3

Is there a practical way to predict the crowd density of a place in real-time?
I was thinking of some way to scrape social media activities and using the geolocation tags to predict the crowd in that particular area?
But I am looking for a more accurate alternative!
Please help!!
All ideas are welcome

Comments
  • 2
    Tracking the amount of devices (e.g. WiFi / Bluetooth mac address). Be aware this probably not privacy friendly and might be illegal as individuals could be tracked, too.

    Of course this only helps you if you want to establish it or is already present.
  • 0
    Real-time human / Object detection via CCTV
    1.Considering the fact that how many persons have passed
    2. By taking a time you can do more accurate prediction that from time x to time y it's more crowded.
    3. Along with human detection count number of vehicles passed, coz it help to detect the traffic.
  • 0
    Yeah, the best way would probably be to collect the count of devices sending out WiFi probes. Normalize the data and you got a pretty accurate value.
    I don't know how privacy friendly that is. You'd probably need to put a disclaimer somewhere.
  • 0
    @PrivateGER What if I am using a device which doesn't have any Wi-Fi or Bluetooth connectivity.
  • 5
    Your local police force may have a way. At least if you're in China.
  • 0
    Why?

    Predicting human behaviour is either futile or trivial.

    If you are looking into real time estimations (not predictions) and if you are considering larger „places”, you need to contact mobile network providers.
  • 6
    This is very close to my PhD research topic. What I would say is that if you choose to go down the route of either WiFi, cell or social media data then this is simply a proxy for your density, and a biased one at that - these figures tend to provide coverage of younger demographics but often don't cover the elderly.
    Beyond this, these data sources may not provide you with a complete map of the density, and only give you data for certain geolocations - do you want to infer information about the points in-between?
  • 0
    The game "Ingress" does this to calculate the amount of energy you can collect in different areas. Backed by Google location data, of course.
  • 0
    Theres a simpler way of doing this.

    Get the average color of the location when the least number of people are present, preferably when no one is present. Basically you're taking a picture of the static background.

    Get the average that it deviates from for a single person.

    Now, when you want to determine density, grab a frame, and look at all the pixels, and determine how far each varies from the baseline and calculate how many people are present based on that.

    Well, thats the gist anyway.

    Firecodes based on "maximum occupancy per square foot" might also be a useful line of inquiry.
  • 1
    I can think about 3 thermal camera with image proccessing. Then u count the red continuous spots

    Maybe you can put 2 senser devices at entrances and detect in wich order they get fired, and know it it is a +1 or -1 (entering room or getting out)
    In the start and end of the day it must be zero, so u have 2 error correcting numbers.
  • 0
    Might have issues on cold days just so you know.

    Sure theres a work around though.

    Sounds like a super fun project.

    I wish I had a hackerspace (god I hate that term), in my area that didn't cost a boatload of money for membership.
  • 1
    I found exactly what I was looking for,
    https://github.com/m-wrzr/....
  • 0
    @sbiewald how do you get the access to other's Bluetooth /wifi residing far away from you?
    What I wanted was --
    "search for any place on the internet and see how busy that place is at a particular point in time."
    But i found a pre existing modified google places api which does the exact same job!
    Check it out for reference https://apify.com/drobnikj/...
Add Comment