S3 animal detection

[rdxny]

I connect smartPss to a Dahua NVR and view the recordings that are on the NVR. I think it will also view from a camera SD card but that needs to be confirmed by somebody who is using it that way. I have the SD cards as a backup for the NVR and have never had the need to go to them, and when I've tested them I used the camera GUI. I don't want to drift the thread too much so will keep this short. After running smartPss and BI in parallel for a few months my opinion is that if smartPss showed trigger thumbnails like BI does, it would be the best pick for me by a large margin. The thumbnails are an elephant-level feature in favor of BI, but you have to give up a few nice things that are in smartPss. Catch-22.

smartPss has a recording capability and it's widely panned as useless, or even worse than useless.

Thanks, was just trying to figure out the usage case for it.

 

[bigredfish]

It can be used as a VMS but I and most others don’t use that component as @tigerwillow1 mentions it sucks.

It’s a great stand alone live viewer, playback, download clips software however. (Minus thumbnails)
I too use it to do the above on recordings or live view via a Dahua NVR

 

[tigerwillow1]

A possible better explanation, at the user level smartPss is Dahua's equivalent of UI3. At the technical level, it's a standalone app as opposed to a web page.

 

[Mibix]

Is it possible to change what it shows in the alerts section for the ONVIF events? I changed their memo/labels but it still shows up with the whole "RuleEngine/MyRuleDetector/PeopleDetect" Instead of just "Person". Even if I set the action for that ONVIF event to "Confirmed Person" it still displays the same. Would just be easier to quickly look at what it identified since I could set different ones for a Person, Vehicle, and Animal. Trying to switch to object detection on my IPC-T58IR-ZE S3 instead of using CodeProject.AI (like my driveway cam is using) and this is my only gripe so far.

 

[wittaj]

You changed the memo to person for the MyRuleDetector, but the alerts are showing it is the FieldDetector ONVIF that is triggering. Change that rule memo to person.

 

[Mibix]

You changed the memo to person for the MyRuleDetector, but the alerts are showing it is the FieldDetector ONVIF that is triggering. Change that rule memo to person.

Derp , ok now I gotta figure out how to separate them, heh

EDIT: not sure this is possible the only ONVIF alerts I ever get are from the RuleEngine/FieldDetector/ObjectsInside or the RuleEngine/CellMotionDetector/Motion :

 

[tigerwillow1]

I doubt that the camera will send the object type over ONVIF. I'm also Sgt. Schultz on this "I know nothing".

 

[wittaj]

It is possible to make this happen. You need to make 3 IVS rules - one for person, one for vehicle, and one for animal.

Then you create a new rule in BI and tell it to listen for each IVS rule.

So you add a rule that says listen for it to contain IVS-1 and then the memo is person. Thn the same for IVS-2 and memo is vehicle and IVS-3 and memo is animal.

Then it will display the appropriate object in the alert.

 

[105437]

It is possible to make this happen. You need to make 3 IVS rules - one for person, one for vehicle, and one for animal.

Then you create a new rule in BI and tell it to listen for each IVS rule.

So you add a rule that says listen for it to contain IVS-1 and then the memo is person. Thn the same for IVS-2 and memo is vehicle and IVS-3 and memo is animal.

Then it will display the appropriate object in the alert.

This! Works perfectly for my use case.

 

[tigerwillow1]

Does anybody know how the IVS individual rule and global sensitivity settings interact? Could it be simple, as in the sum of an individual rule setting plus the global setting is the sensitivity for each rule, or do the two settings adjust something different?

 

[rdxny]

It is possible to make this happen. You need to make 3 IVS rules - one for person, one for vehicle, and one for animal.

Then you create a new rule in BI and tell it to listen for each IVS rule.

So you add a rule that says listen for it to contain IVS-1 and then the memo is person. Thn the same for IVS-2 and memo is vehicle and IVS-3 and memo is animal.

Then it will display the appropriate object in the alert.

Can you show how to make the BI rule in this case? (all entries in the new rule) I can't seem to get it working (listening for "IVS-3" for example). Thanks

 

[Mibix]

Can you show how to make the BI rule in this case? (all entries in the new rule) I can't seem to get it working (listening for "IVS-3" for example). Thanks

I think I got mine working with this setup in BI:

and this is the setup in the camera AI:

Showing up like this now (I only have it configured for my porch camera):

 

[wittaj]

^yes that works as well if you rename each IVS rule to the object and then add the contains option.

 

[Mibix]

This might be off topic but has anyone gotten the EPTZ thing to work with this? I'm using the PC-T58IR-ZE S3 and it should be following and zooming in on the things it identifies and showing them at the bottom?

 

[bigredfish]

You described my "blue" better as teal. I don't get any vehicles in my FOV but I'll assume yellow=vehicle. That leaves me asking if animals have their own color, or if anything but vehicles and people all get green?

FWIW on Dahua cameras with Dahua NVR's and the new FW

• I'm getting reasonably good animal detection
• I get Green bounding boxes for critters. Teal for humans, Yellow for vehicles
• I can change the names of IVS rules on the camera and If I wanted to I could do the same as being discussed on BI by setting separate rules and naming them separately. IE Human rule, critter rule..etc

View attachment Recording 2024-07-25 121607.mp4

 

[rdxny]

I think I got mine working with this setup in BI:

Thanks, that worked

 

[Mibix]

This might be off topic but has anyone gotten the EPTZ thing to work with this? I'm using the PC-T58IR-ZE S3 and it should be following and zooming in on the things it identifies and showing them at the bottom?

It looks like it was just a problem with the browser plugin and it works in the web interface but it would just never send it over the RTSP stream to BI because I guess you could zoom in yourself?