opencvとpillowを用いた顔認証システム（デモあり）

#data capture
cam = cv2.VideoCapture(0) # capture image
cam.set(3, 640) # set video width
cam.set(4, 480) # set video height
face_detector = cv2.CascadeClassifier('haarcascade_frontalface_default.xml') # import classifier
# For each person, enter one numeric face id
face_id = input('\n input user id')
print("\n data collection in progress, please look at the camera slightly twisted")
# Initialize individual sampling face count
count = 0
while(True):
 ret, img = cam.read() # ret is whether the read was successful, it is a boolean value
 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)# Convert to grayscale
 faces = face_detector.detectMultiScale(gray, 1.3, 5,minSize=(100,100))
 for (x,y,w,h) in faces:# here faces is an array array or an empty tuple, for reasons I will analyze later
  cv2.rectangle(img, (x,y), (x+w,y+h), (255,0,0), 2)
  count += 1
 # Save the captured image into the datasets folder
  cv2.imwrite("dataset/User." + str(face_id) + '.' + str(count) + ".jpg", gray[y:y+h,x:x+w])
 cv2.imshow('image', img)
 k = cv2.waitKey(100) & 0xff # Press 'ESC' for exiting video
 if k == 27:
  break
 elif count >= 10: # Take 10 face sample and stop video
  break
# Do a bit of cleanup
print("\n [INFO] Exiting Program and cleanup stuff")
cam.release()
cv2.destroyAllWindows()

import cv2
#trainer
import numpy as np
from PIL import Image
import os
# Path for face image database
path = 'dataset'
recognizer = cv2.face.LBPHFaceRecognizer_create() # Import of recognizer
detector = cv2.CascadeClassifier("haarcascade_frontalface_default.xml")
# function to get the images and label data
def getImagesAndLabels(path):
 imagePaths = [os.path.join(path,f) for f in os.listdir(path)]# here os module can help us to build paths very well, it is recommended to check the usage of related functions first.
 faceSamples = []
 ids = []
 for imagePath in imagePaths:
  PIL_img = Image.open(imagePath).convert('L') # convert to grayscale
  img_numpy = np.array(PIL_img,'uint8')# convert to array
  id = int(os.path.split(imagePath)[-1].split(". ")[1])
  faces = detector.detectMultiScale(img_numpy)
  for (x,y,w,h) in faces:
   faceSamples.append(img_numpy[y:y+h,x:x+w])
   ids.append(id)
 return faceSamples,ids
print ("\n training data in, please wait")
faces,ids = getImagesAndLabels(path)
recognizer.train(faces, np.array(ids))
# Save the model into trainer/trainer.yml
recognizer.write('trainer/trainer.yml') # recognizer.save() worked on Mac, but not on Pi
# Print the numer of faces trained and end program
print("\n {0} faces trained. Program closed".format(len(np.unique(ids))))

# -*- coding: UTF-8 -*-
# recognizer
import cv2
import numpy as np
import os
from PIL import Image, ImageFont, ImageDraw
path_to_ttf = 'C:\Windows\Fonts\Microsoft YaHei UI\msyh.ttc'#ttc file is the font that supports Chinese, I will explain later.
font1= ImageFont.truetype(path_to_ttf, size=20)
recognizer = cv2.face.LBPHFaceRecognizer_create()
recognizer.read('trainer/trainer.yml')#read recognizer
cascadePath = "haarcascade_frontalface_default.xml"
faceCascade = cv2.CascadeClassifier(cascadePath);
font = cv2.FONT_HERSHEY_SIMPLEX
#iniciate id counter
id = 0
# names related to ids: example ==> Marcelo: id=1, etc
names = ['None', 'Duan Lin Chen', 'Paula', 'Ilza', 'Z', 'W']#because we won't just show your code name in face recognition but your specific information.
cam = cv2.VideoCapture(0)
cam.set(3, 640) # set video widht
cam.set(4, 480) # set video height
# minimum recognized face size, during recognition we don't capture information about passersby on the street that are far away from you, which can cause a lot of problems, so we only need to recognize people who want to participate in the recognition, and setting the minimum face recognition size can circumvent this
minW = 0.1*cam.get(3)
minH = 0.1*cam.get(4)
while True:
 ret, img = cam.read()
 gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
 faces = faceCascade.detectMultiScale(
 gray,
 scaleFactor = 1.2,
 minNeighbors = 5,
 minSize = (int(minW), int(minH)),
 ) # Related parameters can be set by searching
 for(x,y,w,h) in faces:
  cv2.rectangle(img, (x,y), (x+w,y+h), (0,255,0), 2)
  id, confidence = recognizer.predict(gray[y:y+h,x:x+w])
 # Check if confidence is less them 100 == > "0" is perfect match
  if (confidence < 100):
   id = names[id]
   confidence = " {0}%".format(round(100 - confidence))# confidence is the confidence index, equal to 100 - probability, I believe that everyone's probability statistics must be better than mine
  else:
   id = "unidentified"
   confidence = " {0}%".format(round(100 - confidence))
  img = Image.fromarray(img)
  draw = ImageDraw.Draw(img)
  draw.text(xy=(x+5,y-5), text=str(id), font=font1,fill=(255,255,255))
  img=np.array(img)
  cv2.putText(img, str(confidence), (x+5,y+h-5), font, 1, (255,255,0), 1)
 cv2.imshow('camera',img)
 k = cv2.waitKey(10) & 0xff # Press 'ESC' for exiting video
 if k == 27:
  break
# Do a bit of cleanup
print("\n [INFO] Exiting Program and cleanup stuff")
cam.release()
cv2.destroyAllWindows()