DLib
a machine learning library for image processing using the C++ providing C/C++, Python, and Java interface. Dlib is able to detect face features and provide a map of landmark points that surround each feature
installation and usage
#pipenv install opencv-python, dlib
#following is an example for video real time face feature detection
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()
