CodeProject.AI Version 2.0
- Thread starter MikeLud1
- Start date
Village Guy
Pulling my weight
Keep the faith. I can't really offer you any constructive help aside from making sure that you have deleted the old program data and made a full un install followed by a full install. I was fighting for days to transition from V1.67 to to 2.07 without success. I did a full removal of V2.07, installed V2.08 and Viola.OMG, when you wish you'd never messed with it when it was workinglol
I've tried repairs, uninstalls (including deleting every reference I can find manually), wmi repair, and still every time I restart the PC I then have to spend 10 minutes stopping, starting, stopping, starting CPAI from BI itself and also from the CP Dashboard. I'm getting really fed up with it now so taking a break before throwing it out the window lol
on 2.0.8 and latest BI.
Anyone else with the same issue?
@Lta, I would try the following: uninstall cpai, delet the codeproject install directory, shutdown BI service, reboot, reinstall CPAI, set BI service to "delayed start", then restart BI service.
@MikeR33, do you have windows services configured so the BI service has a "delayed startup"? I battled with similar problems for a while until I did that. Since switching BI to delayed startup, I've done a few BI updates, CPAI from 2.0.7 to 2.0.8, and numerous reboots... without any Ai issues (including ALPR). Note, I had to reconfigure delayed startup after updating BI... it resets default after upgrading. Not sure if they'll fix the underlying problem, or change the default startup behavior, but either way... would be nice to no longer have to worry about this.
@MikeR33, do you have windows services configured so the BI service has a "delayed startup"? I battled with similar problems for a while until I did that. Since switching BI to delayed startup, I've done a few BI updates, CPAI from 2.0.7 to 2.0.8, and numerous reboots... without any Ai issues (including ALPR). Note, I had to reconfigure delayed startup after updating BI... it resets default after upgrading. Not sure if they'll fix the underlying problem, or change the default startup behavior, but either way... would be nice to no longer have to worry about this.
Thanks for the idea, i'll give that a try at the next reboot. Fortunately it's stable once finally working.
Thanks for the idea, yes that what I tried to do, full uninstall, delete all remaining folders, full reinstall. I had already moved from 1.6 to 2.0.6 and it just worked fine, should have left it alone lol
Village Guy
Pulling my weight
For anyone who thinks AI is infallible, BI triggered and captured my visitor and AI decided it was a ghost 
Attachments
FYI, for anyone who hasn't found this yet, Windows will not start services set to delayed start until all services set to automatic start first.
I tried this, and I waited a long long time, and finally set both services back to automatic.
It turned out that I sorted the service list by startup type, and found that on my machine there was a security chip service (Intel TPM Provisioning Service) that never started since I don't use TPM.
Disabled that service and everything now starts as expected.
dan1234561
n3wb
Hi all - the ALPR is quite slow, because it's CPU only. Anyone know if there are plans to release a GPU/CUDA version?
rbc1225
Getting the hang of it
I could be wrong because I can’t even get mine to work on the CPU but if I am not mistaken the version that when it loads it automatically looks to see if you have a compatible GPU. If you don’t, it loads the CPU version.
dan1234561
n3wb
Weird - the object detection works just fine on the GPU… wonder why ALPR is different?
I know it does. In my case (separate Linux server with docker) it just does not enable - Blue Iris and CodeProject.AI ALPR
I have tagged @MikeLud1 before about this issue but I haven't heard anything back.
Last edited:
dan1234561
n3wb
thanks - wonder if my
thanks - that’s exactly my position to - Linux server running docker.
MikeLud1
IPCT Contributor
I only worked on the Windows version of the ALPR model which supports GPU(CUDA 11.7). Chris from CP.AI worked on the other platforms which as I know does not support GPUs at this time.
dan1234561
n3wb
thank you!
Bloody birds still keep getting detected as a "person", I've tried using the "ipcam-generals" models, still false positives. Tried disabling custom models and just using the standard
YOLO 6.2, but still getting bird people. any ideas guys? i dont really want to increase the min confidence above 60%.
@M4DM4NZ Why are you hesitant to increase minimum confidence? 60% is still on the low side IMHO.
If your goal is to reduce or eliminate false positives, then increase the minimum confidence. (In parallel, you can try other models since they were trained with different image sets.)
BTW, alternative config, you can just increase the min confidence for just person and leave the other objects at default like this...
person:80,car,truck,bicycle
If your goal is to reduce or eliminate false positives, then increase the minimum confidence. (In parallel, you can try other models since they were trained with different image sets.)
BTW, alternative config, you can just increase the min confidence for just person and leave the other objects at default like this...
person:80,car,truck,bicycle
Last edited:
Thanks for the tip Actran, yeah i wanted low confidence because further in the background of the image people can sometimes jump the fence and appear small on the screen in the distance.
but i'll give your ratio trick a try...
Thanks
Thanks truglo for the tip on the delayed service start, it's all working ok now on a reboot. Maybe there should be a change to the BI service when AI is enabled to add the service dependency to the BI service so it can then be a normal auto start but won't do that until the AI is up first rather than use delayed start and risk a no start should a random other auto service fail to function which is nothing to do with BI\AI.