train.py

import tensorflow as tf
import cv2
import numpy as np
import os
import random
import sys
from sklearn.model_selection import train_test_split

my_faces_path = './my_faces'
other_faces_path = './other_faces'
face2_path = './wyj_faces_new'
size = 64

imgs = []
labs = []

def getPaddingSize(img):
    h, w, _ = img.shape  #获取图片的高和宽
    top, bottom, left, right = (0,0,0,0)  #四个坐标
    longest = max(h, w)  #取高和宽中的最大值，以便将图片处理成正方形

    if w < longest:
        tmp = longest - w
        # //表示整除符号
        left = tmp // 2
        right = tmp - left
    elif h < longest:
        tmp = longest - h
        top = tmp // 2
        bottom = tmp - top
    else:
        pass
    return top, bottom, left, right

def readData(path , h=size, w=size):  #用于读取图片
    for filename in os.listdir(path):
        if filename.endswith('.jpg'):
            filename = path + '/' + filename  #设置好路径

            img = cv2.imread(filename)  #将照片读进来保存在img中

            top,bottom,left,right = getPaddingSize(img)
            # 将图片放大， 扩充图片边缘部分
            img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=[0,0,0])  #扩充边界以便对边缘进行处理
            img = cv2.resize(img, (h, w))  #截取图片大小为（h,w）

            imgs.append(img)  #将处理好的img附在imgs列表后面
            labs.append(path)  #把每张图片的path都保存进labs列表

readData(my_faces_path)  #对我的图片进行处理
readData(other_faces_path)  #对其他人脸图片进行处理
readData(face2_path)
# 将图片数据与标签转换成数组
imgs = np.array(imgs)  #将图片数据转换为数组
labs = np.array([[0,1,0] if lab == face2_path else [0,0,1] if lab == my_faces_path else [1,0,0] for lab in labs])  #添加标签作为正确结果以便训练时使用
# 随机划分测试集与训练集，训练集占95%，测试集占5%
train_x,test_x,train_y,test_y = train_test_split(imgs, labs, test_size=0.05, random_state=random.randint(0,100))
# 参数：图片数据的总数，图片的高、宽、通道
train_x = train_x.reshape(train_x.shape[0], size, size, 3)
test_x = test_x.reshape(test_x.shape[0], size, size, 3)
# 将数据转换成小于1的数
train_x = train_x.astype('float32')/255.0
test_x = test_x.astype('float32')/255.0

print('train size:%s, test size:%s' % (len(train_x), len(test_x)))
# 图片块，每次取100张图片
batch_size = 100
num_batch = len(train_x) // batch_size

x = tf.placeholder(tf.float32, [None, size, size, 3])  #喂入图片的分辨率是64*64，通道数为3
y_ = tf.placeholder(tf.float32, [None, 3])

keep_prob_5 = tf.placeholder(tf.float32)
keep_prob_75 = tf.placeholder(tf.float32)

def weightVariable(shape):  #初始化参数
    init = tf.random_normal(shape, stddev=0.01)
    return tf.Variable(init)

def biasVariable(shape):  #附加项
    init = tf.random_normal(shape)
    return tf.Variable(init)

def conv2d(x, W):  #卷积，参数分别是喂入图片的描述、对卷积核的描述、卷积核移动的步长
    return tf.nn.conv2d(x, W, strides=[1,1,1,1], padding='SAME')

def maxPool(x):  #池化，参数分别是喂入图片的描述，池化核的描述、池化核移动的步长
    return tf.nn.max_pool(x, ksize=[1,2,2,1], strides=[1,2,2,1], padding='SAME')

def dropout(x, keep):  #定义在训练过程中的舍弃，参数分别是来自上层的输出和暂时舍弃的概率
    return tf.nn.dropout(x, keep)

def cnnLayer():  #前向传播
    # 第一层
    W1 = weightVariable([3,3,3,32]) # 卷积核大小(3,3)， 输入通道(3)， 输出通道(32)
    b1 = biasVariable([32])  #初始化偏置
    # 卷积
    conv1 = tf.nn.relu(conv2d(x, W1) + b1)  #激活函数采用relu，第一层卷积
    # 池化
    pool1 = maxPool(conv1)  #第一层池化，采用最大池化
    # 减少过拟合，随机让某些权重不更新
    drop1 = dropout(pool1, keep_prob_5)

    # 第二层
    W2 = weightVariable([3,3,32,64])
    b2 = biasVariable([64])
    conv2 = tf.nn.relu(conv2d(drop1, W2) + b2)  #第二层卷积
    pool2 = maxPool(conv2)  #第二层池化
    drop2 = dropout(pool2, keep_prob_5)

    #第三层
    W3 = weightVariable([3,3,64,64])  #第三层卷积
    b3 = biasVariable([64])
    conv3 = tf.nn.relu(conv2d(drop2, W3) + b3)  #第三层池化
    pool3 = maxPool(conv3)
    drop3 = dropout(pool3, keep_prob_5)

    # 全连接层
    Wf = weightVariable([8*8*64, 512])
    bf = biasVariable([512])
    drop3_flat = tf.reshape(drop3, [-1, 8*8*64])
    dense = tf.nn.relu(tf.matmul(drop3_flat, Wf) + bf)
    dropf = dropout(dense, keep_prob_75)

    # 输出层
    Wout = weightVariable([512,3])
    bout = biasVariable([3])
    #out = tf.matmul(dropf, Wout) + bout
    out = tf.add(tf.matmul(dropf, Wout), bout)
    return out

def cnnTrain():  #反向传播
    out = cnnLayer()  #前向传播搭建结构

    cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=out, labels=y_))

    train_step = tf.train.AdamOptimizer(0.001).minimize(cross_entropy)
    # 比较标签是否相等，再求的所有数的平均值，tf.cast(强制转换类型)
    accuracy = tf.reduce_mean(tf.cast(tf.equal(tf.argmax(out, 1), tf.argmax(y_, 1)), tf.float32))
    # 将loss与accuracy保存以供tensorboard使用
    tf.summary.scalar('loss', cross_entropy)
    tf.summary.scalar('accuracy', accuracy)
    merged_summary_op = tf.summary.merge_all()
    # 数据保存器的初始化
    saver = tf.train.Saver()

    with tf.Session() as sess:

        sess.run(tf.global_variables_initializer())  #初始化参数

        summary_writer = tf.summary.FileWriter('./tmp', graph=tf.get_default_graph())

        for n in range(8):
             # 每次取128(batch_size)张图片
            for i in range(num_batch):
                batch_x = train_x[i*batch_size : (i+1)*batch_size]
                batch_y = train_y[i*batch_size : (i+1)*batch_size]
                # 开始训练数据，同时训练三个变量，返回三个数据
                _,loss,summary = sess.run([train_step, cross_entropy, merged_summary_op],
                                           feed_dict={x:batch_x,y_:batch_y, keep_prob_5:0.5,keep_prob_75:0.75})
                summary_writer.add_summary(summary, n*num_batch+i)
                # 打印损失
                print(n*num_batch+i, loss)

                if (n*num_batch+i) % 100 == 0:
                    # 获取测试数据的准确率
                    acc = accuracy.eval({x:test_x, y_:test_y, keep_prob_5:1.0, keep_prob_75:1.0})
                    print(n*num_batch+i, acc)
                    # 准确率大于0.98时保存并退出

        saver.save(sess, './train_faces.model', global_step=n * num_batch + i)
        sys.exit(0)
        print('accuracy less 0.8, exited!')

cnnTrain()