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/

[MikeLud1]

Just as a reminder below are all the labels for CP.AI custom models

CodeProject.AI-Custom-IPcam-Models

IPcam-combined Labels:
- person, bicycle, car, motorcycle, bus, truck, bird, cat, dog, horse, sheep, cow, bear, deer, rabbit, raccoon, fox, skunk, squirrel, pig

IPcam-general Labels (includes dark models images):
- person, vehicle

IPcam-animal Labels:
- bird, cat, dog, horse, sheep, cow, bear, deer, rabbit, raccoon, fox, skunk, squirrel, pig

IPcam-dark Labels:
- Bicycle, Bus, Car, Cat, Dog, Motorcycle, Person

 

[MikeLud1]

So does that mean that, without hardware acceleration, the CPU load is not increased when "use main stream if available" is checked and it's all down to the GPU? But I think the question has been asked previously:- if the image is resized is there any advantage in checking use main stream if available? And is it resized all the way down to 640x480 prior to being analyzed?

There is no advantage in checking use main stream if available it will only slow down the detection.
It is resized to fit a 640x640 image prior to being analyzed because the models were original trained at 640x640

 

[wepee]

@actran

In first screenshot below, make sure your custom model folder is set correctly---the default path is

C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo\custom-models

Thank you very much for your guidance, I just wish this thread started off with a Tutorial on PAGE 1,
so that the OP can update page 1 for any updates or changes when he sees fit.

Otherwise ppl like me as a total beginner will have a hard time reading pages and pages of information,
trying to get the answer.

Anyway, here is my question, how did you tell BI to use custom model folder?
No matter what I do, it is always stay grey out ( I could not enable this option)

Thank you again.


View attachment 2022-10-02_17-28-51.mp4

 

[Dave Lonsdale]

@actran
We did it by changing the "0" to a "1" in the registry line "deepstack_custom" but it's not greyed out in the up-to-date Blue Iris and you will likely need to change the path.


Thank you very much for your guidance, I just wish this thread started off with a Tutorial on PAGE 1,
so that the OP can update page 1 for any updates or changes when he sees fit.

Otherwise ppl like me as a total beginner will have a hard time reading pages and pages of information,
trying to get the answer.

Anyway, here is my question, how did you tell BI to use custom model folder?
No matter what I do, it is always stay grey out ( I could not enable this option)

Thank you again.

View attachment 141384
View attachment 141387

 

[wepee]

For the custom model options in BI are greyed out checkoff Auto start/stop the click ok and reopen the AI settings and the custom model options in BI will not be greyed out

Hi Mike, how to enable use custom folder? I follow your instruction shown above, it does not work.
It is still showing grey out. Do I need to install the latest BI v5.6.x.x? to solve this problem.
Currently, I am using BI v5.5.9.6.

 

[Tusabrat]

Hi Mike, how to enable use custom folder? I follow your instruction shown above, it does not work.
It is still showing grey out. Do I need to install the latest BI v5.6.x.x? to solve this problem.
Currently, I am using BI v5.5.9.6.

For me, I had to stop the AI service first, then the checkbox was available in BI. I made my selection, then turned the AI service back on. Don't know if that will work for you.

Since upgrading to BI 5.6.2.3 and AI 1.6.5, I no longer have to bounce the Service.

 

[wepee]

For me, I had to stop the AI service first, then the checkbox was available in BI. I made my selection, then turned the AI service back on. Don't know if that will work for you.

Nope did not work for me. I think I need to upgrade my BI first and check again.

 

[Tusabrat]

Can you not just put 'cat' in the 'To confirm' box and leave the 'To cancel' box empty?

If you include 'person' in the cancel then the clip isn't included if there is a person in the clip. So it is doing exactly what you asked.

If you only want clips where the AI sees cats, only include 'cat' but don't cancel for anything.

If you are still wanting to record other objects (eg. 'person' or 'cat') then I would create the camera with those objects included, then clone the camera and make one that only detects 'cat'. You'd then use that to scroll through events with the cats.

I think there is a way to filter the flagged events too, but I don't remember how to do that off the top of my head, but you might find it in the help file.

