第3周学习：ResNet+ResNeXt

（一）论文阅读

一、ResNet网络结构

• 网络中的亮点

  • 超深的网络结构（突破1000层）
    • 梯度消失或梯度爆炸：通过对数据标准化；权重初始化；BN解决
    • 退化问题：引入残差解决
  • 提出residual（残差）模块
  • 使用Batch Normalization加速训练（丢弃dropout）

• residual结构：

  1. 针对层数少的
  2. 针对层数多的

• 残差结构里右侧半包围的箭头线如果是虚线，则图片输入和输出大小不同（高、宽、深度），为了实现不同，步长会发生相应改变

若为实线则相同；

• 主分支与侧分支的输出形状应该相同，以实现图形相加

可以通过1x1的卷积核来降维和升维，通过改变步长来对输入图像的长宽进行缩放

（在原论文中第一个1*1卷积层的步长是2，第二个3*3步长是1，但在pytorch的实现过程中第一个1*1的卷积层步长是1，第二个3*3步长是2，这能够在image net的top1上提升大概0.5%的准确率）

• Batch Normalization：

通过该方法能够加速网络的收敛并提升准确率。

• 原理：我们在图像预处理过程中通常会对图像进行标准化处理，这样能够加速网络的收敛，如下图所示，对于Conv1来说输入的就是满足某一分布的特征矩阵，但对于Conv2而言输入的feature map就不一定满足某一分布规律了（注意这里所说满足某一分布规律并不是指某一个feature map的数据要满足分布规律，理论上是指整个训练样本集所对应feature map的数据要满足分布规律）。而我们Batch Normalization的目的就是使我们的feature map满足均值为0，方差为1的分布规律。

                       

• 我们刚刚有说让feature map满足某一分布规律，理论上是指整个训练样本集所对应feature map的数据要满足分布规律，也就是说要计算出整个训练集的feature map然后在进行标准化处理，对于一个大型的数据集明显是不可能的，所以论文中说的是Batch Normalization，也就是我们计算一个Batch数据的feature map然后在进行标准化（batch越大越接近整个数据集的分布，效果越好）。

• pytorch实验注意问题：
  1. 训练时要将traning参数设置为True，在验证时将trainning参数设置为False。在pytorch中可通过创建模型的model.train()和model.eval()方法控制。
  2. batch size尽可能设置大点，设置小后表现可能很糟糕，设置的越大求的均值和方差越接近整个训练集的均值和方差。
  3. 建议将bn层放在卷积层（Conv）和激活层（例如Relu）之间，且卷积层不要使用偏置bias，因为没有用，参考下图推理，即使使用了偏置bias求出的结果也是一样的

• 迁移学习：

  • 优势：
    1. 能够快速的训练出一个理想的结果
    2. 当数据集较小的时候也能训练出理想的效果
  • 原理：将已有网络训练好的浅层参数进行迁移，以此使新的网络直接获得底层通用网络信息，以期更加快速的完成新网络训练
  • 常见的迁移学习方式：
    1. 载入权重后训练所有参数
    2. 载入权重后只训练最后几层
    3. 载入权重后在原网络基础上在添加一层全连接层，仅训练最后一个全连接层

二、ResNeXt网络：ResNet的升级

• 对残差结构进行了更新（左边为原本的残差结构，右边为升级后的）

• 通过这样的改变，原模型在各项性能中均有提升，在相同的计算量下，错误率更低
• 组卷积：将原本的输入分为g个组，参数个数将会变为原本的1/g

• 升级后的残差网络就是加入了组卷积，以上三幅图的内容在数学计算上是等价的

• 仅在层数>3时才有意义，否则与普通卷积没有什么区别

（二）代码作业 

 一、lenet实现猫狗大战

模型如下 ：

 经过几轮训练，损失率逐渐下降:

 

二、resnet实现猫狗大战

本次训练采用的网络结构是ResNet34浅层残差结构模型如下：

import torch.nn as nn
import torch


class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, in_channel, out_channel, stride=1, downsample=None, **kwargs):
        #输入特征矩阵深度，输出特征矩阵深度，步长，下采样参数对应虚线残差结构
        super(BasicBlock, self).__init__()
        self.conv1 = nn.Conv2d(in_channels=in_channel, out_channels=out_channel,
                               kernel_size=3, stride=stride, padding=1, bias=False)
        #卷积核大小kernel_size，bias为偏置，BN不需要使用
        self.bn1 = nn.BatchNorm2d(out_channel)
        self.relu = nn.ReLU()
        self.conv2 = nn.Conv2d(in_channels=out_channel, out_channels=out_channel,
                               kernel_size=3, stride=1, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(out_channel)
        self.downsample = downsample

    def forward(self, x):
        identity = x
        if self.downsample is not None:#if=none对应实线部分则跳过，否则经过下采样，得到输出
            identity = self.downsample(x)

