The CodeProject.AI-Server Discussions

[ChrisMaunder]

Note

Please remember that this project is being maintained by volunteers. Please be patient if responses are slow, or bug fixes and new features aren't implemented as fast as one would hope.

👋 Welcome!

CodeProject.AI Server has moved homebase from CodeProject.com to codeproject.github.io/ to centralise the code, the discussions, and the docs into one self-contained place dedicated to the CodeProject.AI Server product.

This discussion board should focus on the CodeProject.AI Server and installation process only. For all questions on individual modules, please refer to that module's issues board directly. A list of all modules is available at codeproject.github.io/.

Please use this board to:

• 🙋 Ask general questions, or get help or guidance
• 💡 Make suggestions on features or the overall experience
• 💬 Connect with the community

For bugs, please use Issue board. 🐞

[Gordohat]

I have an NVIDIA GeForce GTX 1060 card in my Windows 10 machine. It is a gen 7 Intel core i7 processor.

Should I be able to just run the latest of all the CUDA stuff, including the CUDA drivers, CUDA Toolkit, cuDNN together with the latest NVIDIA graphics and the Microsoft .NET stuff? I am only using CP.AI for use with BlueIris.

If it only will work with specific older versions of the above items, why is this so? I would like to know what is limiting CP.AI from utilising the latest versions and could it be fixed.

[ChrisMaunder]

CodeProject.AI modules are each separate projects so the accurate answer is "it depends on the module", but generally, if your card work with the latest drivers and latest CUDA, then the modules will work too. For the modules we've created we try and ensure they have support for the latest CUDA, but sometimes some things do throw a curve ball, such as the recent NumPy update

[Gordohat]

Thanks. I just need basic YOLO object detection (YOLOv5 6.2 module) for use in Blue Iris on Windows 10. The CP.AI server is running on the same Windows 10 machine. Does that help to be more specific in your answer?

[MikeLud]

You can also use the Object Detection (YOLOv5 .NET) module. This module has the same models as the Object Detection (YOLOv5 6.2) and does not require CUDA to use your GPU

[BeanBagKing]

@ChrisMaunder - There's been some discussion on the future of CodeProject.AI over on the other message boards (for reference: https://www.codeproject.com/Messages/6020926/Re-Code-Project-AI-Server-version-2-8-0). I think the worry is that, with your Aug 31st announcement and with Matthew Dennis no longer involved, that the project may cease to be updated. However, I still see activity from you (this discussion, case in point) over here.

Do you mind commenting on what you think the future of CodeProject.AI looks like at this time? There's a large user base (I feel like anyway) that uses this for Blue Iris, and if there aren't any currently active devs, it would be good to have some forewarning. I know that, as an all-volunteer project, any future plans are subject to sudden change, I'm just trying to get a feel for what's going on. Thank you for all you've contributed over the years, regardless of what the future holds.

[ChrisMaunder]

@matthewDDennis and I will still be contributing for as long as we can. This project is close to our heart, and while we're no longer with CodeProject, we're still able to contribute to the project.

[thecamaroguy]

Since moving here, will the Docker Images be updated and will they be available at the same location -> https://hub.docker.com/r/codeproject/ai-server/tags?page=1&name= or will that move as well?

[matthewDDennis]

It looks like we still have access to the repositories, so they should get updated.
If we must move them, we will update the documentation and the readme file with the latest information.

[jameson71]

It looks like we still have access to the repositories, so they should get updated. If we must move them, we will update the documentation and the readme file with the latest information.

FYI it appears to me that the :latest tag on docker hub still points to 2.6.5 even though 2.8.0 has been up for about 2 months.

[ChrisMaunder]

2.9.X are starting to come online as I get a chance to build and test

[Armand29]

When I went into the server, the service was stopped and the "Install Modules" page was blank. I tried to install 2.8.0 (I downloaded it last week, but the download site doesn't work today) and the page came up, told me I couldn't reach the server, then shut down. Is codeproject.ai dead?

[hspindel]

When I went into the server, the service was stopped and the "Install Modules" page was blank. I tried to install 2.8.0 (I downloaded it last week, but the download site doesn't work today) and the page came up, told me I couldn't reach the server, then shut down. Is codeproject.ai dead?

Same here. Codeproject.ai website seems to be down for maintenance. Adopting a patient attitude while this gets worked out, although it is frustrating that AI has stopped working for my Blue Iris installation.

[ChrisMaunder]

Should be fixed now

[fobrs]

Are there any plans to support the Hexagon NPU on the Snapdragon X Elite processors?

[ChrisMaunder]

Hexagon - not at the moment. Snapdragon: My general comment is that we support arm64 on Linux, macOS and Windows, so many things will work on the new arm chips.

Until I manage to find a shiny new Snapdragon arm64 machine lying around needing a home, my only way of testing arm64 for Windows is via Windows on Arm using VMWare on my M1 mac mini.

[mrkilvo]

Hi All,

I have managed to get this working for Blue Iris, i am sure you are all as frustrated after uninstalling and realising its all messed up and not working, i have only tested this on PC and have Yolov5 3, ALPR and Face Recognition working.

Here is how I got it working

Install CodeProject
Download the Module(s) from the webpage

Unzip and copy the module(s) to the following folder c:\Program Files\CodeProject\AI\modules

run a command prompt as Administrator

Navigate to the Directory c:\Program Files\CodeProject\AI\modules[modulename]

Type ....\setup.bat

Restart codeproject

at least it should get it working for now

[ChrisMaunder]

The install system should be back working as normal now.

