AlexNet卷积神经网络-笔记

AlexNet卷积神经网络-笔记
AlexNet卷积神经网络2012年提出

测试结果为：
通过运行结果可以发现，
在眼疾筛查数据集iChallenge-PM上使用AlexNet，loss能有效下降，
经过5个epoch的训练，在验证集上的准确率可以达到94%左右。

实测准确率为：0.92到0.9350
[validation] accuracy/loss: 0.9275/0.1661
[validation] accuracy/loss: 0.9350/0.2233

S E:\project\python> & D:/ProgramData/Anaconda3/python.exe e:/project/python/PM/AlexNet_PM.py
W0803 14:19:51.270619  6520 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 6.1, Driver API Version: 12.2, Runtime API Version: 10.2
W0803 14:19:51.290621  6520 gpu_resources.cc:91] device: 0, cuDNN Version: 7.6.
start training ... 
epoch: 0, batch_id: 0, loss is: 1.0486
epoch: 0, batch_id: 20, loss is: 0.5316
[validation] accuracy/loss: 0.9275/0.2720
epoch: 1, batch_id: 0, loss is: 0.2918
epoch: 1, batch_id: 20, loss is: 0.2479
[validation] accuracy/loss: 0.9250/0.3421
epoch: 2, batch_id: 0, loss is: 1.7486
epoch: 2, batch_id: 20, loss is: 0.1236
[validation] accuracy/loss: 0.9350/0.2233
epoch: 3, batch_id: 0, loss is: 0.2802
epoch: 3, batch_id: 20, loss is: 0.3339
[validation] accuracy/loss: 0.9275/0.2186
epoch: 4, batch_id: 0, loss is: 0.0429
epoch: 4, batch_id: 20, loss is: 0.1188
[validation] accuracy/loss: 0.9275/0.1661

W0803 14:34:45.152906 17400 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 6.1, Driver API Version: 12.2, Runtime API Version: 10.2
W0803 14:34:45.173938 17400 gpu_resources.cc:91] device: 0, cuDNN Version: 7.6.
#AlexNet 子图层结构
[Conv2D(3, 96, kernel_size=[11, 11], stride=[4, 4], padding=5, data_format=NCHW), 
MaxPool2D(kernel_size=2, stride=2, padding=0), 
Conv2D(96, 256, kernel_size=[5, 5], padding=2, data_format=NCHW), 
MaxPool2D(kernel_size=2, stride=2, padding=0), 
Conv2D(256, 384, kernel_size=[3, 3], padding=1, data_format=NCHW), 
Conv2D(384, 384, kernel_size=[3, 3], padding=1, data_format=NCHW), 
Conv2D(384, 256, kernel_size=[3, 3], padding=1, data_format=NCHW), 
MaxPool2D(kernel_size=2, stride=2, padding=0), 
Linear(in_features=12544, out_features=4096, dtype=float32), 
Dropout(p=0.5, axis=None, mode=upscale_in_train), 
Linear(in_features=4096, out_features=4096, dtype=float32), 
Dropout(p=0.5, axis=None, mode=upscale_in_train), 
Linear(in_features=4096, out_features=2, dtype=float32)]
(10, 3, 224, 224)
[10, 3, 224, 224]
#AlexNet子图层shape[N，Cout,H,W],w参数[Cout,Ci,Kh,Kw]，b参数[Cout]
conv2d_5 [10, 96, 56, 56] [96, 3, 11, 11] [96]
max_pool2d_3 [10, 96, 28, 28]
conv2d_6 [10, 256, 28, 28] [256, 96, 5, 5] [256]
max_pool2d_4 [10, 256, 14, 14]
conv2d_7 [10, 384, 14, 14] [384, 256, 3, 3] [384]
conv2d_8 [10, 384, 14, 14] [384, 384, 3, 3] [384]
conv2d_9 [10, 256, 14, 14] [256, 384, 3, 3] [256]
max_pool2d_5 [10, 256, 7, 7]
linear_3 [10, 4096] [12544, 4096] [4096]
dropout_2 [10, 4096]
linear_4 [10, 4096] [4096, 4096] [4096]
dropout_3 [10, 4096]
linear_5 [10, 2] [4096, 2] [2]
PS E:\project\python> 

