python+opencv实现⼈脸识别笔记三（训练模型+保存模型）
⼀
在进⾏训练之前，我们⾸先需要对图⽚进⾏修改，使其长宽相等。
新建
import tensorflow as tf
import cv2
import numpy as np
import os
import random
import sys
del_selection import train_test_split
my_faces_path = '/home/dong/PycharmProjects/untitled/⼈脸识别/data/me'
other_faces_path = '/home/dong/PycharmProjects/untitled/⼈脸识别/data/zhang'
size = 64
imgs = []
labs = []
def getPaddingSize(img):
h, w, _ = img.shape
top, bottom, left, right = (0, 0, 0, 0)
longest = max(h, w)
if w < longest:
tmp = longest - w  # 相当于 h-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
修改之前图⽚
修改之后的图⽚（长宽相等）
⼆
从⽂件中读取照⽚
继续添加代码
def readData(path, h=size, w=size):
for filename in os.listdir(path):
dswith('.jpg'):
filename = path + '/' + filename
img = cv2.imread(filename)
top, bottom, left, right = getPaddingSize(img)
# 将图⽚放⼤，扩充图⽚边缘部分
img = pyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=[0, 0, 0])
img = size(img, (h, w))
imgs.append(img)
labs.append(path)
readData(my_faces_path)
readData(other_faces_path)
三
设置训练模型
继续添加代码
# 将图⽚数据与标签转换成数组
imgs = np.array(imgs)
labs = np.array([[0, 1] if lab == my_faces_path else [1, 0] for lab in labs])
# 随机划分测试集与训练集
train_x, test_x, train_y, test_y = train_test_split(imgs, labs, test_size=0.3, random_state=random.randint(0, 100)) # 参数：图⽚数据的总数，图⽚的⾼、宽、通道
train_x = shape(train_x.shape[0], size, size, 3)
train_x = shape(train_x.shape[0], size, size, 3)
test_x = shape(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])
y_ = tf.placeholder(tf.float32, [None, 2])
keep_prob_5 = tf.placeholder(tf.float32)
keep_prob_75 = tf.placeholder(tf.float32)
def weightVariable(shape):
init = tf.random_normal(shape, stddev=0.01)
return tf.Variable(init)
def biasVariable(shape):
init = tf.random_normal(shape)
return tf.Variable(init)
def conv2d(x, W):
v2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
def maxPool(x):
ax_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
def dropout(x, keep):
dropout(x, keep)
def cnnLayer():
# 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])
# 卷积
conv1 = lu(conv2d(x, W1) + b1)
# 池化
pool1 = maxPool(conv1)
# 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 = lu(conv2d(drop1, W2) + b2)
pool2 = maxPool(conv2)
drop2 = dropout(pool2, keep_prob_5)
# 第三层
W3 = weightVariable([3, 3, 64, 64])
b3 = biasVariable([64])
conv3 = lu(conv2d(drop2, W3) + b3)
pool3 = maxPool(conv3)
drop3 = dropout(pool3, keep_prob_5)
# 全连接层
# 全连接层
Wf = weightVariable([8 * 8 * 64, 512])
bf = biasVariable([512])
drop3_flat = tf.reshape(drop3, [-1, 8 * 8 * 64])
dense = lu(tf.matmul(drop3_flat, Wf) + bf)    dropf = dropout(dense, keep_prob_75)
# 输出层
Wout = weightVariable([512, 2])
bout = biasVariable([2])
# out = tf.matmul(dropf, Wout) + bout
out = tf.add(tf.matmul(dropf, Wout), bout)
return out
四
进⾏训练
继续添加代码
def cnnTrain():
session如何设置和读取out = cnnLayer()
'''
out = tf.nn.softmax(out)  #归⼀化
c1 = -tf.reduce.sum(y_*tf.log(out1))
cross_entropy = tf.reduce_mean(c1)
'''
cross_entropy = tf.reduce_softmax_cross_entropy_with_logits(logits=out, labels=y_))
# 梯度下降函数，优化器就会按照循环的次数⼀次次沿着loss最⼩值的⽅向优化参数了。
# train_step = tf.train.GradientDescentOptimizer(0.2).minimize(loss)
train_step = tf.train.AdamOptimizer(0.01).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 = _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())
for n in range(10):
# 每次取12(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})
# 打印损失
acc = accuracy.eval({x: test_x, y_: test_y, keep_prob_5: 1.0, keep_prob_75: 1.0})
print(n * num_batch + i, loss, acc)
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
if acc > 0.98 and n > 2:
saver.save(sess, '/home/dong/PycharmProjects/untitled/FaceRecognition-tensorflow/模型/del', global_step=n * num_batch + i )
print('accuracy less 0.98, exited!')
cnnTrain()

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。