Text this: Spatio-temporal framework and algorithms for video-based face recognition