Albert-Laszlo Barabasi talked about this sort of fear a few years ago in his boot Bursts[1]. He figured Google would be the one to do it. Even with dropouts occurring when someone goes into a bus, people's movements tend to have low entropy[2]. Since most people's movements follow pretty predictable routines, over time the system can learn to predict most people's positions given noisy data.
It might not work well for criminals trying to avoid being seen by arial cameras during the day, hiding out away from their usual haunts and obscuring their faces. It will probably be a big win for marketing data analytics firms.
It might not work well for criminals trying to avoid being seen by arial cameras during the day, hiding out away from their usual haunts and obscuring their faces. It will probably be a big win for marketing data analytics firms.
[1] http://www.amazon.com/Bursts-Patterns-Everything-mail-Crusad...
[2] http://www.barabasilab.com/pubs/CCNR-ALB_Publications/201002...