Using MV32 object tracking analytics to approximate distance between objects

[DR1]

Hello,

We're interesting in determining if we could use data via the API from MV32 cameras in a larger square rooms (with a high ceiling) to calculate and approximate distance between objects in more or less real time.

The use case is auditing/assessing if social distancing guidance is being respected in a large gallery/museum space.

I seem to recall stumbling across someone attempting this but cant find the post.

Maybe theres already a 3rd part solution we could use?

Thanks for any guidance that can be offered.

[PhilipDAth · Accepted solution]

Unless you like a lot of maths you wouldn't want to do it with an MV32 because of the fish eye lense.  Everything straight line is a curve.

You would want to look at MV12s or MV22/MV22-X.  The MV22/MV22-X might be needed to get an optical zoom if the camera is going to be high up.

The easiest option would be to instead use people counting via MQTT.  If the room exceeds "x" number of people then generate an alert.

Otherwise you are going to need to use the same people detection via MQTT, but calculate the distance yourself and calibrate those distance measurements.

With regard to people counting and you are able to "cut some code", you could base it off something like my ICU code, and just change it to only trigger when a specific count is exceeded.

https://www.ifm.net.nz/cookbooks/meraki-icu.html