Detecting wildlife with Dahua cameras

....it's pretty impressive it detected the fox!
Of the different camera models I have I've found the 42xx and 52xx series to be the most sensitive at picking up tripwire crossings. Sometimes I have to play a clip a few times before I can find what caused the trigger. I'm experiencing the downside of that tonight with a spider web hanging in front of one of the cameras and it has been continuously triggering for a few hours, just as bad as never triggering. Even thought the 2231 is a 2 MP starlight, its IVS does not trigger as easily as the older 2 MP starlights. It's still mostly ok with picking up animals. I assume they made it less sensitive in the name of reducing false triggers.
 
Of the different camera models I have I've found the 42xx and 52xx series to be the most sensitive at picking up tripwire crossings.
Just to clarify this Z4E is 54XX series? (in case there are a couple different models, but it's a IPC-HFW5442E-Z4E. The Z12E is a 52XX). But yeah, still impressively sensitive.
 
Now I'm confused! Are you picking up animals on a 54xx varifocal? If so, which firmware?
 
Yeah, this is detecting animals on technically, I guess, a 5442 varifocal (IPC-HFW5442E-Z4E) but the FW doesn't work on the 5442 varifocal turret... It looks like it works on 5442 varifocal bullet, dome, the 5442-Z4E (bullet) and the (2MP) 5231-Z12E (bullet)

It's this firmware: AI 4.0 latest new firmware DH_IPC-HX5XXX-Volt_MultiLang_PN_Stream3_V2.800.0000000.0.R.190521.zip

Even before flashing it I was hopeful it would work since it's dated earlier than the 8.R release date (8.R.190902.bin).
 
Last edited:
On a side note, using the Dahua IPC EASY unbricking / recovery over TFTP method, I attempted flashing the DH_IPC-HX5XXX-Volt_MultiLang_NP_Stream3_V2.800.0000000.10.R.191118 FW you mention in the OP so many times this weekend. While the T5442T-ZE (turret) is not officially supported the rest of the 5442 ZE models (varifocal bullet+dome) are, I'm assuming they were released before the turret model.

I can successfully flash 7 of 9 .img files in the 10.R firmware and get the camera to boot, including two that sound important for what we're looking to do (web-x.squashfs.img and firmware-x.squashfs.img). But the camera fails to boot once I flash the remaining two .imgs (romfs-x.squashgs.img and pd-x.squashfs.img). If I swap out those two .imgs for the ones in the first officially support FW revision (12.R, DH_IPC-HX5XXX-Volt_MultiLang_NP_Stream3_V2.800.0000000.12.R.200319, but doesn't detect animals), the camera boots again. But the web UI still says the system version is 12.R.. :(

The firmware-x.squashfs.img must include the "Algorithm vX.XX" listed in web UI version information because flashing that 10.R one with the rest of the files in the even later 16.R fw, the camera booted but threw an "Algorithm version mismatch, please reflash the firmware" error. The ONVIF version must included in the romfs-x.squashgs.img, because I couldn't get that downgraded no matter what I tried since the 10.R romfs-x.squashgs.img wouldn't boot on the 5442 turret.

Also kind of weird, the 12.R that hasn't been successful detecting animals, was seemingly decent at detecting my parent's cat on the camera I tested in their living room lol. Obviously, very optimal conditions...Could have just been regular motion detection, but I had the sensitivity set way low, the threshold high, and the "area" deleted.
 
Last edited:
  • Like
Reactions: JDreaming
Thanks. I'm not surprised I was confused thinking the version vs. camera situation was simpler than it really is.
 
I feel like I have better results with tripwires. It's a more intuitive thing after trying both, nothing that I can defend with hard data. The animals usually don't behave the way I want them to. Some days there is nothing, other days it's like a traveling zoo. A couple of nights ago there were a few coyotes, raccoons, feral cats, and a badger. The night before, a bobcat. Occasionally foxes, porcupines, and possums. One time a scorpion walking by triggered the IVS, but no triggers ever with snakes, maybe because they're going by too slowly. All the squirrels disappeared a couple of weeks ago when it went from summer to winter in about 24 hours flat. When desperate, I use a little RC car to test if the IVS is working.

Sounds like my place. Using the RC Car for testing is a great idea... I will give that a try.
 
