Here, I introduce another library named "Dlib", which is a computer vision library always cooped with opencv.
So to run the demo, we need to install Dlib on our system. 
1. I found tutorials to install dlib, and it worked for my device (Win10). 

	https://www.learnopencv.com/install-opencv-3-and-dlib-on-windows-python-only/ (I have tried and it did work well)
	https://www.pyimagesearch.com/2017/05/01/install-dlib-raspberry-pi/ (I haven't get a chance to test on my Pi)

	Note that to install on windows, make sure you have CMAKE and Visual Studio 2017 installed.


2. How to use:
	a. Add custom face dataset
		1. Open "Facial_Recognition_Registration.py".
		2. If using the laptop camera, make sure "cap = cv2.VideoCapture(0)" (at line 17);
		   If using the external WebCam, make sure "cap = cv2.VideoCapture(1)" (at line 17).
		3. Run "Facial_Recognition_Registration.py"
		4. Enter the label as your name.
		
		Your face dataset:
		1. Folder "/Facial_images" -> "/face_rec" -> "/train", then you can see the folder of your name is in it.
	
	b. Run Facial_Recognition_Enrollment.py	 
	
	c. Test on videostream
		1. In "Facial_Recognition_Inference.py".
		2. Make sure line 27 to match your imaging device, same as above a.2
		3. Run


3. Requirements for face registration:
	a. User can sometimes turn your head a little bit to let us get more face data, but you must make sure that there're 5 colles on each photo.
	b. Don't register you face when too much light is coming.

welcome any try-out and comments!