Python 3 利用 Dlib 19.7 實現攝像頭人臉檢測特征點標定
阿新 • • 發佈：2018-02-26
backend 開發 img com inter 適合 github red for
0.引言
利用python開發，借助Dlib庫捕獲攝像頭中的人臉，進行實時特征點標定；
圖1 工程效果示例（gif）
圖2 工程效果示例（靜態圖片）
（實現比較簡單，代碼量也比較少，適合入門或者興趣學習。）
1.開發環境
python: 3.6.3
dlib: 19.7
OpenCv, numpy
1 import dlib # 人臉識別的庫dlib 2 import numpy as np # 數據處理的庫numpy 3 import cv2 # 圖像處理的庫OpenCv
2.源碼介紹
其實實現很簡單，主要分為兩個部分：攝像頭調用+人臉特征點標定
2.1 攝像頭調用
介紹下opencv中攝像頭的調用方法；
利用 cap = cv2.VideoCapture(0) 創建一個對象；
（具體可以參考官方文檔：https://docs.opencv.org/2.4/modules/highgui/doc/reading_and_writing_images_and_video.html）
1 # 2018-2-26 2 # By TimeStamp 3 # cnblogs: http://www.cnblogs.com/AdaminXie 4 5 """ 6 cv2.VideoCapture(), 創建cv2攝像頭對象/ open the default camera7 8 Python: cv2.VideoCapture() → <VideoCapture object> 9 10 Python: cv2.VideoCapture(filename) → <VideoCapture object> 11 filename – name of the opened video file (eg. video.avi) or image sequence (eg. img_%02d.jpg, which will read samples like img_00.jpg, img_01.jpg, img_02.jpg, ...)12 13 Python: cv2.VideoCapture(device) → <VideoCapture object> 14 device – id of the opened video capturing device (i.e. a camera index). If there is a single camera connected, just pass 0. 15 16 """ 17 cap = cv2.VideoCapture(0) 18 19 20 """ 21 cv2.VideoCapture.set(propId, value)，設置視頻參數; 22 23 propId: 24 CV_CAP_PROP_POS_MSEC Current position of the video file in milliseconds. 25 CV_CAP_PROP_POS_FRAMES 0-based index of the frame to be decoded/captured next. 26 CV_CAP_PROP_POS_AVI_RATIO Relative position of the video file: 0 - start of the film, 1 - end of the film. 27 CV_CAP_PROP_FRAME_WIDTH Width of the frames in the video stream. 28 CV_CAP_PROP_FRAME_HEIGHT Height of the frames in the video stream. 29 CV_CAP_PROP_FPS Frame rate. 30 CV_CAP_PROP_FOURCC 4-character code of codec. 31 CV_CAP_PROP_FRAME_COUNT Number of frames in the video file. 32 CV_CAP_PROP_FORMAT Format of the Mat objects returned by retrieve() . 33 CV_CAP_PROP_MODE Backend-specific value indicating the current capture mode. 34 CV_CAP_PROP_BRIGHTNESS Brightness of the image (only for cameras). 35 CV_CAP_PROP_CONTRAST Contrast of the image (only for cameras). 36 CV_CAP_PROP_SATURATION Saturation of the image (only for cameras). 37 CV_CAP_PROP_HUE Hue of the image (only for cameras). 38 CV_CAP_PROP_GAIN Gain of the image (only for cameras). 39 CV_CAP_PROP_EXPOSURE Exposure (only for cameras). 40 CV_CAP_PROP_CONVERT_RGB Boolean flags indicating whether images should be converted to RGB. 41 CV_CAP_PROP_WHITE_BALANCE_U The U value of the whitebalance setting (note: only supported by DC1394 v 2.x backend currently) 42 CV_CAP_PROP_WHITE_BALANCE_V The V value of the whitebalance setting (note: only supported by DC1394 v 2.x backend currently) 43 CV_CAP_PROP_RECTIFICATION Rectification flag for stereo cameras (note: only supported by DC1394 v 2.x backend currently) 44 CV_CAP_PROP_ISO_SPEED The ISO speed of the camera (note: only supported by DC1394 v 2.x backend currently) 45 CV_CAP_PROP_BUFFERSIZE Amount of frames stored in internal buffer memory (note: only supported by DC1394 v 2.x backend currently) 46 47 value: 設置的參數值/ Value of the property 48 """ 49 cap.set(3, 480) 50 51 """ 52 cv2.VideoCapture.isOpened(), 檢查攝像頭初始化是否成功 / check if we succeeded 53 返回true或false 54 """ 55 cap.isOpened() 56 57 """ 58 cv2.VideoCapture.read([imgage]) -> retval,image, 讀取視頻 / Grabs, decodes and returns the next video frame 59 返回兩個值： 60 一個是布爾值true/false，用來判斷讀取視頻是否成功/是否到視頻末尾 61 圖像對象，圖像的三維矩陣 62 """ 63 flag, im_rd = cap.read()
