前言
在前面的博客中介绍了，如何使用dlib标定人脸（python dlib学习（一）：人脸检测），提取68个特征点（python dlib学习（二）：人脸特征点标定）。这次要在这两个工作的基础之上，将人脸的信息提取成一个128维的向量空间。在这个向量空间上，同一个人脸的更接近，不同人脸的距离更远。度量采用欧式距离，欧氏距离计算不算复杂。
二维情况下：
distance=(x1−x2)2+(y1−y2)2−−−−−−−−−−−−−−−−−−√
distance=(x1−x2)2+(y1−y2)2
三维情况下：
distance=(x1−x2)2+(y1−y2)2+(z1−z2)2−−−−−−−−−−−−−−−−−−−−−−−−−−−−√
distance=(x1−x2)2+(y1−y2)2+(z1−z2)2
将其扩展到128维的情况下即可。
通常使用的判别阈值是0.6，即如果两个人脸的向量空间的欧式距离超过了0.6，即认定不是同一个人；如果欧氏距离小于0.6，则认为是同一个人。这个距离也可以由自己定，只要效果能更好。
实验中使用了两个模型：
shape_predictor_68_face_landmarks.dat：
http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
dlib_face_recognition_resnet_model_v1.dat：
http://dlib.net/files/dlib_face_recognition_resnet_model_v1.dat.bz2
文件夹目录：
两个模型放在model文件夹中，测试图片放在faces中，图片自己随便下几张就行。
完整工程下载链接：
http://pan.baidu.com/s/1boCDZ7T
程序1
不说废话了，直接上代码。
# -*- coding: utf-8 -*-
import sys
import dlib
import cv2
import os
import glob
current_path = os.getcwd() # 获取当前路径
# 模型路径
predictor_path = current_path + "\model\shape_predictor_68_face_landmarks.dat"
face_rec_model_path = current_path + "\model\dlib_face_recognition_resnet_model_v1.dat"
#测试图片路径
faces_folder_path = current_path + "\faces\"
# 读入模型
detector = dlib.get_frontal_face_detector()
shape_predictor = dlib.shape_predictor(predictor_path)
face_rec_model = dlib.face_recognition_model_v1(face_rec_model_path)
for img_path in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
print("Processing file: {}".format(img_path))
# opencv 读取图片，并显示
img = cv2.imread(img_path, cv2.IMREAD_COLOR)
# opencv的bgr格式图片转换成rgb格式
b, g, r = cv2.split(img)
img2 = cv2.merge([r, g, b])
dets = detector(img, 1) # 人脸标定
print("Number of faces detected: {}".format(len(dets)))
for index, face in enumerate(dets):
print('face {}; left {}; top {}; right {}; bottom {}'.format(index, face.left(), face.top(), face.right(), face.bottom()))
shape = shape_predictor(img2, face) # 提取68个特征点
for i, pt in enumerate(shape.parts()):
#print('Part {}: {}'.format(i, pt))
pt_pos = (pt.x, pt.y)
cv2.circle(img, pt_pos, 2, (255, 0, 0), 1)
#print(type(pt))
#print("Part 0: {}, Part 1: {} ...".format(shape.part(0), shape.part(1)))
cv2.namedWindow(img_path+str(index), cv2.WINDOW_AUTOSIZE)
cv2.imshow(img_path+str(index), img)
face_descriptor = face_rec_model.compute_face_descriptor(img2, shape) # 计算人脸的128维的向量
print(face_descriptor)
k = cv2.waitKey(0)
cv2.destroyAllWindows()
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
程序1结果
部分打印结果：
F:Pythonmy_dlib_codesface_recognition>python my_face_recogniton.py
Processing file: F:Pythonmy_dlib_codesface_recognitionfacesjobs.jpg
Number of faces detected: 1
face 0; left 184; top 64; right 339; bottom 219
-0.179784
0.15487
0.10509
-0.0973604
-0.19153
0.000418252
-0.0357536
-0.0206766
0.129741
-0.0628359
....
