CodeProject.AI Version 2.0

I shut down BI... and then I uninstalled v2.0.8. Downloaded v2.1.7 and installed it. Waited until all modules finished installing etc. which took about 10 minutes. I have an NVIDIA P400 GPU, should I stick with Object Detection YOLOv5 6.2 for best accuracy and speed for object detection. The only custom model I use is ipcam-general since I only want to detect person and vehicle. Appreciate the help!
 
I shut down BI... and then I uninstalled v2.0.8. Downloaded v2.1.7 and installed it. Waited until all modules finished installing etc. which took about 10 minutes. I have an NVIDIA P400 GPU, should I stick with Object Detection YOLOv5 6.2 for best accuracy and speed for object detection. The only custom model I use is ipcam-general since I only want to detect person and vehicle. Appreciate the help!

I would test both the YOLOv5 6.2 and the .NET one. See which one gives you better performance. Some state .NET gives better. For me it was opposite so I use the 6.2 one instead. I could never get the .NET one to properly use the GPU in a stable fashion so I stopped messing with it. I get pretty good performance on my Nvidia Quadro P1000 GPU.
 
  • Like
Reactions: David L and 105437
We've been working with Ken at Blue Iris and are hopeful to have a resolution for error 500.

Short version: try the latest Blue Iris release, version 5.7.5.6. Should resolve it.

Long version: The error 500 issue revolves around how Blue Iris processes Face Processing requests and responses. There's no "error," rather Blue Iris is passing images to face recognition when there's no detectable face, and CodeProject.AI returns success = false (and code = 500) as a result. The call was made, the image processed, but the recognition processing failed because there's no face to recognise.

Purely an issue around what constitutes success and failure.
 
  • Like
Reactions: slabbel and David L
I would test both the YOLOv5 6.2 and the .NET one. See which one gives you better performance. Some state .NET gives better. For me it was opposite so I use the 6.2 one instead. I could never get the .NET one to properly use the GPU in a stable fashion so I stopped messing with it. I get pretty good performance on my Nvidia Quadro P1000 GPU.
My under-powered 2GB P400 does much better on 6.2 than .NET, roughly 300ms vs 1400.
 
My under-powered 2GB P400 does much better on 6.2 than .NET, roughly 300ms vs 1400.
I'm running 6.2 since upgrading to v2.1.7 today and I'm seeing ~150ms processing times for object detection. Noticeably faster than v2.0.8. I was able to find this data in the BI log file. It would be nice if CPAI would allow filtering of the CPAI log file for events where objects were found so it could be exported for review.
 
Ready for another Noob question. I wanted to try Mike's custom models so I made a backup of ipcam-general (12/06/2022), I copied over Mike's ipcam-general (03/13/2023)

Made sure Default object detection is unchecked:
1683118044561.png

Rebooted PC but I was no longer getting any detections/alerts of any kind:
1683118227953.png

So this morning I decided to copy the original ipcam-general.onyx back but I left the Default object detection unchecked
1683117939510.png

Now I am only getting Person detections were as before I was getting Person, Car.
I haven't changed anything in the Camera AI section other than add the ipcam-general in Custom models when I tried Mike's.
1683118693395.png
 
Yep, that’s it. If you’re using the general model it’s vehicle not car.

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
license-plate Labels: - DayPlate, NightPlate
 
Thanks Guys that was it:

1683120921529.png
 
  • Like
Reactions: Tinman and MikeLud1
All my models are in the below link and they are included in v2.17.

Thanks @MikeLud1, I assume the CPAI installer pulls the most recent versions. I know sometime in the past you were talking about updating them.
Also, is there much difference in accuracy between using the medium vs. large size in CPAI for the ipcam-general model?
 
I keep seeing posts referring to CPAI 2.17, yet I don't see anywhere to download that version. My server says 2.1.6 beta is the latest version.
 
  • Like
Reactions: David L and ARAMP1
I keep seeing posts referring to CPAI 2.17, yet I don't see anywhere to download that version. My server says 2.1.6 beta is the latest version.
Version 2.1.7 was only a prerelease for testing, they just did a prerelease 2.1.8 for testing, I am installing it right now. Once they workout the bugs they will do a release to the general public.

1683145155258.png
 
Last edited:
To All

Version 2.1.8 was just released. Hopefully this version should address all the issues everyone was having.
I recommend doing a clean install using the below steps

1. Uninstall CP:AI
2. Delete the Codeproject Directory completely - "C:\Program Files\Codeproject"
3. Delete "C:\ProgramData\CodeProject"
4. Restart The Computer
5. Reinstall CodeProject along with the modules that you wish to install. (All modules get install once the main install is completed, go to the Dashboard to see the install status for the modules)
6. Restart Computer after all the modules are installed
7. Run BI and CP:AI

 
Thanks for the update. I installed 2.1.8, following the above instructions. Running the most recent version of BI (5.7.62), and get nothing but AI Error 404. I'll roll back BI to a previous version I know worked better and see what happens.
 
  • Wow
Reactions: David L