注意：
conv2d_5 [10, 96, 56, 56] [96, 3, 11, 11] [96]
中H=56,W=56的计算方法如下：
H=((Hold+2P-K)/S)+1=((224+2*5-11)/4)+1=56.75=>56
同理W=56
测试源代码如下所示：

#AlexNet在眼疾筛查数据集iChallenge-PM上具体实现的代码如下所示：
# -*- coding:utf-8 -*-

# 导入需要的包
import paddle
import numpy as np
from paddle.nn import Conv2D, MaxPool2D, Linear, Dropout
## 组网
import paddle.nn.functional as F

# 定义 AlexNet 网络结构  2012年
class AlexNet(paddle.nn.Layer):
    def __init__(self, num_classes=1):
        super(AlexNet, self).__init__()
        # AlexNet与LeNet一样也会同时使用卷积和池化层提取图像特征
        # 与LeNet不同的是激活函数换成了‘relu’
        self.conv1 = Conv2D(in_channels=3, out_channels=96, kernel_size=11, stride=4, padding=5)
        self.max_pool1 = MaxPool2D(kernel_size=2, stride=2)
        self.conv2 = Conv2D(in_channels=96, out_channels=256, kernel_size=5, stride=1, padding=2)
        self.max_pool2 = MaxPool2D(kernel_size=2, stride=2)
        self.conv3 = Conv2D(in_channels=256, out_channels=384, kernel_size=3, stride=1, padding=1)
        self.conv4 = Conv2D(in_channels=384, out_channels=384, kernel_size=3, stride=1, padding=1)
        self.conv5 = Conv2D(in_channels=384, out_channels=256, kernel_size=3, stride=1, padding=1)
        self.max_pool5 = MaxPool2D(kernel_size=2, stride=2)

        self.fc1 = Linear(in_features=12544, out_features=4096)
        self.drop_ratio1 = 0.5
        self.drop1 = Dropout(self.drop_ratio1)
        self.fc2 = Linear(in_features=4096, out_features=4096)
        self.drop_ratio2 = 0.5
        self.drop2 = Dropout(self.drop_ratio2)
        self.fc3 = Linear(in_features=4096, out_features=num_classes)
    
    def forward(self, x):
        x = self.conv1(x)
        x = F.relu(x)
        x = self.max_pool1(x)
        x = self.conv2(x)
        x = F.relu(x)
        x = self.max_pool2(x)
        x = self.conv3(x)
        x = F.relu(x)
        x = self.conv4(x)
        x = F.relu(x)
        x = self.conv5(x)
        x = F.relu(x)
        x = self.max_pool5(x)
        x = paddle.reshape(x, [x.shape[0], -1])
        x = self.fc1(x)
        x = F.relu(x)
        # 在全连接之后使用dropout抑制过拟合
        x = self.drop1(x)
        x = self.fc2(x)
        x = F.relu(x)
        # 在全连接之后使用dropout抑制过拟合
        x = self.drop2(x)
        x = self.fc3(x)
        return x
#数据处理
#==============================================================================================
import cv2
import random
import numpy as np
import os

# 对读入的图像数据进行预处理
def transform_img(img):
    # 将图片尺寸缩放道 224x224
    img = cv2.resize(img, (224, 224))
    # 读入的图像数据格式是[H, W, C]
    # 使用转置操作将其变成[C, H, W]
    img = np.transpose(img, (2,0,1))
    img = img.astype('float32')
    # 将数据范围调整到[-1.0, 1.0]之间
    img = img / 255.
    img = img * 2.0 - 1.0
    return img

