设置超参数

2. 编写训练函数

1. optimizer.zero_grad()

函数会遍历模型的所有参数，通过内置方法截断反向传播的梯度流，再将每个参数的梯度值设为0，即上一次的梯度记录被清空。

2. loss.backward()

PyTorch的反向传播(即tensor.backward())是通过autograd包来实现的，autograd包会根据tensor进行过的数学运算来自动计算其对应的梯度。

具体来说，torch.tensor是autograd包的基础类，如果你设置tensor的requires_grads为True，就会开始跟踪这个tensor上面的所有运算，如果你做完运算后使用tensor.backward()，所有的梯度就会自动运算，tensor的梯度将会累加到它的.grad属性里面去。

更具体地说，损失函数loss是由模型的所有权重w经过一系列运算得到的，若某个w的requires_grads为True，则w的所有上层参数（后面层的权重w）的.grad_fn属性中就保存了对应的运算，然后在使用loss.backward()后，会一层层的反向传播计算每个w的梯度值，并保存到该w的.grad属性中。

如果没有进行tensor.backward()的话，梯度值将会是None，因此loss.backward()要写在optimizer.step()之前。

3. optimizer.step()

step()函数的作用是执行一次优化步骤，通过梯度下降法来更新参数的值。因为梯度下降是基于梯度的，所以在执行optimizer.step()函数前应先执行loss.backward()函数来计算梯度。

注意：optimizer只负责通过梯度下降进行优化，而不负责产生梯度，梯度是tensor.backward()方法产生的。

# 训练循环
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

3. 编写测试函数

测试函数和训练函数大致相同，但是由于不进行梯度下降对网络权重进行更新，所以不需要传入优化器

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

4. 正式训练

1. model.train()

model.train()的作用是启用 Batch Normalization 和 Dropout。

如果模型中有BN层(Batch Normalization）和Dropout，需要在训练时添加model.train()。model.train()是保证BN层能够用到每一批数据的均值和方差。对于Dropout，model.train()是随机取一部分网络连接来训练更新参数。

2. model.eval()

model.eval()的作用是不启用 Batch Normalization 和 Dropout。

如果模型中有BN层(Batch Normalization）和Dropout，在测试时添加model.eval()。model.eval()是保证BN层能够用全部训练数据的均值和方差，即测试过程中要保证BN层的均值和方差不变。对于Dropout，model.eval()是利用到了所有网络连接，即不进行随机舍弃神经元。

训练完train样本后，生成的模型model要用来测试样本。在model(test)之前，需要加上model.eval()，否则的话，有输入数据，即使不训练，它也会改变权值。这是model中含有BN层和Dropout所带来的的性质。

#SGD调优#

epochs     = 20
train_loss = []
train_acc  = []
test_loss  = []
test_acc   = []
torch.manual_seed(42)
model=C3_net()
model.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')

#Adam调优#

epochs     = 20
train_loss = []
train_acc  = []
test_loss  = []
test_acc   = []
torch.manual_seed(42)
model=C3_net()
model.to(device)
loss_fn    = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-4 # 学习率
opt        = torch.optim.Adam(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)
    
    # 保存最佳模型到 best_model
    if epoch_test_acc > best_acc:
        best_acc   = epoch_test_acc
        best_model = copy.deepcopy(model)
        
    train_acc.append(epoch_train_acc)
    train_loss.append(epoch_train_loss)
    test_acc.append(epoch_test_acc)
    test_loss.append(epoch_test_loss)
    
   
    # 获取当前的学习率
    lr = optimizer.state_dict()['param_groups'][0]['lr']
    
    template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%, Test_loss:{:.3f}, Lr:{:.2E}')
    print(template.format(epoch+1, epoch_train_acc*100, epoch_train_loss, 
                          epoch_test_acc*100, epoch_test_loss, lr))
    
# 保存最佳模型到文件中
PATH = './best_model.pth'  # 保存的参数文件名
torch.save(model.state_dict(), PATH)

print('Done')

