├── dataSet
│ ├── Badis.5.1.jpg
|.. ├── ....
├── models
│ └── trained_from_sqliteDB.yml
├── FaceDataBase.db
├── faceDataBaseGenerator.py
├── faceDataBase_Detector.py
├── faceDataBase_trainer.py
├── haarcascade_eye.xml
└── haarcascade_frontalface_default.xml
The faceDataBaseGenerator
receive in inputs the Id
and the name
of new detected face from a webcam. And take 20-30 images from this face. It should be ONLY one face in front of the webcam at the time, to avoid confusing between Ids
and faces. The script same images to dataSet
folder and corresponding features like Ids
and names
to sqlite database.
The faceDataBase_trainer
train tge model usig recognizer = cv2.face.LBPHFaceRecognizer_create()
and save the model to models
folder.
The faceDataBase_Detector
gets faces profiles from database and compaire them to live webcam detected faces. If confidence is greater that a threshold (0.7), the script associate detected face to selected profile.