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/

[Tinman]

Can someone tell me what these numbers represent ?

 

[DanDenver]

Location of the found object (in the image)

 

[spammenotinoz]

Also, this error which is bogus as the AI is running and works perfect. Ken has told me this will be fixed in 5.6.0.2

View attachment 136943

I tried 5.6.0 and 5.6.0.1 and the GUI lost all triggers and confirmations and kept "not responding". Mobile App was fine, so I emailed support and rolled back to 5.5.9.6, are you not having issues?

 

[spammenotinoz]

When checked, it uses the camera main stream. But as the AI resizes to 640px anyhow, it needs additional CPU power for resizing before analyzing. So in most cases feeding the sub stream is better.

Essential to ensure your sub-stream is of sufficient quality to identify objects (both resolution and bit-rate).

 

[Tinman]

I tried 5.6.0 and 5.6.0.1 and the GUI lost all triggers and confirmations and kept "not responding". Mobile App was fine, so I emailed support and rolled back to 5.5.9.6, are you not having issues?

Nope mine is running fine on 5.6.0.1 using CPU CodeProject AI

 

[Dave Lonsdale]

Essential to ensure your sub-stream is of sufficient quality to identify objects (both resolution and bit-rate).

For me to understand please, is the resolution shown in the Blue Iris Status alert image (example attached) normally different the 640x480 to which you refer? On my system, the image resolution in BI Status is different for different cameras. Where possible in the camera, I use 640x480 for the substream to hopefully save CPU time. Please clarify.
Thanks

 

[slabbel]

I tried 5.6.0 and 5.6.0.1 and the GUI lost all triggers and confirmations and kept "not responding". Mobile App was fine, so I emailed support and rolled back to 5.5.9.6, are you not having issues?

Same here, you may also notice similar errors to the ones below. I sent the details to Ken.

1 17/08/2022 7:43:36.806 AM B2 Alert image not created
2 17/08/2022 7:43:36.902 AM B2 AddToBIDB failed 1

 

[spammenotinoz]

For me to understand please, is the resolution shown in the Blue Iris Status alert image (example attached) normally different the 640x480 to which you refer? On my system, the image resolution in BI Status is different for different cameras. Where possible in the camera, I use 640x480 for the substream to hopefully save CPU time. Please clarify.
Thanks

View attachment 136988

View attachment 136989

Your sub-stream configuration is fine. Nothing to worry about, there will be some minor image size differences in the alerts, that is to be expected as they are re-sized.

 

[gwminor48]

Nope mine is running fine on 5.6.0.1 using CPU CodeProject AI

Thanks for that. I'm considering trying to switch from Deepstack to CodeSenseAI but maybe I should wait. I would be running CPU version too. I'm still on BI 5.5.7.11. I will continue to watch this thread meantime since the GPU version won't apply to my system.

 

[Dave Lonsdale]

Your sub-stream configuration is fine. Nothing to worry about, there will be some minor image size differences in the alerts, that is to be expected as they are re-sized.

OK thanks spammenotinoz but I guess the argument that using 640x480 to save on CPU cycles by not re-encoding for AI analysis is not valid then? It looks like images are still re-encoded to a 16:9 format?

 

[plainmad]

5.5.8 - June 13, 2022 - Code Project’s SenseAI, | Page 6 | IP Cam Talk

That doesn’t seem to help with getting GPU working for custom modules and why I seem to be missing the other modules unless I’m missing something

 

[MikeLud1]

CodeProject.AI 1.5.6.2 GPU CUDA support Update

• Speed issues are fixed (See the below screenshot, Faster then DeepStack)
• GPU CUDA support for both Windows and Docker

CodeProject.AI Server: AI the easy way.

Version 2.6.2. Our fast, free, self-hosted Artificial Intelligence Server for any platform, any language

www.codeproject.com

CodeProject.AI Server: AI the easy way.

Version 2.6.2. Our fast, free, self-hosted Artificial Intelligence Server for any platform, any language

www.codeproject.com

 

[IAmATeaf]

WOW that is some big improvements almost 10 times as quick?

Do you run the windows or socket version of SenseAI?

 

[Tinman]

Wow is right...even CPU times are FAST ! Also this time I did NOT uninstall AI but just installed over previous version with BI 5.6.0.2 running. Everything went smooth !

 

[gwminor48]

Wow is right...even CPU times are FAST ! Also this time I did NOT uninstall AI but just installed over previous version with BI 5.6.0.2 running. Everything went smooth !

Wow, that sounds encouraging. I would be using CPU version too. I guess I'm not gonna know how well it works for me unless I try it, looks like I may be updating BI & trying Code Project SenseAI.

 

[MikeLud1]

I have been running Windows GPU (CUDA) version all night on two of my cameras that I use for ALPR that have a heavy AI workload, so far no issues to report.

CodeProject.AI is very close to being the go to AI for Blue Iris all that we need is for Ken to iron out the integration issues.

 

[IAmATeaf]

Do they have plans to make use of the onboard intel GPUs or is there no point as they are gutless

Mike do you run the docker version or windows installable version? Might give this a go over the weekend, have an i5-6500, 16Gb RAM but additional GPU just what’s on the CPU.

 

[MikeLud1]

Do they have plans to make use of the onboard intel GPUs or is there no point as they are gutless

Mike do you run the docker version or windows installable version? Might give this a go over the weekend, have an i5-6500, 16Gb RAM but additional GPU just what’s on the CPU.

They do have plans on adding Intel GPU support.
Right now I am running the windows version on two cameras. I did do some testing on the Docker version and are seeing same performance as the windows version.
This weekend I also might fully switch to CodeProject.AI for all my cameras

 

[carteriii]

They do have plans on adding Intel GPU support.

Only because you mention Intel specifically . . . earlier in this thread you referenced the post (which I'll copy here) that they were FIRST going to use DirectML that would support more than just Intel (AMD is the one of particular interest to me). Do you know if that is already done and already contributed to the better "CPU" performance, or if they skipped that to jump to CUDA support, or ???

 

[MikeLud1]

Only because you mention Intel specifically . . . earlier in this thread you referenced the post (which I'll copy here) that they were FIRST going to use DirectML that would support more than just Intel (AMD is the one of particular interest to me). Do you know if that is already done and already contributed to the better "CPU" performance, or if they skipped that to jump to CUDA support, or ???
View attachment 137177

I think they were working on Intel and NVidia support in parallel and Nvidia won. Below are some posts related to Intel GPU support