在本节的内容中,将通过具体实例的实现过程,详细讲解使用face_recognition实现基本人脸检测的知识。
5.2.1 输出显示指定人像人脸特征
库face_recognition通过facial_features来处理面部特征,包含了如下八个特征:
- chin:下巴
- left_eyebrow:左眉;
- right_eyebrow:右眉;
- nose_bridge:鼻梁;
- nose_tip:鼻子尖;
- left_eye:左眼;
- right_eye:右眼;
- top_lip:上唇;
- bottom_lip:下唇。
例如在下面的实例文件shibie01.py中,演示了输出显示指定人像人脸特征的过程。
实例5-1:输出显示指定人像人脸特征
源码路径:daima\5\5-1\shibie01.py
# 自动识别人脸特征
# 导入pil模块
from PIL import Image, ImageDraw
# 导入face_recogntion模块,可用命令安装 pip install face_recognition
import face_recognition
# 将jpg文件加载到numpy 数组中
image = face_recognition.load_image_file("111.jpg")
#查找图像中所有面部的所有面部特征
face_landmarks_list = face_recognition.face_landmarks(image)
print("I found {} face(s) in this photograph.".format(len(face_landmarks_list)))
for face_landmarks in face_landmarks_list:
#打印此图像中每个面部特征的位置
facial_features = [
'chin',
'left_eyebrow',
'right_eyebrow',
'nose_bridge',
'nose_tip',
'left_eye',
'right_eye',
'top_lip',
'bottom_lip'
]
for facial_feature in facial_features:
print("The {} in this face has the following points: {}".format(facial_feature, face_landmarks[facial_feature]))
#让我们在图像中描绘出每个人脸特征!
pil_image = Image.fromarray(image)
d = ImageDraw.Draw(pil_image)
for facial_feature in facial_features:
d.line(face_landmarks[facial_feature], width=5)
pil_image.show()
执行后会输出显示图片111.jpg中人像的人脸特征数值:
I found 1 face(s) in this photograph.
The chin in this face has the following points: [(35, 303), (38, 331), (42, 359), (47, 387), (59, 411), (78, 428), (100, 441), (123, 451), (146, 454), (168, 450), (189, 439), (209, 425), (227, 407), (238, 384), (244, 358), (249, 331), (252, 304)]
The left_eyebrow in this face has the following points: [(53, 289), (66, 273), (87, 266), (109, 269), (131, 276)]
The right_eyebrow in this face has the following points: [(162, 277), (181, 269), (203, 266), (224, 271), (236, 286)]
The nose_bridge in this face has the following points: [(144, 303), (144, 319), (144, 334), (143, 351)]
The nose_tip in this face has the following points: [(124, 364), (134, 366), (144, 368), (154, 366), (164, 364)]
The left_eye in this face has the following points: [(77, 304), (89, 299), (103, 300), (115, 310), (102, 313), (87, 311)]
The right_eye in this face has the following points: [(174, 310), (185, 301), (199, 301), (211, 305), (200, 312), (186, 313)]
The top_lip in this face has the following points: [(109, 395), (124, 391), (136, 387), (144, 389), (153, 386), (165, 390), (180, 393), (174, 393), (153, 394), (144, 395), (136, 394), (116, 396)]
The bottom_lip in this face has the following points: [(180, 393), (165, 402), (154, 405), (145, 406), (136, 406), (125, 404), (109, 395), (116, 396), (136, 396), (145, 396), (153, 395), (174, 393)]
并且会使用PIL在人像中标记出人脸特征,如图5-1所示。
图5-1 标记出人脸特征
5.2.2 在指定照片中识别标记出人脸
在下面的实例文件shibie.py中,演示了在指定照片中识别标记出人脸的过程。
实例5-2:在指定照片中识别标记出人脸
源码路径:daima\5\5-2\shibie.py
# 检测人脸
import face_recognition
import cv2
# 读取图片并识别人脸
img = face_recognition.load_image_file("111.jpg")
face_locations = face_recognition.face_locations(img)
print(face_locations)
# 调用opencv函数显示图片
img = cv2.imread("111.jpg")
cv2.namedWindow("原图")
cv2.imshow("原图", img)
# 遍历每个人脸,并标注
faceNum = len(face_locations)
for i in range(0, faceNum):
top = face_locations[i][0]
right = face_locations[i][1]
bottom = face_locations[i][2]
left = face_locations[i][3]
start = (left, top)
end = (right, bottom)
color = (55,255,155)
thickness = 3
cv2.rectangle(img, start, end, color, thickness)
