How to deal with large light changes?

[Blazer]

I wish to monitor an area that has a motion-activated flood light (at night), which lights up most of the area that is in the FOV of the camera.

Can anyone suggest the best settings for ZM so that it will not trigger when the light goes on or off, yet will still trigger if a large blob/person moves? The camera itself also has infared so should be able to detect motion even if the flood light does not activate.

[cordel]

You will have to try to tweak your max alarm settings but as I'm sure your aware, the analysis can not differentiate between what is a person moving and lighting changes. Only the percentage of the diff alarm settings. The only true way around it is to have a PIR trigger ZM.

[lazyleopard]

It ain't easy... It's possible to do some filtering by having small preclusive zones which will only be triggered when the light switches, but you do run the risk of hiding real events. The main problem is that the reference image needs to be altered rapidly, and often ZM will trigger not on the initial switching, but a few cycles later when the differences between reference and actual have subsided from the initial wipe-out.

[zoneminder]

This is an issue that I think needs to be approached in a new and slightly different way. I am thinking of adding something like a suppresion zone which would be like a preclusive one except it could overlap existing zones (in other words be as big as the whole image) and if triggered (by a lighting change) would suppress motion detection for a determined period/number of frames. Something along those lines anyway. Or just add in that if a zone has maxima set then any changes that exceed them will result in that zone being suppressed for a period.

[MJN]

In this particular case, given that the light is motion-activated, would it not be advantageous to allow the light-induced trigger given there is motion and hence it presumably warrants recording?

Mathew

[lazyleopard]

The frames immediately after a switch event tend to have zones at maximum score because everything's changed and the current reference is no use any more. Some way to change the rate at which the reference image is adjusted might be good. Say the detecting zone triggers, so the reference percentage runs at (say) 30% instead of 3% until the zone goes quiet?

Any other ideas how to get from swamped to useful detection reference as quickly as possible?

[nelsda]

I have had great luck creating a fast acting preclusive zone which is either where the light is at or at a small spot that lights up quickly. I only occasionally get an event trigger when the light trips because of wind or something like it and not because a person walked through.

[ml35]

Hi, it's me the dead thread digger; 13 years later, has any work progressed on this?
Running 1.34.22-bionic1 from iconnor ppa

[MJN]

What's the problem you're trying to solve? Is it an issue for you having these false positives?

[ml35]

What's the problem you're trying to solve? Is it an issue for you having these false positives?

yes, here is a sample

[ml35]

?!

[Magic919]

Interrobang.