Dahua IVS vs Deepstack?

[CJ133]

I'm sorry if this is a dumb question but, I couldn't find an exact answer to this.
I also couldn't make up my mind if this belongs in the Dahua section or here..... I'm thinking it's more Blue Iris related though.

I'm using three IPC-T5442T-ZE's with IVS enabled for people only. I also have four IPC-T2231T-ZS's which do not have AI, from what I recall so they're just using basic motion sensing in Blue Iris.

The 5442's do pretty good with not sending alerts from vehicles etc, although it's funny, or maybe scary, they detect motorcycles as people, but it does make sense. It's actually impressive they can see a person moving at 30+mph.

That said, I do sometimes get false alerts from them and of course the plain motion sensing from the other cameras isn't great. Obviously trees blowing etc send alerts but it is what it is. I do believe I've had issues with the 5442's not alerting from people that are further away. Sometimes it does, sometimes it doesn't. It does seem like one camera is better at it than the other.


All of that said, would Deepstack be of any benefit to me? Does it tend to work better than the AI built into the 5442's or worse?

 

[looney2ns]

Re 5442, Have you properly tuned each camera to its FOV? This can affect AI detection. Wittaj Camera tuning Tips.

There is a person on that motorcycle, so it's performing as it should.

Have you performed the Global setup in the cameras?

Some scenes trip wires work best, in others intrusion works best.

Remember you need to draw the IVS rules, so that the camera has a second or two to ID the object prior to it crossing a tripwire or intrusion box.
So don't draw them too close to the edge of the scene.

 

[CJ133]

Re 5442, Have you properly tuned each camera to its FOV? This can affect AI detection. Wittaj Camera tuning Tips.

There is a person on that motorcycle, so it's performing as it should.

Have you performed the Global setup in the cameras?

Some scenes trip wires work best, in others intrusion works best.

Remember you need to draw the IVS rules, so that the camera has a second or two to ID the object prior to it crossing a tripwire or intrusion box.
So don't draw them too close to the edge of the scene.

Now I'll really sound stupid.
What is the "Global setup" ?

I didn't realize the intrusion box worked that way. I have it as wide as I can get it, basically the entire view. I saw no reason not to as I wanted to detect in the entire area. I can go back and change that if it's part of my problem, if not all of the problem.
I tuned the cameras reasonably well, although I'm sure there's room for improvement.

I'm assuming based on your response that Deepstack isn't worth my time as far as the 5442's go, assuming I set the 5442's up correctly?

 

[wittaj]

Regarding Dahua IVS, whether to use BI AI is up to you, but of course, the AI in the camera may be more than sufficient for your needs without needing BI AI. Do you need the orange box around every object? Do you want to identify animals or logos? Or is just human or vehicle sufficient.

The camera AI is useful to many people, but BI has way more motion setting granularity than the cameras, and some people need that additional detail, especially if wanting AI for more than a car or person. For folks that want AI and alerts on animals or specifically a UPS truck then they need the additional AI.

There isn't really a best practice because every field of view is different and use case and needs are different.

To many here, BI motion without AI is more than adequate for what they do.

To many here, camera AI is more than adequate for what they do.

To many here, using the BI AI adds additional functionality that the above alone can not do.

It comes down to testing with each field of view and which one gives you the most consistent results.


While some of that third party stuff is cool like tagging was it a dog or a bear, I don't need all that fancy stuff. If my camera triggers BI to tag an alert for human or vehicle and BI can accomplish what I need by way of a text or email or push or whatever, that is sufficient for my needs. I just want to be alerted if a person or vehicle is on my property and the camera AI does a fine job with that.

However, I do run BI AI on one camera so that it knocks out headlight shine so that the alert image includes the vehicle. The camera AI will trigger for a car, but the alert image was always just the headlights.

The true test....I have found the AI of the cameras to work even in a freakin blizzard....imagine how much the CPU would be maxing out sending all the snow pictures for analysis to CodeProject LOL. My non-AI cams in BI were triggering all night. This picture was ran through Deepstack (without the IVS or red lines on it) and it failed to recognize a person in the picture, but the camera AI did. This pic says it all and the video had the red box over it even in complete white out on the screen:

See this thread on how using just Dahua AI may be sufficient for your needs:

Who uses Dahua AI capable cameras? Reliable AI for triggering events? Pro's/con's?