# 定义训练集数据读取器
def data_loader(datadir, batch_size=10, mode = 'train'):
    # 将datadir目录下的文件列出来，每条文件都要读入
    filenames = os.listdir(datadir)
    def reader():
        if mode == 'train':
            # 训练时随机打乱数据顺序
            random.shuffle(filenames)
        batch_imgs = []
        batch_labels = []
        for name in filenames:
            filepath = os.path.join(datadir, name)
            img = cv2.imread(filepath)
            img = transform_img(img)
            if name[0] == 'H' or name[0] == 'N':
                # H开头的文件名表示高度近似，N开头的文件名表示正常视力
                # 高度近视和正常视力的样本，都不是病理性的，属于负样本，标签为0
                label = 0
            elif name[0] == 'P':
                # P开头的是病理性近视，属于正样本，标签为1
                label = 1
            else:
                raise('Not excepted file name')
            # 每读取一个样本的数据，就将其放入数据列表中
            batch_imgs.append(img)
            batch_labels.append(label)
            if len(batch_imgs) == batch_size:
                # 当数据列表的长度等于batch_size的时候，
                # 把这些数据当作一个mini-batch，并作为数据生成器的一个输出
                imgs_array = np.array(batch_imgs).astype('float32')
                labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
                yield imgs_array, labels_array
                batch_imgs = []
                batch_labels = []

        if len(batch_imgs) > 0:
            # 剩余样本数目不足一个batch_size的数据，一起打包成一个mini-batch
            imgs_array = np.array(batch_imgs).astype('float32')
            labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
            yield imgs_array, labels_array

    return reader

# 定义验证集数据读取器
def valid_data_loader(datadir, csvfile, batch_size=10, mode='valid'):
    # 训练集读取时通过文件名来确定样本标签，验证集则通过csvfile来读取每个图片对应的标签
    # 请查看解压后的验证集标签数据，观察csvfile文件里面所包含的内容
    # csvfile文件所包含的内容格式如下，每一行代表一个样本，
    # 其中第一列是图片id，第二列是文件名，第三列是图片标签，
    # 第四列和第五列是Fovea的坐标，与分类任务无关
    # ID,imgName,Label,Fovea_X,Fovea_Y
    # 1,V0001.jpg,0,1157.74,1019.87
    # 2,V0002.jpg,1,1285.82,1080.47
    # 打开包含验证集标签的csvfile，并读入其中的内容
    filelists = open(csvfile).readlines()
    def reader():
        batch_imgs = []
        batch_labels = []
        for line in filelists[1:]:
            line = line.strip().split(',')
            name = line[1]
            label = int(line[2])
            # 根据图片文件名加载图片，并对图像数据作预处理
            filepath = os.path.join(datadir, name)
            img = cv2.imread(filepath)
            img = transform_img(img)
            # 每读取一个样本的数据，就将其放入数据列表中
            batch_imgs.append(img)
            batch_labels.append(label)
            if len(batch_imgs) == batch_size:
                # 当数据列表的长度等于batch_size的时候，
                # 把这些数据当作一个mini-batch，并作为数据生成器的一个输出
                imgs_array = np.array(batch_imgs).astype('float32')
                labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
                yield imgs_array, labels_array
                batch_imgs = []
                batch_labels = []

        if len(batch_imgs) > 0:
            # 剩余样本数目不足一个batch_size的数据，一起打包成一个mini-batch
            imgs_array = np.array(batch_imgs).astype('float32')
            labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
            yield imgs_array, labels_array

    return reader

# -*- coding: utf-8 -*-
#  识别眼疾图片
import os
import random
import paddle
import numpy as np

DATADIR = './PM/palm/PALM-Training400/PALM-Training400'
DATADIR2 = './PM/palm/PALM-Validation400'
CSVFILE = './PM/labels.csv'
# 设置迭代轮数
EPOCH_NUM = 5

# 定义训练过程
def train_pm(model, optimizer):
    # 开启0号GPU训练
    use_gpu = True
    paddle.device.set_device('gpu:0') if use_gpu else paddle.device.set_device('cpu')

    print('start training ... ')
    model.train()
    # 定义数据读取器，训练数据读取器和验证数据读取器
    train_loader = data_loader(DATADIR, batch_size=10, mode='train')
    valid_loader = valid_data_loader(DATADIR2, CSVFILE)
    for epoch in range(EPOCH_NUM):
        for batch_id, data in enumerate(train_loader()):
            x_data, y_data = data
            img = paddle.to_tensor(x_data)
            label = paddle.to_tensor(y_data)
            #print('image.shape=',img.shape)
            # 运行模型前向计算，得到预测值
            logits = model(img)
            loss = F.binary_cross_entropy_with_logits(logits, label)
            avg_loss = paddle.mean(loss)

