使用Python+OpenCV+Dlib实现人脸检测与人脸特征关键点识别

    import cv2
    import dlib
    
    # Load the detector
    detector = dlib.get_frontal_face_detector()
    
    # Load the predictor
    predictor = dlib.shape_predictor("shape_predictor_68_face_landmarks.dat")
    
    # read the image
    cap = cv2.VideoCapture(0)
    
    while True:
    _, frame = cap.read()
    # Convert image into grayscale
    gray = cv2.cvtColor(src=frame, code=cv2.COLOR_BGR2GRAY)
    
    # Use detector to find landmarks
    faces = detector(gray)
    
    for face in faces:
        x1 = face.left()  # left point
        y1 = face.top()  # top point
        x2 = face.right()  # right point
        y2 = face.bottom()  # bottom point
    
        # Create landmark object
        landmarks = predictor(image=gray, box=face)
    
        # Loop through all the points
        for n in range(0, 68):
            x = landmarks.part(n).x
            y = landmarks.part(n).y
    
            # Draw a circle
            cv2.circle(img=frame, center=(x, y), radius=3, color=(0, 255, 0), thickness=-1)
    
    # show the image
    cv2.imshow(winname="Face", mat=frame)
    
    # Exit when escape is pressed
    if cv2.waitKey(delay=1) == 27:
        break
    
    # When everything done, release the video capture and video write objects
    cap.release()
    
    # Close all windows
    cv2.destroyAllWindows()