forked from AIBigTruth/CNN_faces_recognition
-
Notifications
You must be signed in to change notification settings - Fork 0
/
face_recognition.py
169 lines (153 loc) · 8.6 KB
/
face_recognition.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
"""---------------------------------------------------------
四、人脸识别
1、打开摄像头，获取图片并灰度化
2、人脸检测
3、加载卷积神经网络模型
4、人脸识别
------------------------------------------------------------"""
import tensorflow as tf
import cv2
import dlib
import numpy as np
import os
import random
import sys
from sklearn.model_selection import train_test_split
"""定义参数"""
faces_my_path = './faces_my'
faces_other_path = './faces_other'
batch_size = 128 # 每次取100张图片
learning_rate = 0.01 # 学习率
size = 64 # 图片大小64*64*3
imgs = [] # 存放人脸图片
labs = [] # 存放人脸图片对应的标签
x = tf.placeholder(tf.float32, [None, size, size, 3]) # 输入X：64*64*3
y_ = tf.placeholder(tf.float32, [None, 2]) # 输出Y_：1*2
keep_prob_fifty = tf.placeholder(tf.float32) # 50%，即0.5
keep_prob_seventy_five = tf.placeholder(tf.float32) # 75%，即0.75
"""定义读取人脸数据函数"""
def readData(path , h = size , w = size):
for filename in os.listdir(path):
if filename.endswith('.jpg'):
filename = path + '/' + filename
img = cv2.imread(filename)
top,bottom,left,right = getPaddingSize(img)
"""放大图片扩充图片边缘部分"""
img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=[0, 0, 0])
img = cv2.resize(img, (h, w))
imgs.append(img) # 一张张人脸图片加入imgs列表中
labs.append(path) # 一张张人脸图片对应的path，即文件夹名faces_my和faces_other，即标签
"""定义padding大小函数，有一套公式"""
def getPaddingSize(img):
height, width, _ = img.shape
top, bottom, left, right = (0, 0, 0, 0)
longest = max(height, width)
if width < longest:
tmp = longest - width
left = tmp // 2
right = tmp - left
elif height < longest:
tmp = longest - height
top = tmp // 2
bottom = tmp - top
else:
pass
return top, bottom, left, right
"""定义神经网络层，共五层，卷积层特征提取，池化层降维,全连接层进行分类，共两类：我的人脸（true),不是我的人脸（false）"""
def cnnLayer():
"""第一、二层，输入图片64*64*3，输出图片32*32*32"""
W1 = tf.Variable(tf.random_normal([3, 3, 3, 32])) # 卷积核大小(3,3)， 输入通道(3)， 输出通道(32)
b1 = tf.Variable(tf.random_normal([32]))
conv1 = tf.nn.relu(tf.nn.conv2d(x, W1, strides=[1, 1, 1, 1], padding='SAME')+b1) # 64*64*32，卷积提取特征，增加通道数
pool1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME') # 32*32*32，池化降维，减小复杂度
drop1 = tf.nn.dropout(pool1, keep_prob_fifty)
"""第三、四层，输入图片32*32*32，输出图片16*16*64"""
W2 = tf.Variable(tf.random_normal([3, 3, 32, 64])) # 卷积核大小(3,3)， 输入通道(32)， 输出通道(64)
b2 = tf.Variable(tf.random_normal([64]))
conv2 = tf.nn.relu(tf.nn.conv2d(drop1, W2, strides=[1, 1, 1, 1], padding='SAME') + b2) # 32*32*64
pool2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME') # 16*16*64
drop2 = tf.nn.dropout(pool2, keep_prob_fifty)
"""第五、六层，输入图片16*16*64，输出图片8*8*64"""
W3 = tf.Variable(tf.random_normal([3, 3, 64, 64])) # 卷积核大小(3,3)， 输入通道(64)， 输出通道(64)
b3 = tf.Variable(tf.random_normal([64]))
conv3 = tf.nn.relu(tf.nn.conv2d(drop2, W3, strides=[1, 1, 1, 1], padding='SAME') + b3) # 16*16*64
pool3 = tf.nn.max_pool(conv3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME') # 8*8*64=4096
drop3 = tf.nn.dropout(pool3, keep_prob_fifty)
"""第七层，全连接层，输入图片8*8*64，reshape到1*4096，输出1*512"""