            if batch_id % 20 == 0:
                print("epoch: {}, batch_id: {}, loss is: {:.4f}".format(epoch, batch_id, float(avg_loss.numpy())))
            # 反向传播，更新权重，清除梯度
            avg_loss.backward()
            optimizer.step()
            optimizer.clear_grad()

        model.eval()
        accuracies = []
        losses = []
        for batch_id, data in enumerate(valid_loader()):
            x_data, y_data = data
            img = paddle.to_tensor(x_data)
            label = paddle.to_tensor(y_data)
            # 运行模型前向计算，得到预测值
            logits = model(img)
            # 二分类，sigmoid计算后的结果以0.5为阈值分两个类别
            # 计算sigmoid后的预测概率，进行loss计算
            pred = F.sigmoid(logits)
            loss = F.binary_cross_entropy_with_logits(logits, label)
            # 计算预测概率小于0.5的类别
            pred2 = pred * (-1.0) + 1.0
            # 得到两个类别的预测概率，并沿第一个维度级联
            pred = paddle.concat([pred2, pred], axis=1)
            acc = paddle.metric.accuracy(pred, paddle.cast(label, dtype='int64'))

            accuracies.append(acc.numpy())
            losses.append(loss.numpy())
        print("[validation] accuracy/loss: {:.4f}/{:.4f}".format(np.mean(accuracies), np.mean(losses)))
        model.train()

        paddle.save(model.state_dict(), 'palm.pdparams')
        paddle.save(optimizer.state_dict(), 'palm.pdopt')
# 定义评估过程
def evaluation(model, params_file_path):

    # 开启0号GPU预估
    use_gpu = True
    paddle.device.set_device('gpu:0') if use_gpu else paddle.device.set_device('cpu')

    print('start evaluation .......')

    #加载模型参数
    model_state_dict = paddle.load(params_file_path)
    model.load_dict(model_state_dict)

    model.eval()
    eval_loader = data_loader(DATADIR, 
                        batch_size=10, mode='eval')

    acc_set = []
    avg_loss_set = []
    for batch_id, data in enumerate(eval_loader()):
        x_data, y_data = data
        img = paddle.to_tensor(x_data)
        label = paddle.to_tensor(y_data)
        y_data = y_data.astype(np.int64)
        label_64 = paddle.to_tensor(y_data)
        # 计算预测和精度
        prediction, acc = model(img, label_64)
        # 计算损失函数值
        loss = F.binary_cross_entropy_with_logits(prediction, label)
        avg_loss = paddle.mean(loss)
        acc_set.append(float(acc.numpy()))
        avg_loss_set.append(float(avg_loss.numpy()))
    # 求平均精度
    acc_val_mean = np.array(acc_set).mean()
    avg_loss_val_mean = np.array(avg_loss_set).mean()

    print('loss={:.4f}, acc={:.4f}'.format(avg_loss_val_mean, acc_val_mean))
#==============================================================================================


# 创建模型
model = AlexNet()
# 启动训练过程
opt = paddle.optimizer.Adam(learning_rate=0.001, parameters=model.parameters())

train_pm(model, optimizer=opt)


# 输入数据形状是 [N, 3, H, W]
# 这里用np.random创建一个随机数组作为输入数据
x = np.random.randn(*[10,3,224,224])
x = x.astype('float32')
# 创建LeNet类的实例，指定模型名称和分类的类别数目
model = AlexNet(2)
# 通过调用LeNet从基类继承的sublayers()函数，
# 查看LeNet中所包含的子层
print(model.sublayers())
print(x.shape)
x = paddle.to_tensor(x)
print(x.shape)
for item in model.sublayers():
    # item是LeNet类中的一个子层
    # 查看经过子层之后的输出数据形状
    try:
        x = item(x)
    except:
        x = paddle.reshape(x, [x.shape[0], -1])
        x = item(x)
    if len(item.parameters())==2:
        # 查看卷积和全连接层的数据和参数的形状，
        # 其中item.parameters()[0]是权重参数w，item.parameters()[1]是偏置参数b
        print(item.full_name(), x.shape, item.parameters()[0].shape, item.parameters()[1].shape)
    else:
        # 池化层没有参数
        print(item.full_name(), x.shape)