用python玩玩人脸识别

lioncloud6 2年前
   <p>人脸识别,乍一听还是高大上的东西.</p>    <p>今天我们就分享下怎么玩?</p>    <p>不需要一大坨代码,只要你会装……包.</p>    <p>当然,我们的重点不是装包,归根结底,还是怎么玩.</p>    <p>Ok,那么,我们开始</p>    <h2>一、需要安装什么?</h2>    <p>1.OpenCV 一个图像处理的强大的包.官网 <a href="/misc/goto?guid=4959751276337147569" rel="nofollow,noindex">http://opencv.org/,最新的版本3.3.0</a></p>    <ul>     <li>不需要最新的版本直接brew install opencv3,我这里装好是3.2.0的</li>     <li>如果需要最新的版本3.3.0的话,请按照这里安装 <a href="/misc/goto?guid=4959751276424772601" rel="nofollow,noindex">http://www.pyimagesearch.com/2016/12/19/install-opencv-3-on-macos-with-homebrew-the-easy-way/</a></li>    </ul>    <p>2.Python2.7或者Python3.x</p>    <p>3.依赖包NumPY. 可以pip install numpy</p>    <p>4.依赖包face <em>recognition.主角就是他了.pip install face</em> recognition</p>    <p>注意事项参照地址 <a href="/misc/goto?guid=4959751276510775961" rel="nofollow,noindex">https://github.com/ageitgey/face_recognition</a></p>    <h2>二、怎么玩</h2>    <p>1.我们首先给定一张被照片</p>    <p><img src="https://simg.open-open.com/show/bfb81de5f13de6705b340560b2d9a4d6.png"></p>    <p>2.从摄像头捕捉的图像里获取检测出头像,跟上面的照片的图片进行对比.</p>    <p>先看下结果:</p>    <p><img src="https://simg.open-open.com/show/5ca1dd07ff354e2473057b8c2589932a.png"></p>    <p>从截下来的图片来看. 我自己因为没有放图片上去进行识别,所以是unknown的. 手机上的图片被识别到了写上chu(名字) 标签.</p>    <p>代码:</p>    <pre>  <code class="language-python">import face_recognition    import cv2    import time    #获取摄像头  video_capture = cv2.VideoCapture(0)    # 加载图片  obama_image = face_recognition.load_image_file("IMG_20170723_213850R.jpg")    # 图片先识别一遍人脸,后面用来比较  obama_face_encoding = face_recognition.face_encodings(obama_image)[0]    # 初始化变量  face_locations = []    face_encodings = []    face_names = []    process_this_frame = True    while True:        # 从摄像头获取图像      ret, frame = video_capture.read()        # 缩放1/4大小,方便快速处理      small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25)        # 保证只处理一遍      if process_this_frame:          #找出所有的人脸          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:              # 判断是否匹配给定的人脸              match = face_recognition.compare_faces([obama_face_encoding], face_encoding)              name = "Unknown"                if match[0]:                  name = "CHU<del>"                face_names.append(name)        process_this_frame = not process_this_frame          # 把识别到和为识别到的人脸进行标记      for (top, right, bottom, left), name in zip(face_locations, face_names):          # 原来是1/4的大小,现在放大4倍          top *= 4          right *= 4          bottom *= 4          left *= 4            # 给脸部画框          cv2.rectangle(frame, (left, top), (right, bottom), (0, 0, 255), 2)            # 脸部下面显示标签          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, 1.5, (255, 255, 255), 3)      #识别到时候进行抓屏          for name in face_names:          if name=="CHU</del>":               cv2.imwrite('img-%d.png'%int(time.time()), frame)        # 显示加了头像框视频      cv2.imshow('Video', frame)        # Hit 'q' on the keyboard to quit!      if cv2.waitKey(1) & 0xFF == ord('q'):          break    # 释放摄像头句柄  video_capture.release()    cv2.destroyAllWindows()</code></pre>    <p> </p>    <p>来自:http://tech.dianwoda.com/2017/08/11/yong-pythonwan-wan-ren-lian-shi-bie/</p>    <p> </p>