vgg和PyTorch

1.vgg:

由于使用图片数量过多，耗时久，这里仅取其中四百张图片。

代码片段：

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers, regularizers
import numpy as np
import os
import cv2
import matplotlib.pyplot as plt


os.environ["CUDA_VISIBLE_DEVICES"] = "1"
resize = 224
path ="D:/pc/data/seven/strain/"

def load_data():
    imgs = os.listdir(path)
    num = len(imgs)
    train_data = np.empty((200, resize, resize, 3), dtype="int32")
    train_label = np.empty((200, ), dtype="int32")
    test_data = np.empty((200, resize, resize, 3), dtype="int32")
    test_label = np.empty((200, ), dtype="int32")
    for i in range(200):
        if i % 2:
            train_data[i] = cv2.resize(cv2.imread(path+'/'+ 'dog.' + str(i) + '.jpg'), (resize, resize))
            train_label[i] = 1
        else:
            train_data[i] = cv2.resize(cv2.imread(path+'/' + 'cat.' + str(i) + '.jpg'), (resize, resize))
            train_label[i] = 0
    for i in range(200, 400):
        if i % 2:
            test_data[i-200] = cv2.resize(cv2.imread(path+'/' + 'dog.' + str(i) + '.jpg'), (resize, resize))
            test_label[i-200] = 1
        else:
            test_data[i-200] = cv2.resize(cv2.imread(path+'/' + 'cat.' + str(i) + '.jpg'), (resize, resize))
            test_label[i-200] = 0
    return train_data, train_label, test_data, test_label


def vgg16():
    weight_decay = 0.0005
    nb_epoch = 10
    batch_size = 32

    # layer1
    model = keras.Sequential()
    model.add(layers.Conv2D(64, (3, 3), padding='same',
                            input_shape=(224, 224, 3), kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.3))
    # layer2
    model.add(layers.Conv2D(64, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.MaxPooling2D(pool_size=(2, 2)))
    # layer3
    model.add(layers.Conv2D(128, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer4
    model.add(layers.Conv2D(128, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.MaxPooling2D(pool_size=(2, 2)))
    
    # layer5
    model.add(layers.Conv2D(256, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer6
    model.add(layers.Conv2D(256, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer7
    model.add(layers.Conv2D(256, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.MaxPooling2D(pool_size=(2, 2)))
    # layer8
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer9
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer10
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.MaxPooling2D(pool_size=(2, 2)))
    # layer11
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer12
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.Dropout(0.4))
    # layer13
    model.add(layers.Conv2D(512, (3, 3), padding='same', kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    model.add(layers.MaxPooling2D(pool_size=(2, 2)))
    model.add(layers.Dropout(0.5))
    # layer14
    model.add(layers.Flatten())
    model.add(layers.Dense(512, kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    # layer15
    model.add(layers.Dense(512, kernel_regularizer=regularizers.l2(weight_decay)))
    model.add(layers.Activation('relu'))
    model.add(layers.BatchNormalization())
    # layer16
    model.add(layers.Dropout(0.5))
    model.add(layers.Dense(2))
    model.add(layers.Activation('softmax'))

    return model
#if __name__ == '__main__':
train_data, train_label, test_data, test_label = load_data()
train_data = train_data.astype('float32')
test_data = test_data.astype('float32')
train_label = keras.utils.to_categorical(train_label, 2)
test_label = keras.utils.to_categorical(test_label, 2)
#定义训练方法,超参数设置
model = vgg16()
sgd = tf.keras.optimizers.SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) #设置优化器为SGD
model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])
history = model.fit(train_data, train_label,
              batch_size=20,
              epochs=10,
              validation_split=0.2,  #把训练集中的五分之一作为验证集
              shuffle=True)
scores = model.evaluate(test_data,test_label,verbose=1)
print(scores)
model.save('model/vgg16dogcat.h5')
acc = history.history['accuracy']  # 获取训练集准确性数据
val_acc = history.history['val_accuracy']  # 获取验证集准确性数据
loss = history.history['loss']  # 获取训练集错误值数据
val_loss = history.history['val_loss']  # 获取验证集错误值数据
epochs = range(1, len(acc) + 1)
plt.plot(epochs, acc, 'bo', label='Trainning acc')  # 以epochs为横坐标，以训练集准确性为纵坐标
plt.plot(epochs, val_acc, 'b', label='Vaildation acc')  # 以epochs为横坐标，以验证集准确性为纵坐标
plt.legend()  # 绘制图例，即标明图中的线段代表何种含义

plt.show()

运行结果如下：

2.PyTorch

PyTorch的一些应用

这里列举两个应用

a.模型测试

#2022-03-22
#导入相应的库
import numpy as np

import torch
import torch.nn as nn
import torch.nn.functional as F
import os
from scipy import stats
import pandas as pd


#数据的导入
titanic_data = pd.read_excel(r"C:\Users\L\data_survived.xlsx")
print(titanic_data.columns )#打印列名

#用哑变量将指定字段转成one-hot
#titanic_data = pd.concat([titanic_data,
#                         pd.get_dummies(titanic_data['sex']),
#                         pd.get_dummies(titanic_data['embarked'],prefix="embark"),
#                         pd.get_dummies(titanic_data['pclass'],prefix="class")], axis=1)

print(titanic_data.columns )
# print(titanic_data['sex'])
# print(titanic_data['female'])



#处理None值
titanic_data["age"] = titanic_data["age"].fillna(titanic_data["age"].mean())
titanic_data["fare"] = titanic_data["fare"].fillna(titanic_data["fare"].mean())#乘客票价