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        out += identity
        out = self.relu(out)

        return out


class Bottleneck(nn.Module):
    """
    注意：原论文中，在虚线残差结构的主分支上，第一个1x1卷积层的步距是2，第二个3x3卷积层步距是1。
    但在pytorch官方实现过程中是第一个1x1卷积层的步距是1，第二个3x3卷积层步距是2，
    这么做的好处是能够在top1上提升大概0.5%的准确率。
    可参考Resnet v1.5 https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_pytorch
    """
    expansion = 4

    def __init__(self, in_channel, out_channel, stride=1, downsample=None,
                 groups=1, width_per_group=64):
        super(Bottleneck, self).__init__()

        width = int(out_channel * (width_per_group / 64.)) * groups

        self.conv1 = nn.Conv2d(in_channels=in_channel, out_channels=width,
                               kernel_size=1, stride=1, bias=False)  # squeeze channels
        self.bn1 = nn.BatchNorm2d(width)
        # -----------------------------------------
        self.conv2 = nn.Conv2d(in_channels=width, out_channels=width, groups=groups,
                               kernel_size=3, stride=stride, bias=False, padding=1)
        self.bn2 = nn.BatchNorm2d(width)
        # -----------------------------------------
        self.conv3 = nn.Conv2d(in_channels=width, out_channels=out_channel*self.expansion,
                               kernel_size=1, stride=1, bias=False)  # unsqueeze channels
        self.bn3 = nn.BatchNorm2d(out_channel*self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample

    def forward(self, x):
        identity = x
        if self.downsample is not None:
            identity = self.downsample(x)

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        out = self.relu(out)

        out = self.conv3(out)
        out = self.bn3(out)

        out += identity
        out = self.relu(out)

        return out


class ResNet(nn.Module):

    def __init__(self,
                 block,#残差结构，basicblock或bottleneck
                 blocks_num,#一个列表参数对应每个残差结构个数
                 num_classes=1000,#训练集分类个数
                 include_top=True,
                 groups=1,
                 width_per_group=64):
        super(ResNet, self).__init__()
        self.include_top = include_top
        self.in_channel = 64

        self.groups = groups
        self.width_per_group = width_per_group

        self.conv1 = nn.Conv2d(3, self.in_channel, kernel_size=7, stride=2,
                               padding=3, bias=False)
        self.bn1 = nn.BatchNorm2d(self.in_channel)
        self.relu = nn.ReLU(inplace=True)#以上为第一层
        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)#第二层
        self.layer1 = self._make_layer(block, 64, blocks_num[0])
        self.layer2 = self._make_layer(block, 128, blocks_num[1], stride=2)
        self.layer3 = self._make_layer(block, 256, blocks_num[2], stride=2)
        self.layer4 = self._make_layer(block, 512, blocks_num[3], stride=2)
        if self.include_top:
            self.avgpool = nn.AdaptiveAvgPool2d((1, 1))  # output size = (1, 1)
            self.fc = nn.Linear(512 * block.expansion, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')

    def _make_layer(self, block, channel, block_num, stride=1):
        downsample = None
        if stride != 1 or self.in_channel != channel * block.expansion:#18层，34层会直接跳过
            downsample = nn.Sequential(
                nn.Conv2d(self.in_channel, channel * block.expansion, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(channel * block.expansion))

        layers = []
        layers.append(block(self.in_channel,
                            channel,
                            downsample=downsample,
                            stride=stride,
                            groups=self.groups,
                            width_per_group=self.width_per_group))
        self.in_channel = channel * block.expansion

        for _ in range(1, block_num):
            layers.append(block(self.in_channel,
                                channel,
                                groups=self.groups,
                                width_per_group=self.width_per_group))

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)

        if self.include_top:
            x = self.avgpool(x)
            x = torch.flatten(x, 1)
            x = self.fc(x)

        return x


def resnet34(num_classes=1000, include_top=True):
    # https://download.pytorch.org/models/resnet34-333f7ec4.pth
    return ResNet(BasicBlock, [3, 4, 6, 3], num_classes=num_classes, include_top=include_top)


def resnet50(num_classes=1000, include_top=True):
    # https://download.pytorch.org/models/resnet50-19c8e357.pth
    return ResNet(Bottleneck, [3, 4, 6, 3], num_classes=num_classes, include_top=include_top)


def resnet101(num_classes=1000, include_top=True):
    # https://download.pytorch.org/models/resnet101-5d3b4d8f.pth
    return ResNet(Bottleneck, [3, 4, 23, 3], num_classes=num_classes, include_top=include_top)


