When the author talked about scanning the video for "interesting" parts he mentioned pattern recognition. Does anyone know if data compression is used for this

When the author talked about scanning the video for "interesting" parts he mentioned pattern recognition. Does anyone know if data compression is used for this in applications? The idea being that if you have 12 hours of cctv footage that is mostly just the same view of a warehouse most of that will compress very well but not the 15 or 20 minutes where there were people walking around.