Thanks for the replies.

When I leave the To Cancel field empty, I get non-stop recording as us humans are both categorized as 'dogs' (I have cats,dogs in the To Confirm box as my 8kg cat is also viewed as a dog). At least with 'person' in the cancel box, I don't get non-stop recording when one of us cooking in the kitchen, for example. I guess I will have to live with losing the clips where humans and cats are in the same frame.

Edit: The funny thing is - when I use the 'To Cancel: person', it always recognises us 100% as persons, not dogs.

 

[Tinman]

Hi Mike, how to enable use custom folder? I follow your instruction shown above, it does not work.
It is still showing grey out. Do I need to install the latest BI v5.6.x.x? to solve this problem.
Currently, I am using BI v5.5.9.6.

YES

 

[actran]

Nope did not work for me. I think I need to upgrade my BI first and check again.

@wepee I had assumed you were on the latest BI release. Yes, it looks like you need to update. The current release is 5.6.2.3 as of today. From experience, 5.6.2.3 will allow you to edit the custom folder field.

In older BI releases, it's greyed out.

Alternatively you can specify the custom folder path using regedit. Read CodeProject.AI Server: AI the easy way.

 

[jaydeel]

There is no advantage in checking use main stream if available it will only slow down the detection.

This post provides a reason for checking ”Use main stream if available”... Lo-Res trigger JPG

Basically, per Ken, enabling this setting ensures that Blue Iris saves the image that was ACTUALLY used to confirm/cancel the alert

This probably doesn't matter most of the time, but it might eliminate an opportunity for confusion.

 

[dirk6665]

Reaching out to the graphic gurus out there...

The last update to CP.AI (1.6.5-Beta) has once again thrown my detection times using CPU in the trash (Less than 3 detections per second). Since this i7 machine I have does not physically have any additional PCI "power" connectors for my GeForce 930 ... can someone recommend to me an inexpensive supported Nvidia solution that does NOT require extra power that will get me at least 20 OPS? <OR> can someone (looking at MikeLud1 here) tell me if the CodeProject developers will be supporting USB based TPU's from Coral like this one...



Because to me - this would be IDEAL as it's not bound to any specific machine and its externa;, universal and MUCH cheaper than most NVidia cards (Not to mention I don't have to find a free slot).

--Crossing fingers, eyes and toes

 

[toastie]

In a camera's AI setting may have in the box, Mark as vehicle the entry car,truck,bus,vehicle.
then in the box, To confirm the entry person,car

In the camera AI settings you need to have something in the To confirm box like person,car

 

[biggen]

Reaching out to the graphic gurus out there...

The last update to CP.AI (1.6.5-Beta) has once again thrown my detection times using CPU in the trash (Less than 3 detections per second). Since this i7 machine I have does not physically have any additional PCI "power" connectors for my GeForce 930 ... can someone recommend to me an inexpensive supported Nvidia solution that does NOT require extra power that will get me at least 20 OPS? <OR> can someone (looking at MikeLud1 here) tell me if the CodeProject developers will be supporting USB based TPU's from Coral like this one...

View attachment 141410

Because to me - this would be IDEAL as it's not bound to any specific machine and its externa;, universal and MUCH cheaper than most NVidia cards (Not to mention I don't have to find a free slot).

--Crossing fingers, eyes and toes

Coral support is coming. Go back a few pages in this thread and you can see I asked and was answered.
Not sure on GPU. You can read through the thread here and see what others are using.

 

[dirk6665]

Coral support is coming. Go back a few pages in this thread and you can see I asked and was answered.
Not sure on GPU. You can read through the thread here and see what others are using.

Yes, I did see the post regarding Coral support but just to be clear, I am specifically inquiring about the USB type. I see what others are using as far as GPU but just because someone is using a certain hardware doesn't necessarily mean they would recommend the same.

 

[jrbeddow]

I think many of us are waiting to see what sort of performance we can expect when (if?) Coral support actually takes place; as it is easy to plan support, but another to implement, and it remains an open question until then.

