CodeProject.AI on a separate PC on the network

HBilbrey

n3wb
Sep 12, 2020
3
0
Texas
Can anyone help me with setting up BI host machine with CPAI on a separate pc on the network please?

I have the instance of CPAI setup and running on the other pc on the network, and the BI host (5.7.8.1) is connected and taking requests, but nothing is ever found during the analysis even though people are walking through the scene.

1690077364571.png
1690078386947.png
1690078511636.png

What should be placed in the "Use Custom Models" field of the BI host?

1690077644287.png
1690078433859.png
Any help would be greatly appreciated!
 
If I run the CPAI Explorer and run test images through it, it responds with ""No predictions returned"

During the initial setup everything looked fine, but I found this in the log.

"20:34:24:ObjectDetectionYolo: ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
20:34:24:ObjectDetectionYolo: botocore 1.31.9 requires urllib3<1.27,>=1.25.4, but you have urllib3 2.0.4 which is incompatible.
20:34:24:ObjectDetectionYolo: google-auth 2.22.0 requires urllib3<2.0, but you have urllib3 2.0.4 which is incompatible."

Is this the source of the problem, and if yes, how to correct it?


1690080882644.png
 
Try disabling Half Precision or switch to the Object Detection (YOLOv5 .NET) module. The Object Detection (YOLOv5 .NET) also will use your GPU.
To have BI read what custom models are available make sure CP.AI server is running then restart BI service

1690089368559.png
 
FInally got CodeProject to work with my dual Coral TPU m.2 card, but it is in a different PC than my current BI machine.

I am trying to link my BI to this separate PC and for some reason it can't connect to the codeproject ai instance.

The IP address is 192.168.1.87 at the standard port 32168.

When I hit the "open AI dashboard" from the BI AI main menu, it pops open a browser on the BI machine and just sits there, but never loads the 192.168.1.87:32168 codeproject ai instance.

I can't access the instance from any PC but the PC it is installed on...missing something easy here. Help?
 
FInally got CodeProject to work with my dual Coral TPU m.2 card, but it is in a different PC than my current BI machine.

I am trying to link my BI to this separate PC and for some reason it can't connect to the codeproject ai instance.

The IP address is 192.168.1.87 at the standard port 32168.

When I hit the "open AI dashboard" from the BI AI main menu, it pops open a browser on the BI machine and just sits there, but never loads the 192.168.1.87:32168 codeproject ai instance.

I can't access the instance from any PC but the PC it is installed on...missing something easy here. Help?
On my Windows machines, I had to open the CPAI port on the machine running CPAI.
Linux machine running CPAI I could connect with no issues, but on Windows 10 machines, I needed to allow the connection through Windows Firewall.
HTH

Sent from my iPlay_50 using Tapatalk
 
  • Like
Reactions: greg_mitch
On my Windows machines, I had to open the CPAI port on the machine running CPAI.
Linux machine running CPAI I could connect with no issues, but on Windows 10 machines, I needed to allow the connection through Windows Firewall.
HTH

Sent from my iPlay_50 using Tapatalk

Yep, that did it.

Settings - Windows Firewall - Advanced Settings - New Rule

Allow all traffic on 32168...the TPU started popping off analyzing images instantly. Nice!
 
Hi Guys,
I've been at this for three days now and I'm getting no where fast and would appreciate some of your advanced knowledge to help me get up and running. So, I'm running the latest release of Blue Iris on a Win 10 PC and CodeProject AI on a seperate Win 11 NUC PC, i7 3.1GHz, and 16GB of RAM. I'm seeing communication between the two, but no evidence of AI image processing. I've included some screen shots for someone to identify where or what is not configured correctly. Even the CodeProject Explorer wasn't working for me. Here's what things look like:1.jpg2.jpg3.jpg4.jpg5.jpg6.JPG7.JPG8.JPG
 
Last edited:
Hey Mike, Tried that suggestion but still no joy. CodeProject Explorer yielded the same respose- No predictions returned. The YOLOv5.NET reinstall without using download cache should have worked on that diagnostic tool as well, right?
 
Hey Mike, Tried that suggestion but still no joy. CodeProject Explorer yielded the same respose- No predictions returned. The YOLOv5.NET reinstall without using download cache should have worked on that diagnostic tool as well, right?
Your Anti Vires or Fire Wall might be blocking the downloads, try disabling them. Then retry uninstalling the Object Detection (YOLOv5 .NET) module then reinstall it with Do not use download cache enabled
 
Mike, Your last suggestion didn't work. So I dusted off another Windows 10 computer and installed CodeProject. Tried to keep things simple, installed the YOLOv5 .NET and tried to see what would happen using CodeProject.AI Explorer on a test image. Still nothing. What am I doing so wrong?Test.JPG
 
Mike, Your last suggestion didn't work. So I dusted off another Windows 10 computer and installed CodeProject. Tried to keep things simple, installed the YOLOv5 .NET and tried to see what would happen using CodeProject.AI Explorer on a test image. Still nothing. What am I doing so wrong?View attachment 215578
Can you try going to " " in your web browser and see if you get the same results, that is what comes up on mine when I go to this page, not sure if it will make any difference or not as it should go to the same place!
 
So, I guess I've been using the web browser method of using CodeProject. Upon your latest reply I located CodeProject.AI.Server.exe from the Program Files location, hit the file and this was launched. Does this provide any insight?CP1.JPG
 
If I try to use the Custom Detect with the same image file, I also get the same "No predictions returned" message, not sure if this is what is supposed to happen
So, I guess I've been using the web browser method of using CodeProject. Upon your latest reply I located CodeProject.AI.Server.exe from the Program Files location, hit the file and this was launched. Does this provide any insight?View attachment 215588
Not 100% sure, but I suspect that the errors are due to the fact that the service is already running, which would explain the ports in use etc.! Going to the local web page is the proper way to interact with the CPAI service!
 
I tried the web page method several dozen times now. Nothing. I've deleted the program and directory many times and of course reinstalled, all with no luck. What to do next??
 
I tried the web page method several dozen times now. Nothing. I've deleted the program and directory many times and of course reinstalled, all with no luck. What to do next??
Sorry I have never tried to run CPAI on a separate/mesh system or for that matter tried to use the LPR module, all of my three install are all on the same PC that runs BI and it has always worked with no issues
 
If I run the CPAI Explorer and run test images through it, it responds with ""No predictions returned"

During the initial setup everything looked fine, but I found this in the log.

"20:34:24:ObjectDetectionYolo: ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
20:34:24:ObjectDetectionYolo: botocore 1.31.9 requires urllib3<1.27,>=1.25.4, but you have urllib3 2.0.4 which is incompatible.
20:34:24:ObjectDetectionYolo: google-auth 2.22.0 requires urllib3<2.0, but you have urllib3 2.0.4 which is incompatible."

Is this the source of the problem, and if yes, how to correct it calcolo codice fiscale?


View attachment 168302
By running the AI server on a dedicated machine, you can improve performance for applications like object detection, facial recognition, or image processing without burdening your main system.
 
  • Like
Reactions: Skinny1