Suggestions for eliminating these false triggers

[stebbo]

Generally caused by clouds changing the sunlight. I've already set "cancel shadows" and tried to require motion but still getting a good 10 or so of these a day.

I'm thinking the next thing to try is increasing contrast up a lot and making a new profile for night vision as I'm currently missing triggers with people at night, but any other suggestions on handling this?

 

[Dasstrum]

I suggest setting up 2 more zones: Zone B and Zone C. You can have zone B covering the top left portion of the driveway and Zone C covering the bottom portion. Make sure these two zones (b and c) do not overlap. Then in object detection create a rule "B-C, C-B"

This means it will only trigger when an object crosses from zone b to c or visa versa.

I actually will be making a video on this very subject this week and will have it released soon. You can follow me on this thread: Blue Iris Video Tutorials

 

[Dasstrum]

Also a object travel distance of 50 pixels is a VERY short distance in relation to the camera image. Depending on your cam resolution... if that camera is a 1080p with res of 1920x1080... 50 pixels is only going to roughly be the following distance...

 

[Dasstrum]

One last thing to try it the "reset detector when" setting. Try changing this to a lower number like 30%. If you think about it a person walking down the driveway will probably only account for 10-15% change in the picture. Having a 30% reset might greatly reduce the false triggers.... Also a 2 second make time in my opinion is very high. You will miss important alerts and events if this is set too high or you will miss things leading up to the trigger:

 

[stebbo]

Great - I'll watch that video as I think I just left 'reset detector' to its default, not knowing what it does.

I was hoping a 5 second pre-record would cater for the missing things up to the trigger due to the make time, but I could have been desperate trying to eliminate these unwanted triggers. I'll get more familiar with everything.