5.5.8 - June 13, 2022 - Code Project’s SenseAI Version 1 - See V2 here https://ipcamtalk.com/threads/codeproject-ai-version-2-0.68030/

[VideoDad]

You have the default object detection turned on so that will happen on every camera. The default object detection is a generic vision detection model that has things like plants, toothbrushes, pizza, etc. so many people opt to only use custom models and turn off the default object model under global AI settings.

Alternatively, if you still need the default object detection, you can turn it off for individual cameras by adding ", objects:0" to the list of custom models.

 

[Corvus85]

You have the default object detection turned on so that will happen on every camera. The default object detection is a generic vision detection model that has things like plants, toothbrushes, pizza, etc. so many people opt to only use custom models and turn off the default object model under global AI settings.

Alternatively, if you still need the default object detection, you can turn it off for individual cameras by adding ", objects:0" to the list of custom models.

Yes I understand that, however when I turn it OFF, it doesn't detect anything. And I am unable to check 'use custom model'. Refer to my screenshots. Any ideas?

 

[biggen]

Yes I understand that, however when I turn it OFF, it doesn't detect anything. And I am unable to check 'use custom model'. Refer to my screenshots. Any ideas?

It's a bug talked about several pages back. You have to go in to Windows services and stop the CodeProject AI service. After you stop it, open up BI and see if you can check it. You may have to restart BI.

Note: You also have to be running at least BI version 5.6.2.2 to enable it from within BI.

 

[Corvus85]

It's a bug talked about several pages back. You have to go in to Windows services and stop the CodeProject AI service. After you stop it, open up BI and see if you can check it. You may have to restart BI.

Note: You also have to be running at least BI version 5.6.2.2 to enable it from within BI.

Ok I did that, and restarted BI, and... Use custom models is still greyed out.
I'm running 5.6.2.4.

Any ideas?

 

[Tusabrat]

Ok I did that, and restarted BI, and... Use custom models is still greyed out.

Any ideas?

View attachment 141847

View attachment 141848

Before upgrading, I could get it ungreyed by selecting Auto start/stop. Othertimes, I was able to access it by stopping the Codeproject service. On the latest versions, it isn't an issue.

 

[MikeLud1]

Ok I did that, and restarted BI, and... Use custom models is still greyed out.

Any ideas?

View attachment 141847

View attachment 141848

You need to check Auto start/stop then you can check Use custom model

 

[Corvus85]

You need to check Auto start/stop then you can check Use custom model

Thanks, that worked. Although now for some reason, it's taking longer to analyze? I walked exactly the same path as the other times through the camera's FOV, and it's the middle of the night here so no environmental changes.

In the official Codeproject.AI website, it says not to select 'auto start/stop'. Will it affect things if I leave it checked? Or do I uncheck it after I check 'use custom models'?

 

[Corvus85]

For crying out loud. I thought if I took out 'objects:0' and unchecked 'default object detection' that it would be quicker, but nope!
It's now over 1000ms!

What could explain such long analysis times, even though it's got less objects to analyze?

 

[MikeLud1]

For crying out loud. I thought if I took out 'objects:0' and unchecked 'default object detection' that it would be quicker, but nope!
It's now over 1000ms!

What could explain such long analysis times, even though it's got less objects to analyze?

View attachment 141855

What version CP.AI are you on, the current version is 1.6.6, this does fix some performance issues

 

[Tusabrat]

If your only concern is animals (cats & dogs) you should use the ipcam-animal and ipcam-dark models

Try it again without using zones.

Here are the motion settings I'm using for detecting if our dog is in our living room.
[snip]
And here are my AI settings.
[snip]

Note: This is an older camera that doesn't support sub-streams and works better using edge vector algorithm.

Okay, so I uninstalled CP1.6.6, rebooted, reinstalled CP1.6.5 and it's all perfect now (using just object detection). It's recording the cats, and ignoring the people, just how I want.

 

[Tinman]

For crying out loud. I thought if I took out 'objects:0' and unchecked 'default object detection' that it would be quicker, but nope!
It's now over 1000ms!

What could explain such long analysis times, even though it's got less objects to analyze?

View attachment 141855

Change the BI port for CPAI to default 32168 and then do a complete reboot of pc with BI. Also be on CPAI 1.6.6 for best performance

 

[toastie]

Detect/ignore static objects, I just need to check this, I want my own vehicle parked on the driveway to be ignored and a new vehicle that is driven up to be detected, do I have the box ticked or left unchecked?

CP AI is working probably now better here than I had DeepStack, and that includes the issue I had with car headlights at night. That's a big plus for CP AI.

I've one camera looking mostly at my RV, however at night in rain I'm getting a lot of triggers that go onto being Alerts that CP AI doesn't reject. It's as if the IR illuminated rain in front of the vehicle concludes that the vehicle is moving. In the rain AI also identified a person being in the passenger seat. In rain it also identified the camera's name overlay of being of some significance, I forget what object it concluded.

I'm also surprised that on another camera that looks across the street, AI rightly finds a neighbor's parked vehicle, but in an area I had excluded in Camera settings>Trigger>Configure>Use zones and hot spots. CP AI doesn't need to waste time analyzing an area of a camera's field I have excluded in my Trigger configurations.

Like probably most of us I'd like a BI CP AI system where are all Alerts are of significance and nothing of significance is being missed, and this is where having AI working with a CCTV setup could be a game changer.

 

[Vettester]

do I have the box ticked or left unchecked?

Checked

 

[toastie]

Thanks but I have to be clear, a checked box means static objects are ignored, right?

 

[sebastiantombs]

Yes, but that means they won't trigger an alert. They still need to be analyzed to find out if they're static or not though.

 

[biggen]

Thanks but I have to be clear, a checked box means static objects are ignored, right?

Theoretically anyway. My parked truck in the driveway still gets occasional picked up even though it's static.

 

[jrbeddow]

Theoretically anyway. My parked truck in the driveway still gets occasional picked up even though it's static.

I get this far more frequently for some reason under CodeProjectAI than I did previously under DeepStack. It occurs most often when a person is walking in front of a stationary vehicle. Anyone else notice that trend?

 

[Corvus85]

What version CP.AI are you on, the current version is 1.6.6, this does fix some performance issues

That's the version I'm on.

 

[Corvus85]

Change the BI port for CPAI to default 32168 and then do a complete reboot of pc with BI. Also be on CPAI 1.6.6 for best performance

I'm on the latest version. Why would changing the default BI port make any difference if it's been working fine till now?

 

[jrbeddow]

I'm on the latest version. Why would changing the default BI port make any difference if it's been working fine till now?

The CodeProject team has indicated that while the current setup does allow for the use of port 5000, it is deprecated, as that port is shared with UPnP. As such, they are suggesting migrating to port 32168 when possible.