https://123dok.net/document/yevxnm7z-a-multi-cue-spatio-temporal-framework-for-automatic-frontal-face-clustering-in-video-sequences.html