T-top face landmark detection #119
Description
In T-top, the face descriptor extractor refers to OpenFace (https://github.com/cmusatyalab/openface/tree/master), which uses dlib/shape_predictor_68_face_landmarks.dat for facial landmark detection.
Do you also use dlib/shape_predictor_68_face_landmarks.dat? If not, is there a specific reason why?
We are now working on emotion detection and gaze tracking, both of which require facial landmarks as the first step. Can I reuse or modify the output from the face descriptor extractor for this purpose?