Do smart phones have the performance to handle face (or character or any object) recognition algorithms? Or should this sort of application be handled by a server/cloud?