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
后面的那一堆数字就是人脸在128维向量空间上的值。
程序2
前面只是测试了一下，把要用的值给求到了。这里我封装了一下，把比对功能实现了。没加多少东西，所以不做赘述了。
# -*- coding: utf-8 -*-
import sys
import dlib
import cv2
import os
import glob
import numpy as np
def comparePersonData(data1, data2):
diff = 0
# for v1, v2 in data1, data2:
# diff += (v1 - v2)**2
for i in xrange(len(data1)):
diff += (data1[i] - data2[i])**2
diff = np.sqrt(diff)
print diff
if(diff < 0.6):
print "It's the same person"
else:
print "It's not the same person"
def savePersonData(face_rec_class, face_descriptor):
if face_rec_class.name == None or face_descriptor == None:
return
filePath = face_rec_class.dataPath + face_rec_class.name + '.npy'
vectors = np.array([])
for i, num in enumerate(face_descriptor):
vectors = np.append(vectors, num)
# print(num)
print('Saving files to :'+filePath)
np.save(filePath, vectors)
return vectors
def loadPersonData(face_rec_class, personName):
if personName == None:
return
filePath = face_rec_class.dataPath + personName + '.npy'
vectors = np.load(filePath)
print(vectors)
return vectors
class face_recognition(object):
def __init__(self):
self.current_path = os.getcwd() # 获取当前路径
self.predictor_path = self.current_path + "\model\shape_predictor_68_face_landmarks.dat"
self.face_rec_model_path = self.current_path + "\model\dlib_face_recognition_resnet_model_v1.dat"
self.faces_folder_path = self.current_path + "\faces\"
self.dataPath = self.current_path + "\data\"
self.detector = dlib.get_frontal_face_detector()
self.shape_predictor = dlib.shape_predictor(self.predictor_path)
self.face_rec_model = dlib.face_recognition_model_v1(self.face_rec_model_path)
self.name = None
self.img_bgr = None
self.img_rgb = None
self.detector = dlib.get_frontal_face_detector()
self.shape_predictor = dlib.shape_predictor(self.predictor_path)
self.face_rec_model = dlib.face_recognition_model_v1(self.face_rec_model_path)
def inputPerson(self, name='people', img_path=None):
if img_path == None:
print('No file!
')
return
# img_name += self.faces_folder_path + img_name
self.name = name
self.img_bgr = cv2.imread(self.current_path+img_path)
# opencv的bgr格式图片转换成rgb格式
b, g, r = cv2.split(self.img_bgr)
self.img_rgb = cv2.merge([r, g, b])
def create128DVectorSpace(self):
dets = self.detector(self.img_rgb, 1)
print("Number of faces detected: {}".format(len(dets)))
for index, face in enumerate(dets):
print('face {}; left {}; top {}; right {}; bottom {}'.format(index, face.left(), face.top(), face.right(), face.bottom()))
shape = self.shape_predictor(self.img_rgb, face)
face_descriptor = self.face_rec_model.compute_face_descriptor(self.img_rgb, shape)
# print(face_descriptor)
# for i, num in enumerate(face_descriptor):
# print(num)
# print(type(num))
return face_descriptor
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
程序2结果
测试代码1：
import face_rec as fc
face_rec = fc.face_recognition() # 创建对象
face_rec.inputPerson(name='jobs', img_path='\faces\jobs.jpg') # name中写第一个人名字，img_name为图片名字，注意要放在faces文件夹中
vector = face_rec.create128DVectorSpace() # 提取128维向量，是dlib.vector类的对象
person_data1 = fc.savePersonData(face_rec, vector ) # 将提取出的数据保存到data文件夹，为便于操作返回numpy数组，内容还是一样的
# 导入第二张图片，并提取特征向量
face_rec.inputPerson(name='jobs2', img_path='\faces\jobs2.jpg')
vector = face_rec.create128DVectorSpace() # 提取128维向量，是dlib.vector类的对象
person_data2 = fc.savePersonData(face_rec, vector )
# 计算欧式距离，判断是否是同一个人
fc.comparePersonData(person_data1, person_data2)
1
2
3
4
5
6
7
8
9
10
11
12
13
如果data文件夹中已经有了模型文件，可以直接导入：
import face_rec as fc
face_rec = fc.face_recognition() # 创建对象
person_data1 = fc.loadPersonData(face_rec , 'jobs') # 创建一个类保存相关信息，后面还要跟上人名，程序会在data文件中查找对应npy文件，比如这里就是'jobs.npy'
person_data2 = fc.loadPersonData(face_rec , 'jobs2') # 导入第二张图片
fc.comparePersonData(person_data1, person_data2) # 计算欧式距离，判断是否是同一个人
1
2
3
4
5
程序2结果
Python 2.7.10 |Anaconda 2.3.0 (64-bit)| (default, May 28 2015, 16:44:52) [MSC v.1500 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://binstar.org
>>> import face_rec as fc
>>> face_rec = fc.face_recognition()
>>> face_rec.inputPerson(name='jobs', img_path='\faces\jobs.jpg')
>>> vector = face_rec.create128DVectorSpace()
Number of faces detected: 1
face 0; left 184; top 64; right 339; bottom 219
>>> person_data1 = fc.savePersonData(face_rec, vector )
Saving files to :F:Pythonmy_dlib_codesface_recognitiondatajobs.npy
>>> face_rec.inputPerson(name='jobs2', img_path='\faces\jobs2.jpg')
>>> vector = face_rec.create128DVectorSpace()
Number of faces detected: 1
face 0; left 124; top 39; right 253; bottom 168
>>> person_data2 = fc.savePersonData(face_rec, vector )
Saving files to :F:Pythonmy_dlib_codesface_recognitiondatajobs2.npy
>>> fc.comparePersonData(person_data1, person_data2)
