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/
- Thread starter fenderman
- Start date
what model does object detection .net use as the 'standard model'
confused since I don't see any .pt file in
C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionNet
confused since I don't see any .pt file in
C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionNet
MikeLud1
IPCT Contributor
It use the below onnx models. Also you should not run both Object Detections Net & YOLO at the same time
Thank you.
I have only been running .NET (GPU)
Is their a consenus, that regardless of model .NET or YOLO , in SenseAI is better?
I have only been running .NET (GPU)
Is their a consenus, that regardless of model .NET or YOLO , in SenseAI is better?
MikeLud1
IPCT Contributor
Vettester
Getting comfortable
- Feb 5, 2017
- 822
- 820
MikeLud1
IPCT Contributor
I am using YOLO because .NET does not support custom models at this time. I think it will be added in the future.
As a side note in regards to accuracy: I have found that unchecking the box for "Use motion leading trigger to start detection" has resulted in better detections. At first this seems counterintuitive, until you realize that this means it will start analyzing on the trigger that actually crosses a zone that you care about, rather than motion in some (generally more distant) region that you may not care about at all.
I highly recommend at least testing this to see if works better for you.
I highly recommend at least testing this to see if works better for you.
biggen
Known around here
- May 6, 2018
- 2,691
- 3,065
I have disabled this as well. I was actually getting just head shots of "person" as the alert because it was identifying them too early. I'd rather it wait to get more of the subject in the frame before analyzing and confirming.
Vettester
Getting comfortable
- Feb 5, 2017
- 822
- 820
You need to be careful with this setting. Disabling it doesn't work well with LPR cameras.
Good point, I believe I saw that mentioned in the BI help file as well, but forgot about that use case when I wrote my post.
Yes me. “Auto start/stop with Blue Iris” is definitely checked but I have to press the “Start now” button after a reboot. Was OK with version 1.6.2
biggen
Known around here
- May 6, 2018
- 2,691
- 3,065
Have you tried leaving the auto start/stop unchecked? That is how mine is. My understanding is that when unchecked, BI isn't responsible for it at all since the CP AI runs as a Windows Service and is started at startup and left running.
OK biggen, that seems to have fixed it. Thank you. I notice that this makes “Use custom model folder” (checked) greyed out, but I’m not aware of that being a problem. It is a bug, though, not to start via BI and should be corrected.
biggen
Known around here
- May 6, 2018
- 2,691
- 3,065
It's greyed out but it will still hold any custom models in there you had previously set. If you need to change the custom model location, then just check auto start/stop again, change the model, and uncheck auto start/stop.
In case the AI would stop responding or crash, BI will attempt to restart the AI. If you stop the CPAI it will not restart on its own unless you do reboot the OS.
Granted I have never had CPAI crash yet, but you never know .
wepee
Getting the hang of it
- Jul 16, 2016
- 255
- 59
Help! I have a weird problem, every couple of seconds,
CAPI server routinely will start Object Detection (YOLO) - processing - and spike my CPU usage
Is it normal? I have got only 4 cameras turned on and my shop was closed after working hours.
There is no one is in the shop.
I took a video recording of the instance.
View attachment 2022-10-05_23-30-21.mp4
CAPI server routinely will start Object Detection (YOLO) - processing - and spike my CPU usage
Is it normal? I have got only 4 cameras turned on and my shop was closed after working hours.
There is no one is in the shop.
I took a video recording of the instance.
View attachment 2022-10-05_23-30-21.mp4
Mike, there's been lots of correspondence about model accuracy. I'm getting good results with your ipcam-xxx and would like it to be as good as possible. Hence queries:
- ipcam-combined has both general and animal labels but seems to deliver slightly lower confidence percentages and have longer processing times than when using both models separately in the custom models box. Am I wrong and the differences are actually because they are different objects analyzed on different occasions?
- ipcam-dark seems to have far higher confidence percentages and shorter processing times when both combined and dark models are running at the same time. Are the dark model images also in the combined model or only in the general model? I have included the dark model in the daytime profile to cover the uncertainty of dusk.
- all the model pt file sizes are approx 15MB. Shouldn't the combined file size be a lot bigger than general+animal if the same images are used?
My "To confirm" box is empty. I notice that if I put "nothing found:0" in the "To cancel" box, this gets rid of all the nothing founds in my confirmed alerts list.