5.9.9 - January 17, 2025 - More Pre-6.0 feature releases

[5.9.9.28]

EDIT (3/16/2025): This feature was originally broken. It was fixed in 5.9.9.29.

Via email from Ken...
Starting with 5.9.9.28, an AI DAT file will always be created in the BI temp folder if it's not already being saved to the Alerts folder.

Default Location: C:\ProgramData\Blue Iris\temp
File Name: alert_camshortname.dat

Note: This is somewhat analogous to how Blue Iris saves temp JPG files (named like temp_camshortname.jpg) to the Alerts folder, so that they can be used by actions in the 'On alerts' action set that use the &ALERT_PATH macro (more here).

Once the update is available (and the new feature is verified), I'll be modifying my BI_gif_tool utility.
This new feature will eliminate the need to:
  1. Purposely save AI DAT files (Alert tab > AI configuration > 'Save AI analysis .DAT files')
  2. Or extract the needed JPGs from the BVR file (AFTER a significant delay waiting for the video buffer to be written to disk).
 
Last edited:
5.9.9.25 broke AI for me. Went back to 5.9.9.24 and all is fine again.

Normally in the AI setup it lists the custom models available, but in .25 it shows a folder.
I run Codeproject AI 2.9.97 in docker on a different machine.
And BI reported status. But again by just going back to .24 all is fine.

Maybe I need to configure it different in .25 ?
Did you solve the problem or stay at .24?

Patrick
 
5.9.9.25 broke AI for me. Went back to 5.9.9.24 and all is fine again.

Normally in the AI setup it lists the custom models available, but in .25 it shows a folder.
I run Codeproject AI 2.9.97 in docker on a different machine.
And BI reported status. But again by just going back to .24 all is fine.

Maybe I need to configure it different in .25 ?

Same for me. My setup is similar to yours--CPAI running on another machine, most current version.
 
5.9.9.25 broke AI for me. Went back to 5.9.9.24 and all is fine again.

Normally in the AI setup it lists the custom models available, but in .25 it shows a folder.
I run Codeproject AI 2.9.97 in docker on a different machine.
And BI reported status. But again by just going back to .24 all is fine.

Maybe I need to configure it different in .25 ?
How did you only roll back one version? I upgraded to 9.32 and the newest roll back I can get it 5.9.8.5 (10/03/2025).

Oh and AI broke in .32 for me. I am still using deepstack because it has been flawless up to this update. Reboot did not fix it.
 
How did you only roll back one version? I upgraded to 9.32 and the newest roll back I can get it 5.9.8.5 (10/03/2025).

Oh and AI broke in .32 for me. I am still using deepstack because it has been flawless up to this update. Reboot did not fix it.

You can only do it if you have the update file. BI doesn't not publish every update, but some of them can be found here:


Also, in your Blue Iris 5 directory, you might have an update directory, and if you've downloaded an update, it might be there.
 
You can only do it if you have the update file. BI doesn't not publish every update, but some of them can be found here:


Also, in your Blue Iris 5 directory, you might have an update directory, and if you've downloaded an update, it might be there.
Dang for some reason the last update I downloaded was 5.9.9.10 12/8/24. Guess I will try 5.9.9.19 from the drive.
 
Last edited:
[5.9.9.35]

Well, this seems to be something new -- Blue Onyx AI Server.
Perhaps I've overlooked its discussion in other threads?

View attachment 218064



Quick search of mentions on ipcamtalk ... Google Search
I am running it in docker on my unraid box and it's working really well.

I don't think it's a direct replacement for deepstack or CPAI because last I read the dev doesn't seem to want to iimplement custom models or the coral tpu's which is something I have always wanted to try but CPAI seemed to buggy to me to switch over to.

So far it's working pretty well at 17,971 requests handled on a i7-8700 with 26 other docker containers running.

Average Inference - 373 ms​

Min Inference - 352 ms​

Max Inference - 586 ms​

Average Processing - 391 ms​

Min Processing Time - 358 ms​

Max Processing - 600 ms​

Average Analysis Round Trip Time - 404 ms​

Min Analysis Round Trip Time - 359 ms​

Max Analysis Round Trip Time - 2282 ms​

 
[5.9.9.35]

Well, this seems to be something new -- Blue Onyx AI Server.
Perhaps I've overlooked its discussion in other threads?

View attachment 218064



Quick search of mentions on ipcamtalk ... Google Search
Yes, I'm curious about this as well. I currently use the camera onboard AI which works well for me, but I would like to experiment with this to see how it compares. Anyone know what objects can be detected with Blue Onyx?
 
Yes, I'm curious about this as well. I currently use the camera onboard AI which works well for me, but I would like to experiment with this to see how it compares. Anyone know what objects can be detected with Blue Onyx?
I have googled and looked though all sorts of github pages and documentation without a single reference to a object list. However the Blue onyx tool does have a test image feature. If you want to send me some screenshots or links to images with the objects you are curious about I can run them though my set up.
 
I have googled and looked though all sorts of github pages and documentation without a single reference to a object list. However the Blue onyx tool does have a test image feature. If you want to send me some screenshots or links to images with the objects you are curious about I can run them though my set up.
Aren't the object labels model-dependent? I read that you can configure Blue Onyx to use MikeLud's custom models (see the 1st post here).
 
Aren't the object labels model-dependent? I read that you can configure Blue Onyx to use MikeLud's custom models (see the 1st post here).
Yes, but I am just running the stock rt-detrv2-s.onnx model that comes with it. I am not a wizard so I haven't tried to pull a different model and start that using docker. Plus without GPU acceleration ( just banging it out with a CPU) a smaller model works better I think.

Edit: and looking at the github GPU's aren't supported on linux anyway. Guess I need to try to migrate this to my windows BI machine.
 
Oh I think I found the base object list:


NAMES:
- person - fire hydrant - elephant - skis - wine glass - broccoli - diningtable - toaster
- bicycle - stop sign - bear - snowboard - cup - carrot - toilet - sink
- car - parking meter - zebra - sports ball - fork - hot dog - tvmonitor - refrigerator
- motorbike - bench - giraffe - kite - knife - pizza - laptop - book
- aeroplane - bird - backpack - baseball bat - spoon - donut - mouse - clock
- bus - cat - umbrella - baseball glove - bowl - cake - remote - vase
- train - dog - handbag - skateboard - banana - chair - keyboard - scissors
- truck - horse - tie - surfboard - apple - sofa - cell phone - teddy bear
- boat - sheep - suitcase - tennis racket - sandwich - pottedplant - microwave - hair drier
- traffic light - cow - frisbee - bottle - orange - bed - oven - toothbrush
 
  • Like
Reactions: 105437
Oh I think I found the base object list:


NAMES:
- person - fire hydrant - elephant - skis - wine glass - broccoli - diningtable - toaster
- bicycle - stop sign - bear - snowboard - cup - carrot - toilet - sink
- car - parking meter - zebra - sports ball - fork - hot dog - tvmonitor - refrigerator
- motorbike - bench - giraffe - kite - knife - pizza - laptop - book
- aeroplane - bird - backpack - baseball bat - spoon - donut - mouse - clock
- bus - cat - umbrella - baseball glove - bowl - cake - remote - vase
- train - dog - handbag - skateboard - banana - chair - keyboard - scissors
- truck - horse - tie - surfboard - apple - sofa - cell phone - teddy bear
- boat - sheep - suitcase - tennis racket - sandwich - pottedplant - microwave - hair drier
- traffic light - cow - frisbee - bottle - orange - bed - oven - toothbrush
Appreciate the list! Several of the objects really make me scratch my head and wonder why.