Awesome, I hadn’t considered the 3241. Good news is that it’s also cheaper than the 5442.
Does anyone happen to know the side by side differences between the 3241 and 5442, other than MP?
Who uses Dahua AI capable cameras? Reliable AI for triggering events? Pro's/con's?
- Thread starter Holbs
- Start date
Awesome, I hadn’t considered the 3241. Good news is that it’s also cheaper than the 5442.
Does anyone happen to know the side by side differences between the 3241 and 5442, other than MP?
sebastiantombs
Known around here
The biggest one is resolution and sensor size. I have a few 5442s as well and think the 5442 is better in low light conditions. The 5442 seems to be able to run color at lower light levels and at night the added resolution makes the video much sharper to my tired old eyes. I do plan on replacing that 3241 that can stay in color at night with a 5442, depending on budgetary constraints, and will be sure to get some screen grabs and video exports of both if/when it finally happens.
looney2ns
IPCT Contributor
The 5442's are really hard to beat for nighttime. They stomp the 5231's in my experience. Huge improvement on the ID'ing prospect
I wonder a comparison between 2MP vs 4MP B&W pix at night with all the varying contrast and white balance, etc. Just seems crisper pic with 4MP (of course) than 2MP, especially during motion. I gotta swap out my 5231's at 6' height by both my driveway carriage lights to 5442's 6mm. I think I'll do some test comparisons.
How are you setting up these cameras for AI detection? I know some examples showed using IVS tripwires. I've been testing out different modes and rules and have been less than happy with anything other than tripwires. My goal is to be notified whenever a person enters an area. I started with the basic "Smart Motion Detection" and it had so many false alarms I had to quickly turn it off. It triggered when there was nothing obvious moving in the scene, much less a person. I tried experimenting with the motion detection settings to raise the threshold and that reduced them somewhat. But it seems as though the AI function is not working at all. Every time a bug flew near the camera it triggered for example. It was FAR less accurate than DeepStack had been.
I moved on to IVS using intrusion detection. I drew the area out and set it to trigger on appears or crosses and used the enters option. See attached picture. It works fine when a person enters from the top (note the arrow) but does not when from the sides or bottom. If only one side triggers the detection, how is that different than just using tripwires? So far tripwires seem to be working the way I want but need a few more days of testing to be sure. But that is one of the more time consuming methods to set up vs either the "smart motion" or IVS intrusion.
What am I missing, and are there any tips anyone can share?
I moved on to IVS using intrusion detection. I drew the area out and set it to trigger on appears or crosses and used the enters option. See attached picture. It works fine when a person enters from the top (note the arrow) but does not when from the sides or bottom. If only one side triggers the detection, how is that different than just using tripwires? So far tripwires seem to be working the way I want but need a few more days of testing to be sure. But that is one of the more time consuming methods to set up vs either the "smart motion" or IVS intrusion.
What am I missing, and are there any tips anyone can share?
sebastiantombs
Known around here
The reason it works from the top but doesn't from the sides or the bottom is because of the space at the top and not at the sides or the bottom. Re-draw the box with some space, like at the top, on the sides and the bottom. Then re-test and it will probably work quite well.
Follow up: Using "Appears" and unchecking "Cross" action seems to work somewhat. I'll try out your suggestion on moving the sides when I get a chance.
Can you elaborate on how you're setting it up? I don't mind a few false hits. But triggering every minute or two because a moth buzzed by (or even when the outside lights come on) makes it less useful.
looney2ns
IPCT Contributor
What firmware version is it running?
Be sure to also setup the Global Setup: IVS/Global Setup - Dahua Wiki
Make Minimum size 0,0 and let AI do it's thing. You only put in minimum size AFTER you have seen that a dog for example is tripping AI. None of mine have a minimum size and I get zero false triggers. When you set a minimum size, in some instances I have seen that this takes priority over the AI and then you get moth triggers. YMMV.
Mike A.
Known around here
- May 6, 2017
- 4,141
- 6,945
Event > Video Detection = Motion Detection OFF
Event > Smart Motion Detection = Enabled NOT checked, Human and Motor Vehicle checked. (Not sure this one matters much, think maybe overridden by IVS settings below)
Event > Smart Plan = Light Bulb selected.
Event > IVS = Action, Appears and Cross checked. Enters and Exits checked, Object filter checked with Human and Motor Vehicle checked. Tripwires and an intrusion box created and selected above.
Rest left at defaults.
I'd guess that maybe you still have video motion detection checked?
Edit to add: And now mine F's up and detects a trash truck going by outside of any detection zone. I said something good about it... so It had to screw up. lol
Edit to my edit: Ahhh... Somehow video motion detection got checked when I went into mine. That was the problem. So, again, double check that you don't have that selected.
Last edited:
No, I did turn off video motion detection and smart motion when I enabled the IVS setup.
Thanks for the IVS tips. I'll keep adjusting it. Has anyone gotten good results with the "Smart Motion" though?
The firmware installed is V2.820.15OG001.0.R
Thanks for the IVS tips. I'll keep adjusting it. Has anyone gotten good results with the "Smart Motion" though?
