Eye Aspect Ratio (EAR) and Mouth Aspect Ratio(MAR) detection using Python and Facial landmark predictors.
Eye Aspect Ratio (EAR) and Mouth Aspect Ratio(MAR) can be used to identify-
1.Facial expression of a person classified based on observed EAR and MAR values.
2.Drowsiness detection while dirving.
3.Behaviour of driver i.e attentive while driving using MAR ratio
Pre-trained face-shape decriptors which are trained using DNN and used using dlib library in OpenCV are used to identify eyes and mouth through descriptors generated.
Reference for EAR- http://vision.fe.uni-lj.si/cvww2016/proceedings/papers/05.pdf
Dlib's Facial Descriptor -https://github.com/pyannote/pyannote-data/blob/master/dlib.face.landmarks.dat