机器学习笔记 - 使用Face recognition、OpenCV、Python进行人脸识别

一、安装相关包

安装dlib、cmake、face_recognition。

face_recognition是世界上最简单的人脸识别库，使用 dlib 最先进的深度学习人脸识别技术构建。该模型的准确率为 99.38%。

二、获取人脸的128位数组编码

使用基本演绎法这两个演员的脸测试，只用了这两张正脸进行识别，如果要识别准确，得准备多种角度的照片，才能比较精准。

三、参考代码

# 使用Face recognition获取人脸128位数组from imutils.video import VideoStream
from imutils import paths
import face_recognition
import argparse
import pickle
import cv2
import os
import imutilsdef train():# grab the paths to the input images in our datasetprint("[INFO] quantifying faces...")imagePaths = ['C:Users/xia/Desktop/Watson.png','C:/Users/xia/Desktop/Sherlock.png']# initialize the list of known encodings and known namesknownEncodings = []knownNames = []# loop over the image pathsfor (i, imagePath) in enumerate(imagePaths):# extract the person name from the image pathprint("[INFO] processing image {}/{}".format(i + 1, len(imagePaths)))name = 'Watson'if i==1:name = 'Sherlock'# load the input image and convert it from BGR (OpenCV ordering)# to dlib ordering (RGB)image = cv2.imread(imagePath)rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)# detect the (x, y)-coordinates of the bounding boxes# corresponding to each face in the input imageboxes = face_recognition.face_locations(rgb, model='cnn')# compute the facial embedding for the faceencodings = face_recognition.face_encodings(rgb, boxes)# loop over the encodingsfor encoding in encodings:# add each encoding + name to our set of known names and# encodingsknownEncodings.append(encoding)knownNames.append(name)# dump the facial encodings + names to diskprint("[INFO] serializing encodings...")data = {"encodings": knownEncodings, "names": knownNames}f = open('C:/Users/xia/Desktop/encodings.pickle', "wb")f.write(pickle.dumps(data))f.close()

四、进行人脸识别测试

# 使用Face recognition获取人脸128位from imutils.video import VideoStream
from imutils import paths
import face_recognition
import argparse
import pickle
import cv2
import os
import imutils# 在图片上测试
def reimage():# load the known faces and embeddingsprint("[INFO] loading encodings...")data = pickle.loads(open('C:/Users/xia/Desktop/encodings.pickle', "rb").read())# load the input image and convert it from BGR to RGBimage = cv2.imread('C:/Users/xia/Desktop/v2.jpg')rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)# detect the (x, y)-coordinates of the bounding boxes corresponding# to each face in the input image, then compute the facial embeddings# for each faceprint("[INFO] recognizing faces...")boxes = face_recognition.face_locations(rgb, model='cnn')encodings = face_recognition.face_encodings(rgb, boxes)# initialize the list of names for each face detectednames = []# loop over the facial embeddingsfor encoding in encodings:# attempt to match each face in the input image to our known# encodingsmatches = face_recognition.compare_faces(data["encodings"], encoding)name = "Unknown"# check to see if we have found a matchif True in matches:# find the indexes of all matched faces then initialize a# dictionary to count the total number of times each face# was matchedmatchedIdxs = [i for (i, b) in enumerate(matches) if b]counts = {}# loop over the matched indexes and maintain a count for# each recognized face facefor i in matchedIdxs:name = data["names"][i]counts[name] = counts.get(name, 0) + 1# determine the recognized face with the largest number of# votes (note: in the event of an unlikely tie Python will# select first entry in the dictionary)name = max(counts, key=counts.get)# update the list of namesnames.append(name)# loop over the recognized facesfor ((top, right, bottom, left), name) in zip(boxes, names):# draw the predicted face name on the imagecv2.rectangle(image, (left, top), (right, bottom), (0, 255, 0), 2)y = top - 15 if top - 15 > 15 else top + 15cv2.putText(image, name, (left, y), cv2.FONT_HERSHEY_SIMPLEX,0.75, (0, 255, 0), 2)# show the output imagecv2.imshow("Image", image)cv2.waitKey(0)