Can add the T2431T-AS-S2 to the list of cams with IVS that doesn't trigger for wildlife with no target filter selected. I was hopeful but not surprised since it's a newer cam. Has missed many foxes, a few raccoons, and a huge deer that lingered there the other day that I thought for sure would have triggered it. Gets me every time though.
 
Can add the T2431T-AS-S2 to the list of cams with IVS that doesn't trigger for wildlife with no target filter selected. I was hopeful but not surprised since it's a newer cam. Has missed many foxes, a few raccoons, and a huge deer that lingered there the other day that I thought for sure would have triggered it. Gets me every time though.
Yep ->
In 2023 we will update some old models to a newer version. I will keep update here.

1. IPC-T2431T-AS(1/3CMOS) update to IPC-T2431T-AS S2 version, 1/2.9CMOS, S2 version will support the SMD( accurate detection of vehicle and human ). Currently stocks on Amazon already use the latest new stocks.
2. IPC-LPR237B-IR update to 4MP IPC-LPR437B-IR, this will be updated at Later Jan/Feb. Price will keep same as 2mp we sell.
Seems al new models are moving to vehicle/human detection only. A real shame Dahua is actively removing functionality.
 
IPC-T2231T-ZS right now still can accept animals trigger, you don't know how much complaint we get when false alarms happen, this bring lots of troubles for users too, so not easy to make everyone happy, maybe later we do 1 model for animals watching only. AI algorithm will focus more on human and vehcles.
 
IPC-T2231T-ZS right now still can accept animals trigger, you don't know how much complaint we get when false alarms happen, this bring lots of troubles for users too, so not easy to make everyone happy, maybe later we do 1 model for animals watching only. AI algorithm will focus more on human and vehcles.
Oh don't get me wrong; I like the fact there's build-in human/vehicle detection. For a lot of situations that's hugely beneficial like you said; brings down false alarms.
But it's just sad Dahua then removes functionality like basic IVS detection; I mean; there's even a checkbox to check/uncheck "human" and "motor vehicle" under "Object filter" -> "Effective object"; indicating that when you uncheck those boxes; all movement detected by IVS should trigger...but that doesn't work...
 
Last edited:
The biggest problem with the 2231 is that once you've used a 5442, you don't want the 2231. Compared to the 5442 you lose image quality and audio. Compared to the 5231 you get the same image but lose audio and a couple other things you may or may not care about. The 2231 looks even worse when making the comparison to a 4k-x/t. It bums me that I have a couple of superior-image varifocals sitting in the closet because they're not capable of detecting animals walking right in front of them.
 
  • Like
Reactions: Mike A.
IPC-T2231T-ZS right now still can accept animals trigger, you don't know how much complaint we get when false alarms happen, this bring lots of troubles for users too, so not easy to make everyone happy, maybe later we do 1 model for animals watching only. AI algorithm will focus more on human and vehcles.
The problem is that having the checkboxes for human and vehicle detection raises the expectation that if they are unchecked, the AI will function in a manner that allows for other (ie: animal) movement detection. If they don't make any difference (which seems to be the case), they should be removed, period.
Personally, I'd like the checkboxes to stay, but only if they are genuinely functional.
 
Oh don't get me wrong; I like the fact there's build-in human/vehicle detection. For a lot of situations that's hugely beneficial like you said; brings down false alarms.

Yes, that part works great. Where I want human/vehicle detection, the IVS has functioned near flawlessly for me. No misses that I've ever detected and very few falses. But in some cases I don't want only that. I can understand why the AI is focused that way though. That's probably what +90% of their customers want.
 
The problem is that having the checkboxes for human and vehicle detection raises the expectation that if they are unchecked, the AI will function in a manner that allows for other (ie: animal) movement detection. If they don't make any difference (which seems to be the case), they should be removed, period.
Personally, I'd like the checkboxes to stay, but only if they are genuinely functional.

Exactly. Or add animals to the model and a pet/animal box. Which would be even better. The little Wyze cams that I have just suck in most respects but they do reliably trigger for pets/animals with that selected. If Wyze can do it, Dahua surely can. But again doubt that's high on the priorities list given their market.
 
I think I mentioned earlier but it's still odd to me how often IVS triggers on small dark birds in my yard. Not sure what it is about them that cause it to work. Just your usual starlings or similar that sort of waddle across the lawn feeding. It's still nowhere near 100%. More like maybe 10% at best. But it does trigger while fox/raccoons/cats/deer can walk by perfectly for the rules and pretty much 100% nothing.