This thread got me to thinking, what do you BI guys consider a False Alert?
For me using an NVR and Dahua late model cameras:
1- Using old fashioned
legacy MD, there are NO FALSE ALERTS, ZERO .... I have
told the camera/NVR to alert me on
EVERYTHING that moves. Thats what I get.
I think but havent tested, that IVS with NO filters would do the same?
2- Using AI/IVS, I filter for say Humans and Vehicles only.
ANYTHING NOT a Human or Vehicle that produces an alert is essentially a
False Alert
I use this method on 95% of my cameras and I truly am surprised when and if I EVER get a false alert
Bugs, Rain, Tree branch shadows, Sun, Lightning, Refrigerators, etc... I just dont get alerts from those. MAYBE 1-2 per MONTH from 25 cameras I watch each day
So how do you validate a "False Alert" with BI?
For me with an NVR, Two ways but essentially the same
1- I get an Alert in the App or Email or both (on some cameras I do set for both), with a snapshot.
The snapshot, in theory, shows me the culprit that caused the alert, right?
So here's a sample clip:
Lots of IVS (Human and Vehicle) on this camera today. Lets say 300
Using the NVR AI search filtering IVS events, I can visually see each thumb and see the culprit in the thumb that set off the alert.
But the last example alert I show, the snapshot doesnt show anything?
False Alert? Nope
...Just a split second too slow to snap the photo on a far away IVS rule
View attachment Recording 2024-11-01 204747.mp4
And Using SmartPSS instead, I also see an Alert hash mark, so by playing that same clip, I see it was indeed a VALID alert.
The same holds true on the DMSS App and the NVR regular search timeline.
View attachment Screen Recording 2024-11-01 211713.mp4