Wf = tf.Variable(tf.random_normal([8*8*64,512])) # 输入通道(4096)， 输出通道(512)
bf = tf.Variable(tf.random_normal([512]))
drop3_flat = tf.reshape(drop3, [-1, 8*8*64]) # -1表示行随着列的需求改变，1*4096
dense = tf.nn.relu(tf.matmul(drop3_flat, Wf) + bf) # [1,4096]*[4096,512]=[1,512]
dropf = tf.nn.dropout(dense, keep_prob_seventy_five)
"""第八层，输出层，输入1*512，输出1*2，再add，输出一个数"""
Wout = tf.Variable(tf.random_normal([512,2])) # 输入通道(512)， 输出通道(2)
bout = tf.Variable(tf.random_normal([2]))
out = tf.add(tf.matmul(dropf, Wout), bout) # (1,512)*(512,2)=(1,2) ,跟y_ [0,1]、[1,0]比较给出损失
return out
"""定义人脸识别函数"""
def face_recognise(image):
res = sess.run(predict, feed_dict={x: [image/255.0], keep_prob_fifty: 1.0, keep_prob_seventy_five: 1.0})
if res[0] == 1:
return "Yes,my face"
else:
return "No,other face"
if __name__ == '__main__':
"""1、读取人脸数据"""
readData(faces_my_path)
readData(faces_other_path)
imgs = np.array(imgs) # 将图片数据与标签转换成数组
labs = np.array([[0, 1] if lab == faces_my_path else [1, 0] for lab in labs]) # 标签：[0,1]表示是我的人脸，[1,0]表示其他的人脸
"""2、随机划分测试集与训练集"""
train_x_1, test_x_1, train_y, test_y = train_test_split(imgs, labs, test_size=0.05,
random_state=random.randint(0, 100))
train_x_2 = train_x_1.reshape(train_x_1.shape[0], size, size, 3) # 参数：图片数据的总数，图片的高、宽、通道
test_x_2 = test_x_1.reshape(test_x_1.shape[0], size, size, 3)
train_x = train_x_2.astype('float32') / 255.0 # 归一化
test_x = test_x_2.astype('float32') / 255.0
print('Train Size:%s, Test Size:%s' % (len(train_x), len(test_x)))
num_batch = len(train_x) // batch_size # 22784//128=178
"""3、将读取的人脸图片输出神经网络，输出out(1,2)"""
out = cnnLayer()
"""4、预测， 1表示按行返回out中最大值的索引，而不是out与1比谁大返回谁，predict为索引值，0或1，因为out的shape是（1,2）,一行两列，两个数字"""
predict = tf.argmax(out, 1) # 0或者1
saver = tf.train.Saver()
sess = tf.Session()
saver.restore(sess, tf.train.latest_checkpoint('.'))
"""5、检测人脸，特征提取器: dlib自带的frontal_face_detector"""
detector = dlib.get_frontal_face_detector()
cap = cv2.VideoCapture(0) # 打开摄像头
while True:
_, img = cap.read() # 读取
gray_image = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 灰度化
dets = detector(gray_image, 1)
if not len(dets):
key = cv2.waitKey(30)
if key == 27:
sys.exit(0)
"""--------------------------------------------------------------------
使用enumerate 函数遍历序列中的元素以及它们的下标,i为人脸序号,d为i对应的元素;
left：人脸左边距离图片左边界的距离 ；right：人脸右边距离图片左边界的距离
top：人脸上边距离图片上边界的距离 ；bottom：人脸下边距离图片上边界的距离
----------------------------------------------------------------------"""
for i, d in enumerate(dets):
x1 = d.top() if d.top() > 0 else 0
y1 = d.bottom() if d.bottom() > 0 else 0
x2 = d.left() if d.left() > 0 else 0
y2 = d.right() if d.right() > 0 else 0
"""人脸大小64*64"""
face = img[x1:y1, x2:y2]
face = cv2.resize(face, (size, size))
"""6、人脸识别，输出识别结果"""
print('It recognizes my face? %s' % face_recognise(face))
cv2.rectangle(img, (x2, x1), (y2, y1), (255, 0, 0), 3)
if face_recognise(face) == "Yes,my face":
cv2.putText(img, 'Yes,my face', (x2, y2), cv2.FONT_HERSHEY_COMPLEX, 0.5, (0, 255, 0), 2)
else:
cv2.putText(img, 'No,other face', (x2, y2), cv2.FONT_HERSHEY_COMPLEX, 0.5, (0, 255, 0), 2)
"""通过确定对角线画矩形"""
#cv2.rectangle(img, (x2, x1), (y2, y1), (255, 0, 0), 3)
cv2.imshow('image', img)
key = cv2.waitKey(30)
if key == 27:
sys.exit(0)
sess.close()