def resnext50_32x4d(num_classes=1000, include_top=True):
    # https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pth
    groups = 32
    width_per_group = 4
    return ResNet(Bottleneck, [3, 4, 6, 3],
                  num_classes=num_classes,
                  include_top=include_top,
                  groups=groups,
                  width_per_group=width_per_group)


def resnext101_32x8d(num_classes=1000, include_top=True):
    # https://download.pytorch.org/models/resnext101_32x8d-8ba56ff5.pth
    groups = 32
    width_per_group = 8
    return ResNet(Bottleneck, [3, 4, 23, 3],
                  num_classes=num_classes,
                  include_top=include_top,
                  groups=groups,
                  width_per_group=width_per_group)

训练如下：

import os
import sys
import json

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

from model import resnet34
import torchvision.models.resnet

def main():
    device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
    print("using {} device.".format(device))

    data_transform = {
        "train": transforms.Compose([transforms.RandomResizedCrop(224),
                                     transforms.RandomHorizontalFlip(),
                                     transforms.ToTensor(),
                                     transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])]),
        "val": transforms.Compose([transforms.Resize(256),
                                   transforms.CenterCrop(224),
                                   transforms.ToTensor(),
                                   transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])}

    data_root = os.path.abspath(os.path.join(os.getcwd(), "D:\PycharmProject"))  # get data root path
    image_path = os.path.join(data_root, "data_set","cat_dog_data")  # flower data set path
    assert os.path.exists(image_path), "{} path does not exist.".format(image_path)
    train_dataset = datasets.ImageFolder(root=os.path.join(image_path, "train"),
                                         transform=data_transform["train"])
    train_num = len(train_dataset)

    # {'cat':0, 'dog':1}
    flower_list = train_dataset.class_to_idx
    cla_dict = dict((val, key) for key, val in flower_list.items())
    # write dict into json file
    json_str = json.dumps(cla_dict, indent=4)
    with open('class_indices.json', 'w') as json_file:
        json_file.write(json_str)

    batch_size = 16
    nw = min([os.cpu_count(), batch_size if batch_size > 1 else 0, 8])  # number of workers
    print('Using {} dataloader workers every process'.format(nw))

    train_loader = torch.utils.data.DataLoader(train_dataset,
                                               batch_size=batch_size, shuffle=True,
                                               num_workers=nw)

    validate_dataset = datasets.ImageFolder(root=os.path.join(image_path, "val"),
                                            transform=data_transform["val"])
    val_num = len(validate_dataset)
    validate_loader = torch.utils.data.DataLoader(validate_dataset,
                                                  batch_size=batch_size, shuffle=False,
                                                  num_workers=nw)

    print("using {} images for training, {} images for validation.".format(train_num,
                                                                           val_num))

    net = resnet34()
    # load pretrain weights
    # download url: https://download.pytorch.org/models/resnet34-333f7ec4.pth
    model_weight_path = "./resnet34-pre.pth"
    assert os.path.exists(model_weight_path), "file {} does not exist.".format(model_weight_path)
    net.load_state_dict(torch.load(model_weight_path, map_location='cpu'));
    for param in net.parameters():
        param.requires_grad = False

    # change fc layer structure
    in_channel = net.fc.in_features
    net.fc = nn.Linear(in_channel, 2)
    net.to(device)

    # define loss function
    loss_function = nn.CrossEntropyLoss()

    # construct an optimizer
    params = [p for p in net.parameters() if p.requires_grad]
    optimizer = optim.Adam(params, lr=0.0001)

    epochs = 3
    best_acc = 0.0
    save_path = './resNet34.pth'
    train_steps = len(train_loader)
    for epoch in range(epochs):
        # train
        net.train()
        running_loss = 0.0
        train_bar = tqdm(train_loader, file=sys.stdout)
        for step, data in enumerate(train_bar):
            images, labels = data
            optimizer.zero_grad()
            logits = net(images.to(device))
            loss = loss_function(logits, labels.to(device))
            loss.backward()
            optimizer.step()

            # print statistics
            running_loss += loss.item()

            train_bar.desc = "train epoch[{}/{}] loss:{:.3f}".format(epoch + 1,
                                                                     epochs,
                                                                     loss)

