Dahua TIOC camera does not recognize the IVS rule at night, original support is at a loss...

cam-da

n3wb
Joined
Nov 3, 2023
Messages
17
Reaction score
1
Location
Germany
Hello @wittaj,

I was able to achieve my goal.

The distributor has finally exchanged my camera for one without TIOC.

It looks like the person and vehicle detection is finally working.

I have only seen 2 false alarms so far:

1. once I received the push message even though the vehicle was outside the tripwire (clearly 3m away from the tripwire)

2. once I did not receive a push message at all, although the vehicle had crossed the tripwire by approx. 1m. The line for triggering was reached.
I don't know if it matters if the zones of 2 IVS rules overlap or cross at one or more points.

Do I need to draw the IVS rules so that the lines of 2 rules (intrusion / overlap) do not cross?

Thanks anyway for any tips.

Merry Christmas to all.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,299
Reaction score
49,274
Location
USA
I am glad to hear you finally got a satisfactory resolution!

The occasional false trigger will happen.

For #1, sometimes the camera AI can lag as it processes and you sometimes get that "delay" where the object is shown past the IVS rules. Sometimes based on the field of view, it is simply a limitation, while other times a simple moving of the IVS rules to allow the camera more time to see the object before it crosses can help it react faster once it does pass it.

For #2 - do you know for sure it triggered as in you have a trigger event in the NVR or on the SD card? As you saw from the previous camera, just because an object crosses the line doesn't mean the camera AI "sees" it.

Post a video of that clip and we can probably offer a suggestion.

I personally do not like IVS lines to touch, but many here have used crossing over IVS rules just fine.
 

cam-da

n3wb
Joined
Nov 3, 2023
Messages
17
Reaction score
1
Location
Germany
Hello @wittaj,

I have posted the videos.

In the video1exp you can see the detection of the passing car not touching the line and triggering the alarm. After this error message I moved the line a little bit inwards. So far (after 24 hours) I have not received any false alarms.

In the Video3exp you can see that the vehicle cuts the tripwire, but no alarm is triggered. The lines of the two IVS rules also cross at this point.

The camera is connected to the recorder. I've been paying very close attention over the last few days. For some time now, I have set up the push notifications from the camera and the recorder in parallel on my smartphone. This allows me to compare the differences.
 

Attachments

duplo

Getting comfortable
Joined
May 26, 2022
Messages
301
Reaction score
376
Location
Berlin, Deutschland
I'll assume the basics are setup correctly on the IVS rule....

It certainly looks bright enough.
Its hard to explain but IVS rules, especially Intrusion boxes have a "3D" nature to them. People can literally "walk underneath" the rule line and not trigger the rule.
Have you tried tripwires? I find them more sensitive and I like to use a zig zag effect to increase chances of trigger.

View attachment 176937
never saw someone using overlaping intrusion boxes..
 
Last edited:

bigredfish

Known around here
Joined
Sep 5, 2016
Messages
17,732
Reaction score
49,342
Location
Floriduh
Hello @wittaj,

I have posted the videos.

In the video1exp you can see the detection of the passing car not touching the line and triggering the alarm. After this error message I moved the line a little bit inwards. So far (after 24 hours) I have not received any false alarms.

In the Video3exp you can see that the vehicle cuts the tripwire, but no alarm is triggered. The lines of the two IVS rules also cross at this point.

The camera is connected to the recorder. I've been paying very close attention over the last few days. For some time now, I have set up the push notifications from the camera and the recorder in parallel on my smartphone. This allows me to compare the differences.
IVS rule lines can be tricky. Remember 3D environment.

In the first clip, the yellow truck didnt actually get far enough to cross the intrusion line or tripwire., frankly the nose of the vehicle is "underneath" the line. had he pulled up another 5ft? the taller body of the vehicle likely would have tripped it.

In the clip of the white car on the street, certainly a false alert, but Im thinking it had nothing to do with the car. Wind was actualy shaking the camera, and while it shouldnt have alerted, with all the activity of the landscape and shaking of the camera, I wouldnt be overly surprised by it.


Here's an example.
Not only is space/3D a concern, but so is Time. Time for AI to recognize a human, Time to recognize its crossed the line, and Time to enable tracking.
2 seconds from object detection, to classify as human, to trigger and finally track

Scene with tripwire inside an Intrusion box. Note how lines extend past the target area (crosswalk).
IVS sees a vehicle and human but no alert yet as it should be
picB.jpg

Then 1 second later she actually trips the zone as the lines momentarily blank
picC.jpg

Another second and she has completely crossed the tripwire
picD.jpg

Another second and IVS/AI have confirmed human, put a red bounding box on her and is tracking
picE.jpg

Your lines need to be "lower" in 3D space. I would use similar to above. A zig zag tripwire as "low" as possible inside an Intrusion zone
tripwire.jpg
 
Last edited:

bigredfish

Known around here
Joined
Sep 5, 2016
Messages
17,732
Reaction score
49,342
Location
Floriduh
Here's another example, Intrusion Zone + Tripwire

Notice she is "under" the Intrusion box and in a 2D world has hit the tripwire, but no detection in the 3D space
1.jpg

Now he, who is much taller and closer to the lower portion of the Intrusion box, trips it
2.jpg

Had she been alone and kept going, it would have most certainly picked her up with the"Appears" filter on the Intrusion box., but another 1-2 seconds later
 
Top