2.2 人臉特征點標定
調用預測器“shape_predictor_68_face_landmarks.dat”進行68點標定，這是dlib訓練好的模型，可以直接調用進行人臉68個人臉特征點的標定；
具體可以參考我的另一篇博客（http://www.cnblogs.com/AdaminXie/p/8137580.html）；
2.3 源碼
實現的方法比較簡單：
利用 cv2.VideoCapture() 創建攝像頭對象，然後利用 flag, im_rd = cv2.VideoCapture.read() 讀取攝像頭視頻，im_rd就是視頻中的一幀幀圖像；
然後就類似於單張圖像進行人臉檢測，對這一幀幀的圖像im_rd利用dlib進行特征點標定，然後繪制特征點；
你可以按下s鍵來獲取當前截圖，或者按下q鍵來退出攝像頭；
1 # 2018-2-26 2 # By TimeStamp 3 # cnblogs: http://www.cnblogs.com/AdaminXie 4 # github: https://github.com/coneypo/Dlib_face_detection_from_camera 5 6 import dlib #人臉識別的庫dlib 7 import numpy as np #數據處理的庫numpy 8 import cv2 #圖像處理的庫OpenCv 9 10 # dlib預測器 11 detector = dlib.get_frontal_face_detector() 12 predictor = dlib.shape_predictor(‘shape_predictor_68_face_landmarks.dat‘) 13 14 # 創建cv2攝像頭對象 15 cap = cv2.VideoCapture(0) 16 17 # cap.set(propId, value) 18 # 設置視頻參數，propId設置的視頻參數，value設置的參數值 19 cap.set(3, 480) 20 21 # 截圖screenshoot的計數器 22 cnt = 0 23 24 # cap.isOpened（） 返回true/false 檢查初始化是否成功 25 while(cap.isOpened()): 26 27 # cap.read() 28 # 返回兩個值： 29 # 一個布爾值true/false，用來判斷讀取視頻是否成功/是否到視頻末尾 30 # 圖像對象，圖像的三維矩陣 31 flag, im_rd = cap.read() 32 33 # 每幀數據延時1ms，延時為0讀取的是靜態幀 34 k = cv2.waitKey(1) 35 36 # 取灰度 37 img_gray = cv2.cvtColor(im_rd, cv2.COLOR_RGB2GRAY) 38 39 # 人臉數rects 40 rects = detector(img_gray, 0) 41 42 #print(len(rects)) 43 44 # 待會要寫的字體 45 font = cv2.FONT_HERSHEY_SIMPLEX 46 47 # 標68個點 48 if(len(rects)!=0): 49 # 檢測到人臉 50 for i in range(len(rects)): 51 landmarks = np.matrix([[p.x, p.y] for p in predictor(im_rd, rects[i]).parts()]) 52 53 for idx, point in enumerate(landmarks): 54 # 68點的坐標 55 pos = (point[0, 0], point[0, 1]) 56 57 # 利用cv2.circle給每個特征點畫一個圈，共68個 58 cv2.circle(im_rd, pos, 2, color=(0, 255, 0)) 59 60 # 利用cv2.putText輸出1-68 61 cv2.putText(im_rd, str(idx + 1), pos, font, 0.2, (0, 0, 255), 1, cv2.LINE_AA) 62 cv2.putText(im_rd, "faces: "+str(len(rects)), (20,50), font, 1, (0, 0, 255), 1, cv2.LINE_AA) 63 else: 64 # 沒有檢測到人臉 65 cv2.putText(im_rd, "no face", (20, 50), font, 1, (0, 0, 255), 1, cv2.LINE_AA) 66 67 # 添加說明 68 im_rd = cv2.putText(im_rd, "s: screenshot", (20, 400), font, 0.8, (255, 255, 255), 1, cv2.LINE_AA) 69 im_rd = cv2.putText(im_rd, "q: quit", (20, 450), font, 0.8, (255, 255, 255), 1, cv2.LINE_AA) 70 71 # 按下s鍵保存 72 if (k == ord(‘s‘)): 73 cnt+=1 74 cv2.imwrite("screenshoot"+str(cnt)+".jpg", im_rd) 75 76 # 按下q鍵退出 77 if(k==ord(‘q‘)): 78 break 79 80 # 窗口顯示 81 cv2.imshow("camera", im_rd) 82 83 # 釋放攝像頭 84 cap.release() 85 86 # 刪除建立的窗口 87 cv2.destroyAllWindows()
# 請尊重他人勞動成果，轉載或者使用源碼請註明出處：http://www.cnblogs.com/AdaminXie
# 如果對您有幫助，歡迎在GitHub上star本項目: https://github.com/coneypo/Dlib_face_detection_from_camera
Python 3 利用 Dlib 19.7 實現攝像頭人臉檢測特征點標定