The firmware installed is V2.820.15OG001.0.R
looney2ns
IPCT Contributor
Maybe I don't understand how to set it up, but it seemed to me that the AI function of Smart Motion was not functioning. Maybe it's broken? I'd imagine that the AI algorithm is the same as for IVS, just a difference in the triggering. But it should not be dramatically different than IVS Intrusion in theory.
Firstly, I am new to Dahua. I purchased a couple of their 5442-based cameras due to input received on this forum. In Blue Iris, I can see how to 'tune' the Triggers/Alerts by using the Test Motion Detector and Test With DeepStack features. What are the settings to capture IVS triggering information so that I can see exactly when IVS is triggering and what it's using to trigger?
Reason I ask is, at the far range of the camera's view, I can tune DeepStack to trigger when a person first steps into view. So far IVS seems to require the person to be considerably closer to the camera. The difference being almost 10 seconds. I want to tune IVS to be able to trigger more quickly.
Reason I ask is, at the far range of the camera's view, I can tune DeepStack to trigger when a person first steps into view. So far IVS seems to require the person to be considerably closer to the camera. The difference being almost 10 seconds. I want to tune IVS to be able to trigger more quickly.
IVS tuning is done within the camera itself, not BI.
You either have to watch live view and turn on the IVS rules and have someone walk around or have an SD card in the camera and playback events with the IVS rules turned on to see when they are triggering.
Field of view and how you set up the IVS rules impacts the success or failure of the IVS rule and triggering.
So first steps are have you done the global config to calibrate the camera for your field of view and do you have min object size as 0,0?
Also, many have noted that with SMD 3.0 that your new camera is probably loaded with, that the IVS triggers are delayed as a result of improving them to eliminate false triggers.
Keep in mind the the camera and Deepstack operate under two different algorithms. Deepstack will begin analyzing as soon as an image is sent, so you may get an alert with only half a person in the image, whereas Dahua AI requires the camera to first detect the object and then determine if it meets the criteria and then it triggers the IVS. So depending on your field of view, the object on the far range of the camera view may not trigger Dahua AI as it wants to be sure it is the object.
DeepStack under certain lighting conditions can probably detect objects at a further range than the Dahua AI, but it also may be some false triggers as well. For example, at night, DeepStack will mistake the mailbox across the street as a person, but the Dahua AI never does.
I have noticed that the newest algorithms for Dahua AI tend to trigger after the object has passed the IVS, whereas before as soon as a foot would touch the line it would trigger, but now the person can sometimes be past the line before it triggered.
You either have to watch live view and turn on the IVS rules and have someone walk around or have an SD card in the camera and playback events with the IVS rules turned on to see when they are triggering.
Field of view and how you set up the IVS rules impacts the success or failure of the IVS rule and triggering.
So first steps are have you done the global config to calibrate the camera for your field of view and do you have min object size as 0,0?
Also, many have noted that with SMD 3.0 that your new camera is probably loaded with, that the IVS triggers are delayed as a result of improving them to eliminate false triggers.
Keep in mind the the camera and Deepstack operate under two different algorithms. Deepstack will begin analyzing as soon as an image is sent, so you may get an alert with only half a person in the image, whereas Dahua AI requires the camera to first detect the object and then determine if it meets the criteria and then it triggers the IVS. So depending on your field of view, the object on the far range of the camera view may not trigger Dahua AI as it wants to be sure it is the object.
DeepStack under certain lighting conditions can probably detect objects at a further range than the Dahua AI, but it also may be some false triggers as well. For example, at night, DeepStack will mistake the mailbox across the street as a person, but the Dahua AI never does.
I have noticed that the newest algorithms for Dahua AI tend to trigger after the object has passed the IVS, whereas before as soon as a foot would touch the line it would trigger, but now the person can sometimes be past the line before it triggered.
Last edited:
Just tonight I discovered a trigger was missed completely. Attached are the IVS settings along with a short video of the guy walking right through the Tripwire and Intrusion settings.
This is not the scenario I was trying to solve in my earlier message. That was aimed at people walking the other direction .... from the far end of the alley toward the camera. But since this was the first known entire missed trigger, I wanted to post it and ask for thoughts. In BI with DeepStack, it would have triggered and alerted within the first 2 steps of coming into view.
View attachment Loryta4MPVF.20210730_231230_1.mp4
You need to make the IVS rules larger and think about 3-dimensional.
I would extend the zig zag and the Intrusion box to the top of the fence on the left.
That person's lower body was the only thing within the IVS triggers and his dark pants matched the dark of the ground, but his head is at the top of the fence, so it was missed. Extend the zones to where his head is and you should start getting more.
Also, are you on default settings or have you dialed this in to your area? Did you do the Global Config? It seems like there should be more contrast and I would expect IR to be brighter at that close proximity.
I would extend the zig zag and the Intrusion box to the top of the fence on the left.
That person's lower body was the only thing within the IVS triggers and his dark pants matched the dark of the ground, but his head is at the top of the fence, so it was missed. Extend the zones to where his head is and you should start getting more.
Also, are you on default settings or have you dialed this in to your area? Did you do the Global Config? It seems like there should be more contrast and I would expect IR to be brighter at that close proximity.
Last edited:
