Estimating 3D facial pose in video with just three points | IEEE Conference Publication | IEEE Xplore