# 显示识别结果
cv2.namedWindow("识别")
cv2.imshow("识别", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
执行后将分别显示原始照片111.jpg效果和识别标记处人脸的效果,如图5-2所示。
图5-2 原始照片效果和标记出人脸效果
5.2.3 识别出照片中的所有人脸
假设我们有一张照片888.jpg,如图5-3所示。
图5-3 照片888.jpg
实例5-3:识别出指定照片中的人脸
这是一幅3人合影照,我们应该如何识别出这张照片中的人脸呢?在下面的实例文件shibie02.py中,演示了提取出照片888.jpg中所有人脸的过程。
源码路径:daima\5\5-3\shibie02.py
# 识别图片中的所有人脸并显示出来
# filename : shibie02.py
# 导入pil模块
from PIL import Image
# 导入face_recogntion模块,可用命令安装 pip install face_recognition
import face_recognition
# 将jpg文件加载到numpy 数组中
image = face_recognition.load_image_file("888.jpg")
# 使用默认的给予HOG模型查找图像中所有人脸
# 这个方法已经相当准确了,但还是不如CNN模型那么准确,因为没有使用GPU加速
# 另请参见: find_faces_in_picture_cnn.py
face_locations = face_recognition.face_locations(image)
# 使用CNN模型
# face_locations = face_recognition.face_locations(image, number_of_times_to_upsample=0, model="cnn")
# 打印:我从图片中找到了 多少 张人脸
print("I found {} face(s) in this photograph.".format(len(face_locations)))
# 循环找到的所有人脸
for face_location in face_locations:
# 打印每张脸的位置信息
top, right, bottom, left = face_location
print("Top: {}, Left: {}, Bottom: {}, Right: {}".format(top, left, bottom, right))
# 指定人脸的位置信息,然后显示人脸图片
face_image = image[top:bottom, left:right]
pil_image = Image.fromarray(face_image)
pil_image.show()
执行后首先会输出照片888.jpg中人脸的位置信息:
I found 3 face(s) in this photograph.
Top: 163, Left: 79, Bottom: 271, Right: 187
Top: 125, Left: 182, Bottom: 254, Right: 311
Top: 329, Left: 104, Bottom: 403, Right: 179
然后输出显示识别的三个人脸,如图5-4所示。
图5-4 提取出的3个人脸
这是一幅3人合影照,我们应该如何识别出这张照片中的人脸呢?在下面的实例文件shibie02-1.py中,演示了提取出照片888.jpg中所有人脸的过程。
实例5-4:识别出多人张照片中的人脸
源码路径:daima\5\5-3\shibie02-1.py
import face_recognition
import cv2
# 读取图片并识别人脸
img = face_recognition.load_image_file("888.jpg")
face_locations = face_recognition.face_locations(img)
print(face_locations)
# 调用opencv函数显示图片
img = cv2.imread("888.jpg")
cv2.namedWindow("原图")
cv2.imshow("原图", img)
# 遍历每个人脸,并标注
faceNum = len(face_locations)
for i in range(0, faceNum):
top = face_locations[i][0]
right = face_locations[i][1]
bottom = face_locations[i][2]
left = face_locations[i][3]
start = (left, top)
end = (right, bottom)
color = (55, 255, 155)
thickness = 3
cv2.rectangle(img, start, end, color, thickness)
# 显示识别结果
cv2.namedWindow("识别")
cv2.imshow("识别", img)
cv2.waitKey(0)
cv2.destroyAllWindows()
执行后会输出显示照片888.jpg中的所有的人脸信息,如图5-5所示。
图5-5 识别出的3个人脸
5.2.4 判断在照片中是否包含某个人脸
假设我们有一张照片201.jpg,如图5-6所示。
图5-6 照片201.jpg
这是一幅单人合影照,假设这个人的名字叫小毛毛。请问我们应该如何识别出在照片888.jpg中有小毛毛呢?在下面的实例文件shibie03.py中,演示了识别判断在照片888.jpg中是否包含小毛毛人脸的过程。
实例5-5:识别判断在照片888.jpg中是否包含小毛毛人脸
源码路径:daima\5\5-4\shibie03.py
# 识别人脸鉴定是哪个人
import face_recognition
#将jpg文件加载到numpy数组中
chen_image = face_recognition.load_image_file("201.jpg")
#要识别的图片
unknown_image = face_recognition.load_image_file("888.jpg")
#获取每个图像文件中每个面部的面部编码
#由于每个图像中可能有多个面,所以返回一个编码列表。
#但是由于我知道每个图像只有一个脸,我只关心每个图像中的第一个编码,所以我取索引0。
chen_face_encoding = face_recognition.face_encodings(chen_image)[0]
print("chen_face_encoding:{}".format(chen_face_encoding))
unknown_face_encoding = face_recognition.face_encodings(unknown_image)[0]
print("unknown_face_encoding :{}".format(unknown_face_encoding))
known_faces = [
chen_face_encoding
]
#结果是True/false的数组,未知面孔known_faces阵列中的任何人相匹配的结果
results = face_recognition.compare_faces(known_faces, unknown_face_encoding)
print("result :{}".format(results))
print("这个未知面孔是 小毛毛 吗? {}".format(results[0]))
print("这个未知面孔是 我们从未见过的新面孔吗? {}".format(not True in results))
