-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathvideo_sex_rec.py
272 lines (245 loc) · 9.58 KB
/
video_sex_rec.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
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
from PIL import Image
import dlib
import cv2
import os.path
import PIL.Image
from pylab import *
import os
import numpy as np
import shutil
'''
通过BP神经网络学习图片并检测性别
'''
'''参数选择'''
picture_show=0 #是否显示照片
pixel=40 #像素大小
pixel_mat=pixel*pixel #像素总个数
file_num=0 #文件数量
photonum=0 #照片总数量
boy_photonum=0 #男生照片总数量
girl_photonum=0 #女生照片总数量
'''路径选择'''
girlpath = '../beauty' #女孩照片路径，用于训练
boypath = '../nan' #男孩照片路径，用于训练
photopath = '../faces' #截取彩色照片路径
graypath = '../grayfaces' #黑白照片路径
numpath= '../faces' #用于读取照片数量路径
def deletphoto(dir):
delList = os.listdir(dir)
for f in delList:
filePath = os.path.join(dir, f)
if os.path.isfile(filePath):
os.remove(filePath)
print(filePath + " was removed!")
elif os.path.isdir(filePath):
shutil.rmtree(filePath, True)
print('Directory: " + filePath +"was removed!')
'''
利用别人写好的人脸分类器来截取图片中的人脸，并保存到spath中
i:截取图片保存名
path:原图片路径
spath：处理过后图片路径
'''
def get_face_from_photo(i,path,spath):
detector = dlib.get_frontal_face_detector() #获取人脸分类，采用dlib进行人脸识别，dlib更多的人脸识别模型，可以检测脸部68甚至更多的特征点
filenames = os.listdir(path) # 读取path路径下的图片，获得所有的图片名字
for f1 in filenames:
f = os.path.join(path,f1)
img = cv2.imread(f, cv2.IMREAD_COLOR)
iimag = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB)) #转换文件格式
counts = detector(img, 1) #人脸检测
for index, face in enumerate(counts): #在图片中标注人脸，并显示
left = face.left()
top = face.top()
right = face.right()
bottom = face.bottom()
j =str(i)
j = j+'.jpg'
save_path = os.path.join(spath,j)
region = (left,top,right,bottom)
cropImg = iimag.crop(region) #裁切图片
cropImg.save(save_path) #保存裁切后的图片
if(picture_show==0):
print(i)
i +=1
cv2.rectangle(img, (left, top), (right, bottom), (0, 255, 0), 3)
if(picture_show==1):
cv2.namedWindow(f, cv2.WINDOW_AUTOSIZE)
cv2.imshow(f, img)
print("read_ok")
if(picture_show==1):
k = cv2.waitKey(0)
cv2.destroyAllWindows()
return i
'''
将人脸图片转化为pixel*pixel的灰度图片
path:人脸图片
spath:灰度图片
'''
def change_photo_size(path,spath):
filenames = os.listdir(path)
for filename in filenames:
f = os.path.join(path,filename)
iimag = PIL.Image.open(f).convert('L').resize((pixel,pixel))
savepath = os.path.join(spath,filename)
iimag.save(savepath)
'''
读取训练图片
k:照片数量
photo_path:照片保存的路径
'''
def read_photo_for_train(k,photo_path):
for i in range(k):
j = i
j = str(j)
st = '.jpg'
j = j+st
j = os.path.join(photo_path,j)
im1 = array(Image.open(j).convert('L'))
im1 = im1.reshape((pixel_mat,1))
if i == 0: #把所有的图片灰度值放到一个矩阵中,一列代表一张图片的信息
im = im1
else:
im = np.hstack((im,im1))
return im
'''
sigmoid函数实现
'''
def layerout(w,b,x):
y = np.dot(w,x) + b
t = -1.0*y
y = 1.0/(1+exp(t))
return y
'''
训练样本：百度的图片。女生标签为0，男生标签为1.