五、摄像头实时测试

下面是实时视频的参考代码，使用cpu+hpg也是很卡，得上gpu测试。

# 使用Face recognition获取人脸128位from imutils.video import VideoStream
from imutils import paths
import face_recognition
import argparse
import pickle
import cv2
import os
import imutils# 实时视频，但是需要gpu，即使用cpu+hog，也会卡顿
def revideo():# load the known faces and embeddingsprint("[INFO] loading encodings...")data = pickle.loads(open('C:/Users/zyh/Desktop/encodings.pickle', "rb").read())# initialize the video stream and pointer to output video file, then# allow the camera sensor to warm upprint("[INFO] starting video stream...")vs = VideoStream(src=0).start()writer = None#time.sleep(2.0)# loop over frames from the video file streamwhile True:# grab the frame from the threaded video streamframe = vs.read()# convert the input frame from BGR to RGB then resize it to have# a width of 750px (to speedup processing)rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)rgb = imutils.resize(frame, width=750)r = frame.shape[1] / float(rgb.shape[1])# detect the (x, y)-coordinates of the bounding boxes# corresponding to each face in the input frame, then compute# the facial embeddings for each faceboxes = face_recognition.face_locations(rgb, model='hog')encodings = face_recognition.face_encodings(rgb, boxes)names = []# loop over the facial embeddingsfor encoding in encodings:# attempt to match each face in the input image to our known# encodingsmatches = face_recognition.compare_faces(data["encodings"],encoding)name = "Unknown"# check to see if we have found a matchif True in matches:# find the indexes of all matched faces then initialize a# dictionary to count the total number of times each face# was matchedmatchedIdxs = [i for (i, b) in enumerate(matches) if b]counts = {}# loop over the matched indexes and maintain a count for# each recognized face facefor i in matchedIdxs:name = data["names"][i]counts[name] = counts.get(name, 0) + 1# determine the recognized face with the largest number# of votes (note: in the event of an unlikely tie Python# will select first entry in the dictionary)name = max(counts, key=counts.get)# update the list of namesnames.append(name)# loop over the recognized facesfor ((top, right, bottom, left), name) in zip(boxes, names):# rescale the face coordinatestop = int(top * r)right = int(right * r)bottom = int(bottom * r)left = int(left * r)# draw the predicted face name on the imagecv2.rectangle(frame, (left, top), (right, bottom),(0, 255, 0), 2)y = top - 15 if top - 15 > 15 else top + 15cv2.putText(frame, name, (left, y), cv2.FONT_HERSHEY_SIMPLEX,0.75, (0, 255, 0), 2)# 写入视频# # if the video writer is None *AND* we are supposed to write# # the output video to disk initialize the writer# if writer is None and args["output"] is not None:#     fourcc = cv2.VideoWriter_fourcc(*"MJPG")#     writer = cv2.VideoWriter(args["output"], fourcc, 20,#                              (frame.shape[1], frame.shape[0]), True)# # if the writer is not None, write the frame with recognized# # faces to disk# if writer is not None:#     writer.write(frame)# check to see if we are supposed to display the output frame to# the screenif 1 > 0:cv2.imshow("Frame", frame)key = cv2.waitKey(1) & 0xFF# if the `q` key was pressed, break from the loopif key == ord("q"):break# do a bit of cleanupcv2.destroyAllWindows()vs.stop()# check to see if the video writer point needs to be releasedif writer is not None:writer.release()

六、打开视频测试

# 使用Face recognition获取人脸128位from imutils.video import VideoStream
from imutils import paths
import face_recognition
import argparse
import pickle
import cv2
import os
import imutilsdef revideo1():# load the known faces and embeddingsprint("[INFO] loading encodings...")data = pickle.loads(open('C:/Users/xia/Desktop/encodings.pickle', "rb").read())# initialize the video stream and pointer to output video file, thencap = cv2.VideoCapture('C:/Users/xia/Desktop/123.mp4')# loop over frames from the video file streamwhile(cap.isOpened()):# grab the frame from the threaded video streamret, frame = cap.read()# convert the input frame from BGR to RGB then resize it to have# a width of 750px (to speedup processing)rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)rgb = imutils.resize(frame, width=750)r = frame.shape[1] / float(rgb.shape[1])# detect the (x, y)-coordinates of the bounding boxes# corresponding to each face in the input frame, then compute# the facial embeddings for each faceboxes = face_recognition.face_locations(rgb, model='hog')encodings = face_recognition.face_encodings(rgb, boxes)names = []# loop over the facial embeddingsfor encoding in encodings:# attempt to match each face in the input image to our known# encodingsmatches = face_recognition.compare_faces(data["encodings"], encoding)name = "Unknown"# check to see if we have found a matchif True in matches:# find the indexes of all matched faces then initialize a# dictionary to count the total number of times each face# was matchedmatchedIdxs = [i for (i, b) in enumerate(matches) if b]counts = {}# loop over the matched indexes and maintain a count for# each recognized face facefor i in matchedIdxs:name = data["names"][i]counts[name] = counts.get(name, 0) + 1# determine the recognized face with the largest number# of votes (note: in the event of an unlikely tie Python# will select first entry in the dictionary)name = max(counts, key=counts.get)# update the list of namesnames.append(name)# loop over the recognized facesfor ((top, right, bottom, left), name) in zip(boxes, names):# rescale the face coordinatestop = int(top * r)right = int(right * r)bottom = int(bottom * r)left = int(left * r)# draw the predicted face name on the imagecv2.rectangle(frame, (left, top), (right, bottom),(0, 255, 0), 2)y = top - 15 if top - 15 > 15 else top + 15cv2.putText(frame, name, (left, y), cv2.FONT_HERSHEY_SIMPLEX,0.75, (0, 255, 0), 2)# check to see if we are supposed to display the output frame to# the screenif 1 > 0:cv2.imshow("Frame", frame)key = cv2.waitKey(1) & 0xFF# if the `q` key was pressed, break from the loopif key == ord("q"):break# do a bit of cleanupcap.release()cv2.destroyAllWindows()

七、理论参考

https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78https://medium.com/@ageitgey/machine-learning-is-fun-part-4-modern-face-recognition-with-deep-learning-c3cffc121d78

dlib C++ Library: High Quality Face Recognition with Deep Metric Learninghttp://blog.dlib.net/2017/02/high-quality-face-recognition-with-deep.html

八、其它（安装dlib gpu版本）：

git clone https://github.com/davisking/dlib.git
cd dlib
mkdir build
cd build
cmake .. -DDLIB_USE_CUDA=1 -DUSE_AVX_INSTRUCTIONS=1
cmake --build .
cd ..
python3.6 setup.py install --yes USE_AVX_INSTRUCTIONS --yes DLIB_USE_CUDA