#删去无用的列
#titanic_data = titanic_data.drop(['age','sibsp','parch','fare','female','male','embark_C','embark_Q','embark_S','class_1','class_2','class_3'], axis=1)
#print(titanic_data.columns )

#titanic_data.to_excel(r"C:\Users\L\1112.xlsx")

#分离样本和标签
labels = titanic_data["survived"].to_numpy()

data_features = titanic_data.drop(['survived'], axis=1)
data = data_features.to_numpy()

labels 

#样本的属性名称
feature_names = list(titanic_data.columns)

#将样本分为训练和测试两部分
from sklearn.model_selection import train_test_split
train_features,test_features,train_labels,test_labels = train_test_split(data,labels,test_size=0.3,random_state=10)



class Mish(nn.Module):#Mish激活函数
    def __init__(self):
        super().__init__()
        print("Mish activation loaded...")
    def forward(self,x):
        x = x * (torch.tanh(F.softplus(x)))
        return x

torch.nn.Linear（in_features，out_features，bias = True ）
对传入数据应用线性变换：y = A x+ b
参数：
in_features - 每个输入样本的大小
out_features - 每个输出样本的大小
bias - 如果设置为False，则图层不会学习附加偏差。默认值：True

torch.manual_seed(0)  #设置随机种子
#构建一个三层网络12-8-2
class ThreelinearModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(12, 8)
        self.mish1 = Mish()
        self.linear2 = nn.Linear(8, 9)
        self.mish2 = Mish()
        self.linear3 = nn.Linear(9, 6)
        self.softmax = nn.Softmax(dim=1)
        self.criterion = nn.CrossEntropyLoss() #定义交叉熵函数

    def forward(self, x): #定义一个全连接网络
        lin1_out = self.linear1(x)
        out1 = self.mish1(lin1_out)
        out2 = self.mish2(self.linear2(out1))

        return self.softmax(self.linear3(out2))
    

    def getloss(self,x,y): #实现LogicNet类的损失值计算接口
        y_pred = self.forward(x)
        loss = self.criterion(y_pred,y)#计算损失值得交叉熵
        return loss

# 深度学习中经常看到epoch、 iteration和batchsize，下面按自己的理解说说这三个的区别：
（1）batchsize：批大小。在深度学习中，一般采用SGD训练，即每次训练在训练集中取batchsize个样本训练；
（2）iteration：1个iteration等于使用batchsize个样本训练一次；
（3）epoch：1个epoch等于使用训练集中的全部样本训练一次；
举个例子，训练集有1000个样本，batchsize=10，那么：
训练完整个样本集需要：
100次iteration，1次epoch。



net = ThreelinearModel()#模型实例化
num_epochs = 200  #迭代数次
optimizer = torch.optim.Adam(net.parameters(), lr=0.04)  #优化器选择Adam，只有一个参数学习率


#把特征和标签值转成torch张量形式
input_tensor = torch.from_numpy(train_features).type(torch.FloatTensor)
label_tensor = torch.from_numpy(train_labels)


losses = []#定义列表，用于接收每一步的损失值
for epoch in range(num_epochs): 
    loss = net.getloss(input_tensor,label_tensor)
    losses.append(loss.item())
    optimizer.zero_grad()#清空之前的梯度。
    loss.backward()#反向传播损失值
    optimizer.step()#更新参数
    if epoch % 20 == 0:
        print ('Epoch {}/{} => Loss: {:.2f}'.format(epoch+1, num_epochs, loss.item()))
        #.item()方法 是得到一个元素张量里面的元素值

os.makedirs('models', exist_ok=True)
#os.makedirs自动创建一个文件夹models
#如果exist_ok为True，则在目标目录已存在的情况下不会触发FileExistsError异常。
torch.save(net.state_dict(), 'models/titanic_model.pht') #保存模型




print(len(losses))

import matplotlib.pyplot as plt
def moving_average(a, w=10):#定义函数计算移动平均损失值
    if len(a) < w:
        return a[:]
    return [val if idx < w else sum(a[(idx-w):idx])/w for idx, val in enumerate(a)]
#enumerate() 函数用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列，同时列出数据和数据下标，一般用在 for 循环当中。
def plot_losses(losses):
    avgloss= moving_average(losses) #获得损失值的移动平均值
    print(len(avgloss))
    plt.figure(1)
    plt.subplot(211)
    plt.plot(range(len(avgloss)), avgloss, 'b--')
    plt.xlabel('step number')
    plt.ylabel('Training loss')
    plt.title('step number vs. Training loss')
    plt.show()



plot_losses(losses)

#输出训练结果
out_probs = net(input_tensor).detach().numpy()
out_classes = np.argmax(out_probs, axis=1)
print("Train Accuracy:", sum(out_classes == train_labels) / len(train_labels))

#测试模型
test_input_tensor = torch.from_numpy(test_features).type(torch.FloatTensor)
out_probs = net(test_input_tensor).detach().numpy()
out_classes = np.argmax(out_probs, axis=1)
print("Test Accuracy:", sum(out_classes == test_labels) / len(test_labels))

运行结果：

2.PyTorch自动微分

import torch
x=torch.ones(2,2,requires_grad=True)
print(x)
y=x+2
print(y)
print(y.grad_fn)
z=y*y*3
out=z.mean()
print(z,out)
a=torch.randn(2,2)
a=((a*3)/(a-1))
print(a.requires_grad)
a.requires_grad_(True)
print(a.requires_grad)
b=(a*a).sum()
print(b.grad_fn)
out.backward()
print(x.grad)

 

3008