训练方法：简单的梯度下降法
参考：https://blog.csdn.net/yunyunyx/article/details/80539222
'''
'''
设置一个隐藏层，为每张照片像素值-->pixel_mat 隐藏层神经元个数-->1
输入为每张图片的灰度像素矩阵
x_train:训练样本的像素数据
y_train：训练样本的标签
w：输出层权重
b：输出层偏置
w_h：隐藏层权重
b_h：隐藏层偏置
step：循环步数
'''
def mytrain(x_train,y_train):
step=int(input('mytrain迭代步数：'))
a=double(input('学习因子：'))
inn = pixel_mat #输入神经元个数
hid = int(input('隐藏层神经元个数：')) #隐藏层神经元个数
out = 1 #输出层神经元个数
w = np.random.randn(out,hid)
w = np.mat(w)
b = np.mat(np.random.randn(out,1))
w_h = np.random.randn(hid,inn)
w_h = np.mat(w_h)
b_h = np.mat(np.random.randn(hid,1))
for i in range(step):
#打乱训练样本
r=np.random.permutation(photonum)
x_train = x_train[:,r]
y_train = y_train[:,r]
for j in range(photonum):
x = np.mat(x_train[:,j])
x = x.reshape((pixel_mat,1))
y = np.mat(y_train[:,j])
y = y.reshape((1,1))
hid_put = layerout(w_h,b_h,x)
out_put = layerout(w,b,hid_put)
#更新公式的实现
o_update = np.multiply(np.multiply((y-out_put),out_put),(1-out_put)) #计算输出单元误差项，y->tk
h_update = np.multiply(np.multiply(np.dot((w.T),np.mat(o_update)),hid_put),(1-hid_put)) #隐藏单元误差项
outw_update = a*np.dot(o_update,(hid_put.T)) #从隐藏层到输出层的dw
outb_update = a*o_update
hidw_update = a*np.dot(h_update,(x.T))
hidb_update = a*h_update
w = w + outw_update #更新参数
b = b+ outb_update
w_h = w_h +hidw_update
b_h =b_h +hidb_update
return w,b,w_h,b_h
'''
预测结果pre大于0.5，为男；预测结果小于或等于0.5为女
'''
def mytest(x_test,w,b,w_h,b_h):
hid = layerout(w_h,b_h,x_test);
pre = layerout(w,b,hid);
print(pre)
if pre > 0.5:
print("hello,boy!")
else:
print("hello,girl!")
return pre
'''主函数'''
deletphoto(photopath)
deletphoto(graypath)
#框出人脸，并保存到faces中,i为保存的名字
i = 0
i = get_face_from_photo(i,girlpath,photopath) #女孩
girl_photonum=i
print(girl_photonum)
i = get_face_from_photo(i,boypath,photopath) # 男孩
boy_photonum=i-girl_photonum
print(boy_photonum)
change_photo_size(photopath,graypath) #将人脸图片转化为28*28的灰度图片
photonum=(len([lists for lists in os.listdir(numpath) if os.path.isfile(os.path.join(numpath, lists))])) #获取图片数量
print(photonum)
im = read_photo_for_train(photonum,graypath)
#归一化
immin = im.min()
immax = im.max()
im = (im-immin)/(immax-immin)
x_train = im
#制作标签，女生为0，男生为1
y1 = np.zeros((1,girl_photonum))
y2 = np.ones((1,boy_photonum))
y_train = np.hstack((y1,y2))
#开始训练
print("----------------------开始训练-----------------------------------------")
w,b,w_h,b_h = mytrain(x_train,y_train)
print("-----------------------训练结束------------------------------------------")
# 测试
print("--------------------视频测试-----------------------------------------")
camera= cv2.VideoCapture(0)
while(True):
ret, f = camera.read() #图片来自视频中读取的内容
detector = dlib.get_frontal_face_detector() #获取人脸分类，采用dlib进行人脸识别，dlib更多的人脸识别模型，可以检测脸部68甚至更多的特征点
iimag = Image.fromarray(cv2.cvtColor(f, cv2.COLOR_BGR2RGB)) #把灰度处理后的图片进行处理，把OpenCV转换成PIL.Image格式
counts = detector(f, 1) # 人脸检测
for index, face in enumerate(counts): # 在图片中标注人脸，并显示
left = face.left()
top = face.top()
right = face.right()
bottom = face.bottom()
region = (left, top, right, bottom)
cropImg = iimag.crop(region) #将识别出的人脸提取出来
finaImg=cropImg.convert('L').resize((pixel,pixel)) #把大小统一转换成28*28像素
im1 = array(finaImg)
im1 = im1.reshape((pixel_mat, 1))
im = im1
#归一化
immin = im.min()
immax = im.max()
im = (im - immin) / (immax - immin)
i=0
x_test = im
xx = x_test[:, i]
xx = xx.reshape((pixel_mat, 1))
mytest(xx, w, b, w_h, b_h)
pre=mytest(xx, w, b, w_h, b_h)
font = cv2.FONT_HERSHEY_SIMPLEX # 使用默认字体
if pre > 0.5:
img = cv2.putText(f, 'man', (left, top - 30), font, 1.2, (255, 255, 255),2) # 添加文字，1.2表示字体大小，（0,40）是初始的位置，
# 保存
else:
img = cv2.putText(f, 'girl', (left, top - 30), font, 1.2, (255, 255, 255), 2) # 添加文字，1.2
cv2.rectangle(f, (left, top), (right, bottom), (0, 255, 0), 2)
cv2.imshow("camera", f)
if cv2.waitKey(1) & 0xff == ord("q"):
break
camera.release()
cv2.destroyAllWindows()