        # validate
        net.eval()
        acc = 0.0  # accumulate accurate number / epoch
        with torch.no_grad():
            val_bar = tqdm(validate_loader, file=sys.stdout)
            for val_data in val_bar:
                val_images, val_labels = val_data
                outputs = net(val_images.to(device))
                # loss = loss_function(outputs, test_labels)
                predict_y = torch.max(outputs, dim=1)[1]
                acc += torch.eq(predict_y, val_labels.to(device)).sum().item()

                val_bar.desc = "valid epoch[{}/{}]".format(epoch + 1,
                                                           epochs)

        val_accurate = acc / val_num
        print('[epoch %d] train_loss: %.3f  val_accuracy: %.3f' %
              (epoch + 1, running_loss / train_steps, val_accurate))

        if val_accurate > best_acc:
            best_acc = val_accurate
            torch.save(net.state_dict(), save_path)

    print('Finished Training')


if __name__ == '__main__':
    main()

利用迁移学习的方法，由于训练集较大，有20000张照片而设备以及时间有限，我只进行了一轮训练结果如下

 

 可以看到在验证集上准确率达到了0.968，较为准确

测试结果如下：

import json
import os

import torch
from PIL import Image
from torchvision import transforms

from model import resnet34

import csv #调用数据保存文件
import pandas as pd #用于数据输出
def main():
    device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

    data_transform = transforms.Compose(
        [transforms.Resize(256),
         transforms.CenterCrop(224),
         transforms.ToTensor(),
         transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])

    # load image
    # 指向需要遍历预测的图像文件夹
    imgs_root = "D:/PycharmProject/data_set/cat_dog_data/test"
    assert os.path.exists(imgs_root), f"file: '{imgs_root}' dose not exist."
    # 读取指定文件夹下所有jpg图像路径
    img_path_list = [os.path.join(imgs_root, i) for i in os.listdir(imgs_root) if i.endswith(".jpg")]

    # read class_indict
    json_path = './class_indices.json'
    assert os.path.exists(json_path), f"file: '{json_path}' dose not exist."

    json_file = open(json_path, "r")
    class_indict = json.load(json_file)

    # create model
    model = resnet34(num_classes=2).to(device)

    # load model weights
    weights_path = "./resNet34.pth"
    assert os.path.exists(weights_path), f"file: '{weights_path}' dose not exist."
    model.load_state_dict(torch.load(weights_path, map_location=device))
    list = [n for n in range(0, 2000)]
    # prediction
    model.eval()
    batch_size =8 # 每次预测时将多少张图片打包成一个batch
    with torch.no_grad():
        for ids in range(0, len(img_path_list) // batch_size):
            img_list = []
            for img_path in img_path_list[ids * batch_size: (ids + 1) * batch_size]:
                assert os.path.exists(img_path), f"file: '{img_path}' dose not exist."
                img = Image.open(img_path)
                img = data_transform(img)
                img_list.append(img)

            # batch img
            # 将img_list列表中的所有图像打包成一个batch
            batch_img = torch.stack(img_list, dim=0)
            # predict class
            output = model(batch_img.to(device)).cpu()
            predict = torch.softmax(output, dim=1)
            probs, classes = torch.max(predict, dim=1)

            for idx, (pro, cla) in enumerate(zip(probs, classes)):
                print("image: {}  class: {}  prob: {:.3}".format(img_path_list[ids * batch_size + idx],
                                                                 class_indict[str(cla.numpy())],
                                                                 pro.numpy()))

                if format(class_indict[str(cla.numpy())]) == "cat":
                    list[ids * batch_size + idx] = '0'
                else:
                    list[ids * batch_size + idx] = '1'
    test = pd.DataFrame(data=list)  # 将数据放进表格
    test.to_csv('D:\PycharmProject\data_set\cat_dog_data/test.csv')  # 数据存入csv,存储位置及文件名称


if __name__ == '__main__':
    main()

 

但提交后结果不是很理想，可能是为了节省时间训练的时候固定住了前面的参数，只训练了最后一层，加上训练轮次比较少，所以结果不理想，后续我还会继续尝试

 

(三)思考题

1、Residual learning

见上文
2、Batch Normailization 的原理

我们在图像预处理过程中通常会对图像进行标准化处理，这样能够加速网络的收敛，如下图所示，对于Conv1来说输入的就是满足某一分布的特征矩阵，但对于Conv2而言输入的feature map就不一定满足某一分布规律了（注意这里所说满足某一分布规律并不是指某一个feature map的数据要满足分布规律，理论上是指整个训练样本集所对应feature map的数据要满足分布规律）。而我们Batch Normalization的目的就是使我们的feature map满足均值为0，方差为1的分布规律。

3、为什么分组卷积可以提升准确率？即然分组卷积可以提升准确率，同时还能降低计算量，分数数量尽量多不行吗？ 

文章中提到分组卷积可以极大的降低计算量，同等情况下可以训练更深的网络，但是分组过后组与组之间的信息被分隔开来，所以不能分太多。