3
cb219
4y

Didn't really know how to categorize, bit of a question/discussion/curiosity, so I put it here.🤷

Just today I read an article that stated about the Netherlands, where the police will use an "AI surveillance camera" (yey buzzwords incoming 🙄, but it would actually make sense(?)🤷) to detect and punish drivers, holding a smartphone. Pictures without smartphone shall be deleted. How would this system work without having non-smartphone pictures? It needs to build a classifier, doesn't it? (To be clear, the system only reports those images to an officer for further analysis and actions.)

I mean let's consider that the images are somehow pre-processed, then some convolution(s) for feature extraction, then maybe some more intermediate steps and at the end apply the results on a classifier. How would that classifier work? Would a probability between 0 and 1 suffice? And if so, report those from 0,5 and above? Or would there be better techniques?

Comments
  • 0
    It doesn't need to store any *more* images.
    Presumably they already have tons of traffic footage which they're using to train the detector, my guess would be that they just use a deep net object localisation algorithm like YOLO which also outputs confidence values and is fast enough to run realtime. If the confidence is more than a certain threshold the frame gets sent to the authorities I guess with the location of the phone marked and stuff (because YOLO can also find object positions, not just detect them).
  • 0
    I know the UK was/is looking at a similar sort of idea but that one would listen out for signals from the phone and disregard any which also had Bluetooth too. Not sure how feasible it really is. I guess it would also need a human to verify it was the driver using the phone, not another occupant.
  • 2
    I'm dutch and it's currently in a testing phase on two highways.

    I hate surveillance but phone usage in the car is so bad that I approve of it.

    I'm tagging @linuxxx because of privacy and being dutch
  • 2
    They're training algorithms to detect it when someone is holding a phone to their ear in a car or truck. The cameras for this have been in place for a while as far as I'm aware.

    I think they're actually applying a form of AI and machine learning to this, involving a buzzword doesn't mean that the buzzword is false in this context by default.

    A picture is taken and if the algo detects (possibly) someone using a phone while driving, it is stored and send to a human being to be verified.

    As long as they don't store the images of people not using a phone, I'm fine with it (because fuck people using their phones while driving), otherwise no so much (the government doesn't have a single reason to store those images).

    (@Codex404)
  • 0
    I think that the issuing tickets/punishing people should be done through review only.
  • 1
    Australia already has this in trails.

    it picks you up even when the passenger has their phone out or you have your hands near your face.

    https://roadsafety.transport.nsw.gov.au/...
  • 1
    @C0D4 heard that in china there were already people that were fined because a camera tracked them when they actually just scratched their face.😅
  • 1
    @cb219 that's probably where this one came from. It had a lot of false positives, but then this makes sense for a pilot "AI"
  • 0
    @C0D4 gotta start somewhere 🤷
    would be interesting to see smth similar be even considered in germany, where these things are almost impossible to realize.🙄
Add Comment