answered 2018-12-06 08:25:27 -0600
Here is the 2014 CVPR paper it is based on "Face alignment at 3000 fps via regressing local binary features"