 

[CJ133]

Regarding Dahua IVS, whether to use BI AI is up to you, but of course, the AI in the camera may be more than sufficient for your needs without needing BI AI. Do you need the orange box around every object? Do you want to identify animals or logos? Or is just human or vehicle sufficient.

The camera AI is useful to many people, but BI has way more motion setting granularity than the cameras, and some people need that additional detail, especially if wanting AI for more than a car or person. For folks that want AI and alerts on animals or specifically a UPS truck then they need the additional AI.

There isn't really a best practice because every field of view is different and use case and needs are different.

To many here, BI motion without AI is more than adequate for what they do.

To many here, camera AI is more than adequate for what they do.

To many here, using the BI AI adds additional functionality that the above alone can not do.

It comes down to testing with each field of view and which one gives you the most consistent results.


While some of that third party stuff is cool like tagging was it a dog or a bear, I don't need all that fancy stuff. If my camera triggers BI to tag an alert for human or vehicle and BI can accomplish what I need by way of a text or email or push or whatever, that is sufficient for my needs. I just want to be alerted if a person or vehicle is on my property and the camera AI does a fine job with that.

However, I do run BI AI on one camera so that it knocks out headlight shine so that the alert image includes the vehicle. The camera AI will trigger for a car, but the alert image was always just the headlights.

The true test....I have found the AI of the cameras to work even in a freakin blizzard....imagine how much the CPU would be maxing out sending all the snow pictures for analysis to CodeProject LOL. My non-AI cams in BI were triggering all night. This picture was ran through Deepstack (without the IVS or red lines on it) and it failed to recognize a person in the picture, but the camera AI did. This pic says it all and the video had the red box over it even in complete white out on the screen:

See this thread on how using just Dahua AI may be sufficient for your needs:

Who uses Dahua AI capable cameras? Reliable AI for triggering events? Pro's/con's?

Headlights is a constant problem with one of my cameras that doesn't have the AI built in.
It's almost the only thing it ever detects, actually and I don't want it.

I think something must have changed because for a while if either of my front cameras had an alert, there was 100% a person there be it a motorcycle, pedestrian etc. Now, it's doing false alerts quite a bit and I'm not sure why. I've yet to figure out what it's thinking is a person.

I'm going to go over the thread looney2ns posted regarding the tuning you recommend as well as the link you just shared. I have no doubt I can improve things.
I just wasn't sure what I should be putting my effort into.

 

[wittaj]

Yeah, I use BI AI on a few cams to deal with the headlight issue.

At night, my one camera, that has a straight on angle of the street to get a side profile of a car, would either find a car but the alert image would be the lightshine on the street or just a part of the vehicle, or it would trigger out nothing found due to headlight bounce off the street.

Once I added a cancel banana in the field, it now will go thru all the images and select the best one, which gives me the whole vehicle in the frame and it eliminated the nothing found as well, but I have to use the burn label onto image for the vehicle to be in the center of the frame, so every couple days I simply delete these out of the Alerts folder, but they remain in the thumbnail alerts. It makes for scrubbing video much quicker as I can skip looking at video of known vehicles.

 

[looney2ns]

Now I'll really sound stupid.
What is the "Global setup" ?

I didn't realize the intrusion box worked that way. I have it as wide as I can get it, basically the entire view. I saw no reason not to as I wanted to detect in the entire area. I can go back and change that if it's part of my problem, if not all of the problem.
I tuned the cameras reasonably well, although I'm sure there's room for improvement.

I'm assuming based on your response that Deepstack isn't worth my time as far as the 5442's go, assuming I set the 5442's up correctly?

It's depends, in my experience with the proper setup of AI in the cams, that's all you need for people and cars.
BI motion can be very reliable when tuned, and by tuned, this isn't something that happens in 10 minutes typically, it takes time and patience.
For non AI cams, Use BI zones judicially, use zone crossing and other BI tools.
Blue Iris Support - YouTube