Obviously, the ~$60 USB unit would be amazing if that works well, as that unit would be easy to install and appears to use under 5 watts even under load. Google may find us flooding them with orders at that point.

 

[wepee]

YES

Alright thanks!

 

[sebastiantombs]

@MikeLud1

Hey Mike. I just installed the demo version of BI 5.6.2.3 on a spare i7-6700K, 32GB RAM, M2 boot drive with a Radeon card in it, along with the latest SenseAI version. I had several problems and still have one in particular that's irritating.

First, just getting custom models enabled was a PITA between shut downs, reboots and so on.

Next, all I got was timeouts from SenseAI using the ipcam-general model, the same, or equivalent to, the general model I'm using on DS on the main machine. Turns out that SenseAI sees the Radeon card and enables GPU even though it doesn't work on a Radeon card and the dashboard was reporting YOLO as using CPU. So I edited the YOLO config .json and got object detection working on the CPU. The only thing I have enabled is YOLO. Detection is terrible. It's only detecting about a quarter, or less, when compared to DeepStack. Admittedly conditions aren't the greatest with the rain than hasn't let up all day getting on the lenses, but DS is still doing a far, far, better job at detection. Detection times aren't too bad considering it's a CPU version, 200-400ms.

The lagging and annoying thing is that none of the menus in the control panel work at all. If I wanted to enable face detection I can't because the menu doesn't open up at all. I ran a repair of the Sense AI install, after shutting down BI/SenseAI and rebooted yet again and that didn't cure it. I was able to initially, but somewhere along the line it failed and has resisted all efforts to get it back.

I think the developers need to know that it enabled GPU detection with an incompatible card. And as I think about it, I switched to CPU immediately when I opened the control panel prior to configuring SenseAI in Blue Iris.

 

[wepee]

@wepee I had assumed you were on the latest BI release. Yes, it looks like you need to update. The current release is 5.6.2.3 as of today. From experience, 5.6.2.3 will allow you to edit the custom folder field.

In older BI releases, it's greyed out.

Alternatively you can specify the custom folder path using regedit. Read CodeProject.AI Server: AI the easy way.

Ok thanks for the reply.

I won't take the registry modification route though.

I purposely hold off the upgrade, because BI updates are notoriously messed up a working BI system.
Now that v5.6.2.3 is already out, I will have to take the chance of upgrading it (finger-crossed hopefully no issue)

 

[MikeLud1]

@MikeLud1

Hey Mike. I just installed the demo version of BI 5.6.2.3 on a spare i7-6700K, 32GB RAM, M2 boot drive with a Radeon card in it, along with the latest SenseAI version. I had several problems and still have one in particular that's irritating.

First, just getting custom models enabled was a PITA between shut downs, reboots and so on.

Next, all I got was timeouts from SenseAI using the ipcam-general model, the same, or equivalent to, the general model I'm using on DS on the main machine. Turns out that SenseAI sees the Radeon card and enables GPU even though it doesn't work on a Radeon card and the dashboard was reporting YOLO as using CPU. So I edited the YOLO config .json and got object detection working on the CPU. The only thing I have enabled is YOLO. Detection is terrible. It's only detecting about a quarter, or less, when compared to DeepStack. Admittedly conditions aren't the greatest with the rain than hasn't let up all day getting on the lenses, but DS is still doing a far, far, better job at detection. Detection times aren't too bad considering it's a CPU version, 200-400ms.

The lagging and annoying thing is that none of the menus in the control panel work at all. If I wanted to enable face detection I can't because the menu doesn't open up at all. I ran a repair of the Sense AI install, after shutting down BI/SenseAI and rebooted yet again and that didn't cure it. I was able to initially, but somewhere along the line it failed and has resisted all efforts to get it back.

I think the developers need to know that it enabled GPU detection with an incompatible card. And as I think about it, I switched to CPU immediately when I opened the control panel prior to configuring SenseAI in Blue Iris.

Can you send me the log file so I can share the Radeon issue you are having. There was one change in the latest version that might slow down CPUs that was eliminating the low, medium, high modes and running everything at high mode (640x640).