0.490491048429
It's the same person
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
官方例程
#!/usr/bin/python
# The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt
#
# This example shows how to use dlib's face recognition tool. This tool maps
# an image of a human face to a 128 dimensional vector space where images of
# the same person are near to each other and images from different people are
# far apart. Therefore, you can perform face recognition by mapping faces to
# the 128D space and then checking if their Euclidean distance is small
# enough.
#
# When using a distance threshold of 0.6, the dlib model obtains an accuracy
# of 99.38% on the standard LFW face recognition benchmark, which is
# comparable to other state-of-the-art methods for face recognition as of
# February 2017. This accuracy means that, when presented with a pair of face
# images, the tool will correctly identify if the pair belongs to the same
# person or is from different people 99.38% of the time.
#
# Finally, for an in-depth discussion of how dlib's tool works you should
# refer to the C++ example program dnn_face_recognition_ex.cpp and the
# attendant documentation referenced therein.
#
#
#
#
# COMPILING/INSTALLING THE DLIB PYTHON INTERFACE
# You can install dlib using the command:
# pip install dlib
#
# Alternatively, if you want to compile dlib yourself then go into the dlib
# root folder and run:
# python setup.py install
# or
# python setup.py install --yes USE_AVX_INSTRUCTIONS
# if you have a CPU that supports AVX instructions, since this makes some
# things run faster. This code will also use CUDA if you have CUDA and cuDNN
# installed.
#
# Compiling dlib should work on any operating system so long as you have
# CMake and boost-python installed. On Ubuntu, this can be done easily by
# running the command:
# sudo apt-get install libboost-python-dev cmake
#
# Also note that this example requires scikit-image which can be installed
# via the command:
# pip install scikit-image
# Or downloaded from http://scikit-image.org/download.html.
import sys
import os
import dlib
import glob
from skimage import io
if len(sys.argv) != 4:
print(
"Call this program like this:
"
" ./face_recognition.py shape_predictor_68_face_landmarks.dat dlib_face_recognition_resnet_model_v1.dat ../examples/faces
"
"You can download a trained facial shape predictor and recognition model from:
"
" http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
"
" http://dlib.net/files/dlib_face_recognition_resnet_model_v1.dat.bz2")
exit()
predictor_path = sys.argv[1]
face_rec_model_path = sys.argv[2]
faces_folder_path = sys.argv[3]
# Load all the models we need: a detector to find the faces, a shape predictor
# to find face landmarks so we can precisely localize the face, and finally the
# face recognition model.
detector = dlib.get_frontal_face_detector()
sp = dlib.shape_predictor(predictor_path)
facerec = dlib.face_recognition_model_v1(face_rec_model_path)
win = dlib.image_window()
# Now process all the images
for f in glob.glob(os.path.join(faces_folder_path, "*.jpg")):
print("Processing file: {}".format(f))
img = io.imread(f)
win.clear_overlay()
win.set_image(img)
# Ask the detector to find the bounding boxes of each face. The 1 in the
# second argument indicates that we should upsample the image 1 time. This
# will make everything bigger and allow us to detect more faces.
dets = detector(img, 1)
print("Number of faces detected: {}".format(len(dets)))
# Now process each face we found.
for k, d in enumerate(dets):
print("Detection {}: Left: {} Top: {} Right: {} Bottom: {}".format(
k, d.left(), d.top(), d.right(), d.bottom()))
# Get the landmarks/parts for the face in box d.
shape = sp(img, d)
# Draw the face landmarks on the screen so we can see what face is currently being processed.
win.clear_overlay()
win.add_overlay(d)
win.add_overlay(shape)
# Compute the 128D vector that describes the face in img identified by
# shape. In general, if two face descriptor vectors have a Euclidean
# distance between them less than 0.6 then they are from the same
# person, otherwise they are from different people. Here we just print
# the vector to the screen.
face_descriptor = facerec.compute_face_descriptor(img, shape)
print(face_descriptor)
# It should also be noted that you can also call this function like this:
# face_descriptor = facerec.compute_face_descriptor(img, shape, 100)
# The version of the call without the 100 gets 99.13% accuracy on LFW
# while the version with 100 gets 99.38%. However, the 100 makes the
# call 100x slower to execute, so choose whatever version you like. To
# explain a little, the 3rd argument tells the code how many times to
# jitter/resample the image. When you set it to 100 it executes the
# face descriptor extraction 100 times on slightly modified versions of
# the face and returns the average result. You could also pick a more
# middle value, such as 10, which is only 10x slower but still gets an
# LFW accuracy of 99.3%.
dlib.hit_enter_to_continue()
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
吐槽：
dlib的确很方便，不用花多少时间就能自己做到一些目标功能。官方文档讲的很详细，很容易入门。看这个文档（dlib python api）差不多就能学会用了。导师已经安排了研究生阶段的学习任务了，后面也要忙起来了。dlib的学习虽然是我10月份才开的坑，为了善始善终我也要尽快整理完这些东西。以后要回到”泡馆”生活了。
原文链接：https://blog.csdn.net/hongbin_xu/article/details/78390982