• ironhydroxide@sh.itjust.works
    link
    fedilink
    English
    arrow-up
    3
    ·
    2 hours ago

    Yeah this is something a well trained neural network would make simple though. So long as you have the processing power, and enough storage.

    I’d imagine you’d be able to purchase some identifying information from data brokers and eventually link ids to people or families.

    • toast@retrolemmy.com
      link
      fedilink
      English
      arrow-up
      1
      ·
      2 hours ago

      Yes and no, I think. It isn’t really huge amounts of data, and the patterns aren’t super complex, so a neural network would pick up a lot. But, given how far the signals travel, the intermittent nature of the signals, and how little they can initially be associated with a particular vehicle, I think in most environments associating a particular set of signals with a particular car would require some human field work. Sure, there are circumstances where automated pairing would be trivial (like at a toll booth), but catching signals in the wild and processing by neural net alone might be ok for analyzing traffic patterns while not being enough for surveillance.