Need help tweaking motion detection on a thermal camera in a tree...

erkme73

BIT Beta Team
Nov 9, 2014
1,620
1,538
AI sucks on thermal video. I don't know how to train models, so I'm stuck using the standard motion detection as a trigger.

My camera is a tree (about 20' up), and as such, subject to swaying in the wind. Consequently, as it moves, the motion triggers with every gust of wind as the entire field changes.

3.55t 2025-10-05 09.21.25.867 PM.jpg

I suspect it has something to do with this option:

1759719087618.png

But reading the instructions is clear as mud to me.

1759719141579.png

About the only thing that I get is that the lower value makes it more sensitive - and even that is a bit confusing. Does it make the reset more sensitive or the motion detection? I assume a lower value will mean it takes less (smaller) motion to reset the MD... but what's a good threshold for it to ignore swaying? I'm trying 30% but would rather not have to do trial an error on this camera for fear of missing opportunities to capture a hog/yote. Animals wouldn't take up more than 5% of the total screen...
 
No, that setting is for something like a bug goes across the screen or headlight shine that takes over X% of the field of view. It "resets" the sensor so that you don't get an alert for what is likely a false trigger.

Your make time is pretty small and likely the problem.

How long would it take for something to come into the field of view and get halfway across - that is the time to think about. I bet you could change it to 1 second and likely still trigger for what you want, and eliminate the false you are getting.
 
  • Like
Reactions: erkme73
Interesting. Sometimes they move very slowly so I'd be concerned it won't pick up enough movement to trigger, but I will try that as well in a bit.

My thought with the reset is that when the image sways, I have easily 30% or more of the pixels changing (much like a bug/headlights) for which I don't want an alert. I've turned it on and set it to 30%, and so far no more alerts.... Of course it could be that the wind just stopped blowing (Murphy at work). I'll change one variable at a time to figure out which one is giving me my best resolution. Thanks for the suggestion...
 
You can use IVS with the thermal camera. It's "dumb" IVS, no human/vehicle filters.
 
  • Like
Reactions: erkme73