Epoch: 1, Train_acc:76.7%, Train_loss:1.097, Test_acc:56.0%，Test_loss:2.127
Epoch: 2, Train_acc:92.3%, Train_loss:0.230, Test_acc:83.6%，Test_loss:0.718
Epoch: 3, Train_acc:97.6%, Train_loss:0.178, Test_acc:89.3%，Test_loss:0.359
Epoch: 4, Train_acc:97.2%, Train_loss:0.114, Test_acc:86.7%，Test_loss:0.538
Epoch: 5, Train_acc:97.4%, Train_loss:0.214, Test_acc:84.9%，Test_loss:0.556
Epoch: 6, Train_acc:97.6%, Train_loss:0.096, Test_acc:88.0%，Test_loss:0.541
Epoch: 7, Train_acc:98.7%, Train_loss:0.047, Test_acc:88.0%，Test_loss:0.469
Epoch: 8, Train_acc:99.4%, Train_loss:0.042, Test_acc:85.3%，Test_loss:0.627
Epoch: 9, Train_acc:99.3%, Train_loss:0.016, Test_acc:89.3%，Test_loss:0.469
Epoch:10, Train_acc:99.7%, Train_loss:0.009, Test_acc:88.4%，Test_loss:0.571
Epoch:11, Train_acc:99.9%, Train_loss:0.008, Test_acc:88.0%，Test_loss:0.501
Epoch:12, Train_acc:99.9%, Train_loss:0.007, Test_acc:88.9%，Test_loss:0.436
Epoch:13, Train_acc:99.9%, Train_loss:0.005, Test_acc:90.2%，Test_loss:0.513
Epoch:14, Train_acc:99.6%, Train_loss:0.026, Test_acc:88.4%，Test_loss:1.820
Epoch:15, Train_acc:97.8%, Train_loss:0.076, Test_acc:87.6%，Test_loss:0.887
Epoch:16, Train_acc:99.8%, Train_loss:0.007, Test_acc:89.3%，Test_loss:0.686
Epoch:17, Train_acc:99.4%, Train_loss:0.008, Test_acc:90.2%，Test_loss:0.597
Epoch:18, Train_acc:99.0%, Train_loss:0.071, Test_acc:92.4%，Test_loss:0.424
Epoch:19, Train_acc:99.1%, Train_loss:0.056, Test_acc:88.9%，Test_loss:0.662
Epoch:20, Train_acc:99.2%, Train_loss:0.044, Test_acc:88.4%，Test_loss:0.712
Done

修改优化

epochs     = 20
train_loss = []
train_acc  = []
test_loss  = []
test_acc   = []
torch.manual_seed(42)
model=C3_net()
model.to(device)
loss_fn    = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 5e-4 # 学习率
opt        = torch.optim.SGD(model.parameters(),lr=learn_rate,momentum=3e-4, weight_decay=5e-4)
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')

Epoch: 1, Train_acc:78.3%, Train_loss:0.633, Test_acc:67.1%，Test_loss:0.764
Epoch: 2, Train_acc:86.2%, Train_loss:0.369, Test_acc:86.2%，Test_loss:0.350
Epoch: 3, Train_acc:91.8%, Train_loss:0.285, Test_acc:86.2%，Test_loss:0.361
Epoch: 4, Train_acc:93.1%, Train_loss:0.249, Test_acc:82.2%，Test_loss:0.402
Epoch: 5, Train_acc:95.0%, Train_loss:0.188, Test_acc:86.2%，Test_loss:0.316
Epoch: 6, Train_acc:96.1%, Train_loss:0.156, Test_acc:86.2%，Test_loss:0.364
Epoch: 7, Train_acc:97.7%, Train_loss:0.127, Test_acc:88.4%，Test_loss:0.317
Epoch: 8, Train_acc:97.9%, Train_loss:0.133, Test_acc:85.3%，Test_loss:0.442
Epoch: 9, Train_acc:97.6%, Train_loss:0.121, Test_acc:85.3%，Test_loss:0.341
Epoch:10, Train_acc:98.4%, Train_loss:0.122, Test_acc:86.2%，Test_loss:0.374
Epoch:11, Train_acc:98.9%, Train_loss:0.086, Test_acc:88.9%，Test_loss:0.276
Epoch:12, Train_acc:98.7%, Train_loss:0.083, Test_acc:89.3%，Test_loss:0.271
Epoch:13, Train_acc:99.2%, Train_loss:0.081, Test_acc:90.2%，Test_loss:0.280
Epoch:14, Train_acc:98.6%, Train_loss:0.087, Test_acc:88.4%，Test_loss:0.390
Epoch:15, Train_acc:98.7%, Train_loss:0.077, Test_acc:92.0%，Test_loss:0.272
Epoch:16, Train_acc:99.3%, Train_loss:0.055, Test_acc:90.7%，Test_loss:0.389
Epoch:17, Train_acc:99.2%, Train_loss:0.052, Test_acc:90.2%，Test_loss:0.264
Epoch:18, Train_acc:99.6%, Train_loss:0.049, Test_acc:89.3%，Test_loss:0.271
Epoch:19, Train_acc:99.1%, Train_loss:0.054, Test_acc:89.3%，Test_loss:0.261
Epoch:20, Train_acc:99.3%, Train_loss:0.063, Test_acc:88.4%，Test_loss:0.324
Done