[pbradley0]

I'm having trouble adding new modules now. I'm running the server in docker and have cloned the module repository and then ran the setup.sh from inside the docker image and it completes without error. The module shows up in the list of available modules on the server webpage but its status is listed as "Not Available." I've tested this with YOLOv8 and the Coral module and they are both doing it. Anyone else having this issue? Am I missing a step?

[thecamaroguy]

What is the command you used to create your docker and pull request?

[MikeLud]

The below link should have what you are looking for.

https://codeproject.github.io/codeproject.ai/install/running_in_docker.html

[kmjsmith]

yeh im having an issue getting python installed as seen in the photo

[kmjsmith]

[thecamaroguy]

Under the Appsettings.json you will find:

"ModuleInstallTimeout": "00:20:00", around line 92 or something.

Increase that value. I put it to 2 hours, but depending on your internet speeds you might want to increase it to like 20 hours
"ModuleInstallTimeout": "20:20:00",

[kmjsmith]

well it seems to install but it gives the error as seen in my second image even though i have cuda 12.0 installed

[thecamaroguy]

Mine did this and appeared to have installed, but always gave this error. I worked with Chris via the forums and he suggested I remove the module and then adjust this value and reinstall the module. After doing that, it all worked for me and I have not had issue. When a new docker comes out, I pull it down, modify that line, then upgrade any modules. Been flawless since.

[ChrisMaunder]

@kmjsmith There should be an install.log file in the directory for that module (or click the "info" button on the module on the dashboard). If you can paste the install log I can make some guess.

[hspindel]

General question from Code Project AI newbie:

I am using CodeProject (server version 2.6.5) with Blue Iris. Nvidia Quadro P620 GPU. Windows 11. The YOLOv5.3.1 module is working well for me. The install automatically provided ipcam-animal, ipcam-combined, ipcam-dark, ipcam-general, and license-plate.

Thought I would try YOLOv8. It installs fine and works, but the install only provides ipcam-general and license-plate.

Are the additional modules available for YOLOv8? Can I just copy them over from YOLOv5.3.1?

Would there be any benefit to me from using YOLOv8?

[MikeLud]

Are the additional modules available for YOLOv8? No

Can I just copy them over from YOLOv5.3.1? No

Would there be any benefit to me from using YOLOv8? I found there is not much benefit over the other YOLO modules

[hspindel]

Thank you very much for the thorough reply!

[hspindel]

CodeProject (server version 2.6.5) with Blue Iris. Nvidia Quadro P620 GPU. Windows 11. YOLOv5.3.1.

Don't know if this is a new problem, but I just noticed in the logs a complaint that custom-models\actionnetv2.pt does not exist. I verified that this .pt file actually does not exist in the specified directory.

Is this a problem, or should I ignore it? AI object recognition works fine. Is there a change I can make to a config file to tell it not to look for actionnetv2.pt or a way to add it?

What does actionnetv2.pt do?

Thank you.

[tbc725]

I think the actionnetv2.pt is for action detection. Like if the person is standing or sitting, etc.

[hspindel]

Thank you, that's good to know.

[thecamaroguy]

Started to get the following - "Unable to check for updates" on my Docker environment. 2.8.0

It's probably because of the move, but just curious if there is anything I can change to make it point to a new location (if there is one)

[ChrisMaunder]

How about now?

[DeFlanko]

@ChrisMaunder as it pertains to Docker Images -- is this the best way to track the latest Cuda12 builds for docker?

[ChrisMaunder]

Yep!

[fotingo]

Hello,
I am on version 2.6.5 and ever since then, I get "Unable to check for Updates error"

I also notice a lot of false positives, like detecting "person" when it's just a shadow. That started happening after the update.
Which version should I get, 2.8 or 2.9.3?

Is 2.8 the stable version and 2.9.3 beta?

[thecamaroguy]

Thank you for updating the docker images to 2.9.4. So far, everything looks great!

[thecamaroguy]

Well, maybe I was too quick:

when downloading Yolov8 Models...I get the following:
Preparing to download model 'objectdetection-coco-yolov8-pt-nsmlx.zip' for module ObjectDetectionYOLOv8
15:07:47:Downloading model 'objectdetection-coco-yolov8-pt-nsmlx.zip' to '/app/downloads/modules/packages/ObjectDetectionYOLOv8/objectdetection-coco-yolov8-pt-nsmlx.zip'
15:07:53:Preparing to download model 'objectsegmentation-coco-yolov8-pt-nsmlx.zip' for module ObjectDetectionYOLOv8
15:07:53:Downloading model 'objectsegmentation-coco-yolov8-pt-nsmlx.zip' to '/app/downloads/modules/packages/ObjectDetectionYOLOv8/objectsegmentation-coco-yolov8-pt-nsmlx.zip'
15:07:56:Preparing to download model 'objectdetection-custom-yolov8-pt-m.zip' for module ObjectDetectionYOLOv8
15:07:56:Downloading model 'objectdetection-custom-yolov8-pt-m.zip' to '/app/downloads/modules/packages/ObjectDetectionYOLOv8/objectdetection-custom-yolov8-pt-m.zip'
15:08:06:Unable to download and install objectdetection-custom-yolov8-pt-m.zip: Unable to download 'objectdetection-custom-yolov8-pt-m.zip'. Error: Could not find a part of the path '/app/downloads/modules/packages/ObjectDetectionYOLOv8/objectdetection-custom-yolov8-pt-m.zip'.
15:08:17:Unable to download and install objectdetection-coco-yolov8-pt-nsmlx.zip: Response timeout. Try increasing the timeout value
15:08:23:Unable to download and install objectsegmentation-coco-yolov8-pt-nsmlx.zip: Response timeout. Try increasing the timeout value

What timeout do I need to increase?