Face Recognition
我是真的没有想到人脸识别下限这么低。—— 桜井雪子
装库
- python
- python-opencv
- numpy
- pillow
Linux 和 MacOS 需要小心，不要和系统的 python 环境冲突
一个简单的方法：装一个 PyCharm，在 PyCharm 里面换源、装库
摄像头拍照
import cv2 as cv
import time
camera = cv.VideoCapture(0) # 摄像头
while True:
flag, frame = camera.read()
cv.imshow('save_image', frame)
if not flag:
break
key = cv.waitKey(1) & 0xFF
if key == ord('q'):
break
if key == ord('s'):
image_name = time.strftime("%F %a %T", time.localtime()) + '.jpg'
cv.imwrite('<图片所在目录>' + image_name, frame)
print(image_name + ' saved')
camera.release()
cv.destroyAllWindows()
haarcascades 人脸识别
图片
import cv2 as cv
def imageFaceRecognition(image):
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
faceLocation = faceDetector.detectMultiScale(image)
for x, y, w, h in faceLocation:
cv.rectangle(image, (x, y, w, h), color = (0, 255, 0), thickness = 2)
cv.putText(image, 'human', (x, y), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
cv.imshow('image', image)
image = cv.imread('<图片文件位置>')
face(image)
cv.waitKey(0)
cv.destroyAllWindows()
摄像头
import cv2 as cv
import time
camera = cv.VideoCapture(0) # 摄像头
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
while True:
flag, frame = camera.read()
if not flag:
break
key = cv.waitKey(1) & 0xFF
if key == ord('q'):
break
faceLocation = faceDetector.detectMultiScale(frame)
for x, y, w, h in faceLocation:
cv.rectangle(frame, (x, y, w, h), color = (0, 255, 0), thickness = 2)
cv.putText(frame, 'human', (x, y), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
cv.imshow('frame', frame)
camera.release()
cv.destroyAllWindows()
视频
import cv2 as cv
video = cv.VideoCapture('<视频文件>')
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
while True:
flag, frame = video.read()
if not flag:
break
key = cv.waitKey(1) & 0xFF
if key == ord('q'):
break
faceLocation = faceDetector.detectMultiScale(frame)
for x, y, w, h in faceLocation:
cv.rectangle(frame, (x, y, w, h), color = (0, 255, 0), thickness = 2)
cv.putText(frame, 'human', (x, y), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
cv.imshow('frame', frame)
video.release()
cv.destroyAllWindows()
人脸识别分类
~~说白了就是训练完能认出来这个人是谁，但是实在不知道怎么起这个小节的名字了~~
人脸图像集：Labeled Faces in the Wild
训练
import cv2 as cv
import numpy as np
import os
from PIL import Image
def getFaceIndex(path):
faceArray = []
indexArray = []
imagePaths = sorted(os.listdir(path))
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
index = 0
for imagePath in imagePaths:
grayImage = Image.open(path + '/' + imagePath).convert('L')
imageNPArray = np.array(grayImage, 'uint8')
faceLocation = faceDetect.detectMultiScale(imageNPArray)
for x, y, w, h in faceLocation:
faceArray.append(imageNPArray[y:y+h, x:x+w])
indexArray.append(index)
index += 1
return faceArray, indexArray
faceArray , indexArray = getFaceIndex('<训练集目录>')
faceRecognizer = cv.face.LBPHFaceRecognizer.create()
faceRecognizer.train(faceArray, np.array(indexArray))
faceRecognizer.write('训练出的 .yml 文件')
图片
import cv2 as cv
import os
def getImageFileName(path):
imageFileNameArray = []
imagePaths = os.listdir(path)
for imagePath in sorted(imagePaths):
imageFileName = imagePath.split('_0')[0]
imageFileNameArray.append(imageFileName)
return imageFileNameArray
faceRecognizer = cv.face.LBPHFaceRecognizer.create()
faceRecognizer.read('训练出的 .yml 文件')
def imageFaceClassifier(image):
imageGray = cv.cvtColor(image, cv.COLOR_BGR2GRAY)
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
faceLocation = faceDetector.detectMultiScale(imageGray)
for x, y, w, h in faceLocation:
cv.rectangle(image, (x, y, w, h), color = (0, 255, 0), thickness = 2)
index, score = faceRecognizer.predict(imageGray[y:y+h, x:x+w])
if score <= 80:
cv.putText(image, getImageFileName('<训练集目录>')[index], (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
else:
cv.putText(image, "unknown", (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 0.75, color = (0, 255, 0), thickness = 2) # 文字不能直接用中文
cv.imshow('image', image)
image = cv.imread('<待检测图片>')
imageFaceClassifier(image)
cv.waitKey(0)
cv.destroyAllWindows()
摄像头
import cv2 as cv
import os
def getImageFileName(path):
imageFileNameArray = []
imagePaths = os.listdir(path)
for imagePath in sorted(imagePaths):
imageFileName = imagePath.split('_0')[0]
imageFileNameArray.append(imageFileName)
return imageFileNameArray
faceRecognizer = cv.face.LBPHFaceRecognizer.create() # 获取识别器
faceRecognizer.read('训练出的 .yml 文件')
camera = cv.VideoCapture(0)
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
while True:
flag, frame = camera.read()
if not flag:
break
key = cv.waitKey(1) & 0xFF
if key == ord('q'):
break
frameGray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
faceLocation = faceDetector.detectMultiScale(frameGray)
for x, y, w, h in faceLocation:
cv.rectangle(frame, (x, y, w, h), color = (0, 255, 0), thickness = 2)
index, score = faceRecognizer.predict(frameGray[y:y+h, x:x+w])
if score <= 80:
cv.putText(frame, getImageFileName('<训练集目录>')[index], (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
else:
cv.putText(frame, "unknown", (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
cv.imshow('frame', frame)
camera.release()
cv.destroyAllWindows()
视频
import cv2 as cv
import os
def getImageFileName(path):
imageFileNameArray = []
imagePaths = os.listdir(path)
for imagePath in sorted(imagePaths):
imageFileName = imagePath.split('_0')[0]
imageFileNameArray.append(imageFileName)
return imageFileNameArray
faceRecognizer = cv.face.LBPHFaceRecognizer.create() # 获取识别器
faceRecognizer.read('训练出的 .yml 文件')
cap = cv.VideoCapture('<视频文件>')
faceDetector = cv.CascadeClassifier('<分类器 `.xml` 文件>')
while True:
flag, frame = cap.read()
if not flag:
break
key = cv.waitKey(1) & 0xFF
if key == ord('q'):
break
frameGray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
faceLocation = faceDetect.detectMultiScale(frameGray)
for x, y, w, h in faceLocation:
cv.rectangle(frame, (x, y, w, h), color = (0, 255, 0), thickness = 2)
index, score = faceRecognizer.predict(frameGray[y:y+h, x:x+w])
if score <= 75:
cv.putText(frame, getImageFileName('<训练集目录>')[index], (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
else:
cv.putText(frame, "unknown", (x + 10, y - 10), fontFace = cv.FONT_HERSHEY_SIMPLEX, fontScale = 1, color = (0, 255, 0), thickness = 2)
cv.imshow("frame", frame)
cap.release()
cv.destroyAllWindows()