Motion senors settings with DeepStack

[CrazyAsYou]

Just curious what motion setting people are using in conjuction with Deepstack AI?

I am currently using the settings below on a 1080p camera pointing at my driveway/car and pickup lots of movement/triggers to pass to Deepstack, perhaps at little too much at the moment with passing car headlights and reflections off my own car but Deepstack AI is set to "person" only and is filtering those out pretty well. I cannot work out if moving from simple back to edge-vector tracking adds any value when using AI.

 

[wittaj]

Everyone's settings is going to be different based on their field of view and what they are trying to accomplish along with their CPU power.

Really comes down to your setting - is the headlight shine and reflections off of your own car resulting in missed AI triggers? If not, then let it go as AI is working.

If you are missing things, then you set up zones and what not to knock out things like headlights best you can (and I recognize in some instances that isn't possible. I have a few of my cameras that I cannot knock those out, so this works very well for not triggering cars when all I care about is getting human alerts.

I think Edge Vector may be worth giving a try.