Why does 1 second pause in movement stop zone A>B detection

[Bignose2]

Hi,

I have a camera to detect a car coming into my drive, need A>B so not when leaving.

Works well most of the time but not always, appreciate if too fast of really poor light but quite a few times when not expected to miss, hence more testing.

Video 1, works fine but it drives without pausing.

The 2nd video, there is a half second pause, the yellow box disappears, but the car does move into B, to the same final resting point, in fact quite a bit further, of the detected car. Tried all sorts of make & break times.

Feel I am missing something obvious.
My thinking is it should find the car in A, as it does, even with a brief stop, then pick it up as it moves into B.

If this is the case, a person could easily beat security by stopping for a seconds every few steps.

Shown with G zone clear
detected
https://www.dropbox.com/scl/fi/o98nlvny ... oypq3&dl=0
no detected
https://www.dropbox.com/scl/fi/5ase11kx ... yiiuq&dl=0





Apart from videos above have Zone G as whole/joining screen, nothing in any other zone.
I know the video does NOT show this, I used the blackout to highlight the zones.

It does NOT trigger in exactly the same way when G is solid/enabled.


I know the failing video is a bit darker & bit of rain but do not think it is that. Tried every sensor, movement settings, pixel move distance etc etc.

[MikeBwca]

...

[MikeBwca]

...

[Bignose2]

Hi,

Thanks for your replies,

I had tried increasing sensitivity, reduce movement & all possible settings but never helped.
I do use AI but this is after movement, I need to detect this A>B.

I have read many times over the years inc. Youtube etc. from what seem like experts that strongly advise that the gap & encompassing extra zone is the best way to do it.

BI Help, I am sure you know, does not say its better but should be no problem (sure I read somewhere officially this is the preferred method)
"This means there must be continuous zone coverage through
areas where an object is being tracked. Instead of worrying about zones overlapping or
abutting, it is much easier to just setup one zone to be used as an overall “mask” and then
draw additional zones to be used with object detection and tracking. For example, if zone A
is left as representing the entire image, you may then draw smaller zones B and C anywhere
on the image, and then apply an object tracking rule B>C without consideration of how
those zones are aligned or spaced."

What is really weird now, if I reduce B, basically chop off a large rectangle off the bottom it detects OK & earlier, before the pause in fact.
This should have no bearing as zone B is now half the size and the top half that the car moves into has not been touched.

This Triggers



This does NOT ?



At least I can use at the moment but really would like to know why as difficult to plan zones & other detection if it does not do what I would think.

I will try removing the gap & reduce the size of G but still strange?

thanks again