MorphUs app performs automatic face morphing using facial detection and OpenGL shaders for accelerated playback. It also exports to video files and apple watch as a series of still images.
It was previously closed source but I've decided to open source it because it uses a face landmark detection model which is not free for comercial use.
http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2
Previously it used the Face++ API which is no longer free and required a http call to their server. Note that this version requires that you uninstall the previous version because Face++ provided 84 landmarks while the dlib version has 63 facial landmarks.
In terms of the algorithms for morphing it uses the thin plate spline method, thanks to slides from a course I took at the University of East Anglia provide by my supervisor at the time Dr Rudy Lapeer.
I used a matrix library I had written previously in c++ with algorithms from the book numerical recipies in c but ported to c++ using templates for types. But it includes functions for calculating inverse matrices using LU decomposition.
The remainder is in objective C, sorry it's not swift - it is a few years old now.