执行后会输出如下识别结果,这说明在照片888.jpg中存在照片201.jpg的这个人。
result :[True]
这个未知面孔是 小毛毛 吗? True
这个未知面孔是 我们从未见过的新面孔吗? False
假设分别存在3张图片laoguan.jpg(老管的单人照)、maomao.jpg(毛毛的单人照)和unknown.jpg(未知某人的单人照,肯定是老管或毛毛这两人之一),如图5-7所示。
图5-7 三张素材图片
5.2.5 识别出在照片中的人到底是谁
我们应该如何识别出照片unknown.jpg中的人是谁呢?在下面的实例文件shibie04.py中,演示了识别判断在照片unknown.jpg中的人到底是谁的过程。
实例5-6:识别判断在照片unknown.jpg中的人到底是谁
源码路径:daima\5\5-5\shibie04.py
import face_recognition
jobs_image = face_recognition.load_image_file("laoguan.jpg");
obama_image = face_recognition.load_image_file("maomao.jpg");
unknown_image = face_recognition.load_image_file("unknown.jpg");
laoguan_encoding = face_recognition.face_encodings(jobs_image)[0]
maomao_encoding = face_recognition.face_encodings(obama_image)[0]
unknown_encoding = face_recognition.face_encodings(unknown_image)[0]
results = face_recognition.compare_faces([laoguan_encoding, maomao_encoding], unknown_encoding )
labels = ['老管', '毛毛']
print('结果:'+str(results))
for i in range(0, len(results)):
if results[i] == True:
print('这个人是:'+labels[i])
执行后会成功输出识别别结果:
结果:[False, True]
这个人是:毛毛
5.2.6 摄像头实时识别
假设我们保存一张“小毛毛”的照片xiaomaomao.jpg,然后用摄像头识别不同的照片,如果是小毛毛本人的照片,则摄像区域自动识别并显示“小毛毛”。如果摄像头不是小毛毛的照片,则在摄像区域显示“unknown”。通过下面的实例文件shibie05.py可以实现上述识别功能。
实例5-7:用摄像头识别不同的照片
源码路径:daima\5\5-6\shibie05.py
import face_recognition
import cv2
video_capture = cv2.VideoCapture(0)#笔记本摄像头是0,外接摄像头设备是1
obama_img = face_recognition.load_image_file("xiaomaomao.jpg")
obama_face_encoding = face_recognition.face_encodings(obama_img)[0]
face_locations = []
face_encodings = []
face_names = []
process_this_frame = True
while True:
ret, frame = video_capture.read()
small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25)
if process_this_frame:
face_locations = face_recognition.face_locations(small_frame)
face_encodings = face_recognition.face_encodings(small_frame, face_locations)
face_names = []
for face_encoding in face_encodings:
match = face_recognition.compare_faces([obama_face_encoding], face_encoding)
if match[0]:
name = "小毛毛"
else:
name = "unknown"
face_names.append(name)
process_this_frame = not process_this_frame
for (top, right, bottom, left), name in zip(face_locations, face_names):
top *= 4
right *= 4
bottom *= 4
left *= 4
cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 2)
cv2.rectangle(frame, (left, bottom - 35), (right, bottom), (0, 0, 255), 2)
font = cv2.FONT_HERSHEY_DUPLEX
cv2.putText(frame, name, (left+6, bottom-6), font, 1.0, (255, 255, 255), 1)
cv2.imshow('Video', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
cv2.destroyAllWindows()
再看下面的实例文件facerec_from_webcam.py,功能是实时识别摄像头中的人脸。本实例使用库OpenCV从摄像头读取视频。首先准备两幅素材图片obama.jpg和biden.jpg,然后识别摄像头中的人脸。如果摄像头中的人脸是两幅素材图片obama.jpg和biden.jpg的人脸,则在摄像头视频中用矩形标签注明识别结果。如果如果摄像头中的人脸不是两幅素材图片obama.jpg和biden.jpg的人脸,则在摄像头视频中用矩形标签注明“Unknown”。
实例5-8:实时识别摄像头中的人脸
源码路径:daima\5\5-6\facerec_from_webcam.py
import face_recognition
import cv2
import numpy as np
video_capture = cv2.VideoCapture(0)
#加载实例图片并学习如何识别它
obama_image = face_recognition.load_image_file("obama.jpg")
obama_face_encoding = face_recognition.face_encodings(obama_image)[0]
#加载第二个实例图片并学习如何识别它。
biden_image = face_recognition.load_image_file("biden.jpg")
biden_face_encoding = face_recognition.face_encodings(biden_image)[0]
#创建已知面编码及其名称的数组
known_face_encodings = [
obama_face_encoding,
biden_face_encoding
]
known_face_names = [
"Barack Obama",
"Joe Biden"
]
while True:
#抓取一帧视频