SGD会下降的比较慢且容易遇到局部最小值导致训练无法继续向下，通过增加momentum，稍微调大学习率可以使训练更好的进行，增加weight_decay是为了减少过拟合

epochs     = 20
train_loss = []
train_acc  = []
test_loss  = []
test_acc   = []
torch.manual_seed(42)
model=C3_net()
model.to(device)
loss_fn    = nn.CrossEntropyLoss() # 创建损失函数
learn_rate = 1e-4 # 学习率
#opt        = torch.optim.Adam(model.parameters(),lr=learn_rate)

opt        =torch.optim.Adam(model.parameters(),
                lr=0.0001,
                betas=(0.9, 0.999),
                eps=1e-08,
                weight_decay=0,
                amsgrad=False)



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')

Epoch: 1, Train_acc:77.6%, Train_loss:1.456, Test_acc:79.1%，Test_loss:0.683
Epoch: 2, Train_acc:91.2%, Train_loss:0.264, Test_acc:85.8%，Test_loss:0.389
Epoch: 3, Train_acc:97.6%, Train_loss:0.093, Test_acc:90.7%，Test_loss:0.335
Epoch: 4, Train_acc:98.4%, Train_loss:0.176, Test_acc:90.2%，Test_loss:0.359
Epoch: 5, Train_acc:98.0%, Train_loss:0.060, Test_acc:85.3%，Test_loss:0.539
Epoch: 6, Train_acc:98.4%, Train_loss:0.041, Test_acc:85.8%，Test_loss:0.484
Epoch: 7, Train_acc:99.7%, Train_loss:0.021, Test_acc:86.7%，Test_loss:0.479
Epoch: 8, Train_acc:99.8%, Train_loss:0.047, Test_acc:88.4%，Test_loss:0.515
Epoch: 9, Train_acc:98.2%, Train_loss:0.081, Test_acc:86.7%，Test_loss:0.550
Epoch:10, Train_acc:99.4%, Train_loss:0.017, Test_acc:88.0%，Test_loss:0.461
Epoch:11, Train_acc:99.9%, Train_loss:0.003, Test_acc:88.0%，Test_loss:0.403
Epoch:12, Train_acc:99.8%, Train_loss:0.007, Test_acc:87.6%，Test_loss:0.468
Epoch:13, Train_acc:99.3%, Train_loss:0.027, Test_acc:83.1%，Test_loss:0.805
Epoch:14, Train_acc:99.7%, Train_loss:0.013, Test_acc:85.8%，Test_loss:0.893
Epoch:15, Train_acc:98.0%, Train_loss:0.165, Test_acc:83.1%，Test_loss:0.753
Epoch:16, Train_acc:97.6%, Train_loss:0.139, Test_acc:84.9%，Test_loss:1.345
Epoch:17, Train_acc:99.1%, Train_loss:0.050, Test_acc:85.3%，Test_loss:0.765
Epoch:18, Train_acc:99.6%, Train_loss:0.011, Test_acc:86.7%，Test_loss:0.695
Epoch:19, Train_acc:99.7%, Train_loss:0.014, Test_acc:85.8%，Test_loss:0.668
Epoch:20, Train_acc:99.7%, Train_loss:0.172, Test_acc:88.4%，Test_loss:0.691
Done

四、 结果可视化

import matplotlib.pyplot as plt
#隐藏警告
import warnings
warnings.filterwarnings("ignore")               #忽略警告信息
plt.rcParams['font.sans-serif']    = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False      # 用来正常显示负号
plt.rcParams['figure.dpi']         = 100        #分辨率

epochs_range = range(epochs)

plt.figure(figsize=(12, 3))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, train_acc, label='Training Accuracy')
plt.plot(epochs_range, test_acc, label='Test Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Training Loss')
plt.plot(epochs_range, test_loss, label='Test Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()

findfont: Font family ['sans-serif'] not found. Falling back to DejaVu Sans.
findfont: Generic family 'sans-serif' not found because none of the following families were found: SimHei
findfont: Font family ['sans-serif'] not found. Falling back to DejaVu Sans.
findfont: Generic family 'sans-serif' not found because none of the following families were found: SimHei

模型评估

model.eval()
epoch_test_acc, epoch_test_loss = test(test_dl, model, loss_fn)

epoch_test_acc, epoch_test_loss

(0.8844444444444445, 0.7120169894769788)

五 识别本地图片

local_test_image = PIL.Image.open ("/deep learning/p3/p3_testdata/r1.jpg").convert('RGB')
#local_test_data = torchvision.transforms.functional.resize(local_test_data,[224,224])
local_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
])
local_test_data = local_transforms(local_test_image)
PIL.Image.open ("/deep learning/p3/p3_testdata/r1.jpg").convert('RGB')##显示照片

_,result=torch.max(model(local_test_data.to(device).unsqueeze(0)),1)
classeNames[result]

'rain'

优化之后的模型预测效果并不好，将雨预测为云，用Adam效果最好