This mainly uses this API for the face detection and recognition, which is itself based on Dlib with a trained model.
Faces are saved into a local model, used to recognize persons in photos.
There are the following scripts:
model.py: used to create, visualize, add or remove the model.photo_face_reco.py: used to detect and recognize faces in a given photo.video_face_reco.py: used to detect and recognize faces on the webcam.video_face_reco_rpi.py: same thanvideo_face_reco.py, but for raspberry pi.
For first use, run python model.py add <path_photo> to initialize the model with a face. When adding a photo to the model, make sure there is only one face, clearly displayed, e.g. no sunglasses, no profile, etc.
By default, it creates a model.pkl file.
Once you have a model with at least one face, you can run python photo_face_reco.py <path_photo>. It will try to recognize faces from the given photo thanks to the ones saved in the model. The result is saved in a file new.jpg.