ret, frame = video_capture.read()
#将图像从BGR颜色(OpenCV使用)转换为RGB颜色(人脸识别可以使用的颜色)
rgb_frame = frame[:, :, ::-1]
#找到视频帧中的所有人脸和人脸编码
face_locations = face_recognition.face_locations(rgb_frame)
face_encodings = face_recognition.face_encodings(rgb_frame, face_locations)
#在这一帧视频中遍历每个人脸
for (top, right, bottom, left), face_encoding in zip(face_locations, face_encodings):
#查看该脸是否与已知人脸匹配
matches = face_recognition.compare_faces(known_face_encodings, face_encoding)
name = "Unknown"
#或者使用与新人脸距离最小的已知人脸
face_distances = face_recognition.face_distance(known_face_encodings, face_encoding)
best_match_index = np.argmin(face_distances)
if matches[best_match_index]:
name = known_face_names[best_match_index]
# 在人脸上画一个方框
cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 2)
# 在人脸下方绘制一个带有名称的标签
cv2.rectangle(frame, (left, bottom - 35), (right, bottom), (0, 0, 255), cv2.FILLED)
font = cv2.FONT_HERSHEY_DUPLEX
cv2.putText(frame, name, (left + 6, bottom - 6), font, 1.0, (255, 255, 255), 1)
#显示结果图像
cv2.imshow('Video', frame)
#按下键盘中的q键将退出
if cv2.waitKey(1) & 0xFF == ord('q'):
break
#释放资源
video_capture.release()
cv2.destroyAllWindows()
执行后的效果如图5-8所示。
图5-8 执行效果
上述实例文件facerec_from_webcam_faster.py的效率一般,比较消耗计算机内存资源。在下面的实例文件中,我们对基本的识别功能进行了如下调整,使得识别速度更加快速。
实例5-9:实时识别摄像头中的人脸高效版
(1)以1/4分辨率处理每个视频帧,仍以全分辨率显示。
(2)每隔一帧视频检测人脸。
源码路径:daima\5\5-6\facerec_from_webcam_faster.py
import face_recognition
import cv2
import numpy as np
video_capture = cv2.VideoCapture(0)
#加载实例图片并学习如何识别它
obama_image = face_recognition.load_image_file("obama.jpg")
obama_face_encoding = face_recognition.face_encodings(obama_image)[0]
#加载第二个实例图片并学习如何识别它。
biden_image = face_recognition.load_image_file("biden.jpg")
biden_face_encoding = face_recognition.face_encodings(biden_image)[0]
#创建已知面编码及其名称的数组
known_face_encodings = [
obama_face_encoding,
biden_face_encoding
]
known_face_names = [
"Barack Obama",
"Joe Biden"
]
#初始化一些变量
face_locations = []
face_encodings = []
face_names = []
process_this_frame = True
while True:
#抓取一帧视频
ret, frame = video_capture.read()
#将视频帧调整为1/4大小以快速实现人脸识别
small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25)
#将图像从BGR颜色(OpenCV使用)转换为RGB颜色(人脸识别使用)
rgb_small_frame = small_frame[:, :, ::-1]
#每隔一帧的处理视频以节省时间
if process_this_frame:
#查找当前视频帧中的所有面和面编码
face_locations = face_recognition.face_locations(rgb_small_frame)
face_encodings = face_recognition.face_encodings(rgb_small_frame, face_locations)
face_names = []
for face_encoding in face_encodings:
# 查看该脸是否与已知人脸匹配
matches = face_recognition.compare_faces(known_face_encodings, face_encoding)
name = "Unknown"
face_distances = face_recognition.face_distance(known_face_encodings, face_encoding)
best_match_index = np.argmin(face_distances)
if matches[best_match_index]:
name = known_face_names[best_match_index]
face_names.append(name)
process_this_frame = not process_this_frame
#显示结果
for (top, right, bottom, left), name in zip(face_locations, face_names):
#缩放脸部位置,因为我们在中检测到的帧已缩放为1/4大小
top *= 4
right *= 4
bottom *= 4
left *= 4
# 在脸上画一个方框
cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 2)
#在脸部下方绘制一个带有名称的标签
cv2.rectangle(frame, (left, bottom - 35), (right, bottom), (0, 0, 255), cv2.FILLED)
font = cv2.FONT_HERSHEY_DUPLEX
cv2.putText(frame, name, (left + 6, bottom - 6), font, 1.0, (255, 255, 255), 1)
#显示结果图像
cv2.imshow('Video', frame)
#按下键盘中的q键将退出
if cv2.waitKey(1) & 0xFF == ord('q'):
break
video_capture.release()
cv2.destroyAllWindows()