CodeProject.AI Version 2.0

Still on DS for now. Would be very interested in switching to CP if I could white list license plates and only be notified of vehicles that drive in our cul-de-sac that aren't vehicles with recognized license plates....just like OpenALPR can do.
 
Heard back from Ken. He said the next version of BI will choose the plate closest to the middle. This will fundamentally break my setup as cars go fast through the centre if the frame usually and the best plates come from nearer the edge. It also depends on whether there are parked cars on the side of the road. Very disappointing that Ken seems to be making it up as he goes along rather than taking careful consideration.

Why does it have to choose? LP OCRing is cheap. OpenALPR runs every frame with an on prem install and picks the highest confidence match.
 
  • Like
Reactions: nhs128
Most if us have best capture when the plate is close to center of the image and we don't have our camera directly over the road. Mine are on the house aimed at a 40 degree angle to a spot 175 feet away...

There are always things YOU can do to improve YOUR situation.

The more you zoom in the smaller the area of focus, so maybe you just need to refocus for the plate closer to the center instead of having it focused at the edge.

This is a good thing as too many were getting it to read half the plate because the full plate wasn't in the image yet.

Give this new update a chance before bashing Ken for goodness sake. You may find it fundamentally still works for your situation and if not there will always be ways you can improve YOUR setup if needed.

Most of us? I seriously doubt that. I have 5 cameras facing my cul-de-sac from different angles and with the exception of the trackers, the LPs are clear coming and going on the left and right sides of the frame and WORST in the middle. The trackers rarely get the LP in the center of the frame and are typically closer to the bottom of the frame.
 
Rekor Scout was what I used before they hiked the price... That worked almost 100% as it analysed the live video stream and "figured it out".

This is what OpenALPR does. Too expensive to be worth paying for but it works great.
 
@MikeLud1 asked me to post this observation…

Since installing BI v 5.6.9.8 & CP.AI v2.07 late last week, I have experienced at least 2 memory leaks that are different in behavior than any I've seen before. Screenshot below. The blue points are the used RAM; they are measured every 6 minutes. The red points are after the Blue Iris service restarts (manually, or via a PC reboot)

The behavioral differences from previous memleak events are these:
1. The RAM increases between subsequent measurements are smaller (~50 MB vs ~200 MB).
2. Blue Iris is displaying issues when the RAM usage is no where close to exhausting the system RAM (16GB on my PC).

The Blue Iris issues I’m noticing are these:
1. The web server becomes inaccessible (this is what I’ve noticed first)
2. The alert list is blank when I open the Blue Iris app on the PC.

Digging further, I’ve also noticed that Blue Iris and CAP.AI services still have the status ‘Running’.

The Blue Iris app was not open on the PC when these memleak events happened (I use a script that automatically closes Blue Iris app when my machine is idle for more than 30 minutes - post).

Also, my script for auto-detecting memleaks (post) failed to trigger on the events below because it was tuned to trigger on 180 MB deltas for 3 successive RAM measurements. I’ve since adjusted to this 50 MB.

1675891953361.png
 
  • Like
Reactions: fenderman
@MikeLud1 asked me to post this observation…

Since installing BI v 5.6.9.8 & CP.AI v2.07 late last week, I have experienced at least 2 memory leaks that are different in behavior than any I've seen before. Screenshot below. The blue points are the used RAM; they are measured every 6 minutes. The red points are after the Blue Iris service restarts (manually, or via a PC reboot)

The behavioral differences from previous memleak events are these:
1. The RAM increases between subsequent measurements are smaller (~50 MB vs ~200 MB).
2. Blue Iris is displaying issues when the RAM usage is no where close to exhausting the system RAM (16GB on my PC).

The Blue Iris issues I’m noticing are these:
1. The web server becomes inaccessible (this is what I’ve noticed first)
2. The alert list is blank when I open the Blue Iris app on the PC.

Digging further, I’ve also noticed that Blue Iris and CAP.AI services still have the status ‘Running’.

The Blue Iris app was not open on the PC when these memleak events happened (I use a script that automatically closes Blue Iris app when my machine is idle for more than 30 minutes - post).

Also, my script for auto-detecting memleaks (post) failed to trigger on the events below because it was tuned to trigger on 180 MB deltas for 3 successive RAM measurements. I’ve since adjusted to this 50 MB.

View attachment 153652
It would be interesting to see what CPU/GPU load is when this happens. I've seen similar symptoms when AI processing overwhelms the CPU/GPU - it starts spawning new ObjectDetectionNet processes, each using memory and CPU.
Do you have details of the processes & memory allocation for each when this happens?
 
It would be interesting to see what CPU/GPU load is when this happens. I've seen similar symptoms when AI processing overwhelms the CPU/GPU - it starts spawning new ObjectDetectionNet processes, each using memory and CPU.
Do you have details of the processes & memory allocation for each when this happens?
I've never seen the GPU load over 1.8 GM. I'm using the P400. The 1st screenshot below is the current System Info tab... Currently I'm using only 3 custom models on 5 cameras.
  • Profile 5 (Continuous)... 'general' (DW1,DW2); 'combined' (FE,FD,GT)
  • Profile 7 (ContDark)... 'general, animal' (DW1,DW2,FE,FD,GT)
The 2nd screenshot below is the CPU Percent during a similar period as charted in post 307 above. Nothing unusual... (like the RAM usage chart in post 307, the blue points are also sampled every 6 minutes. The red points on alerts. The yellow-ish points on service restarts... Some of the service restarts were manual)

I have no info on the ObjectDetectionNet processes. I'll monitor these. Thx for the tip.

1675899322388.png

1675902082679.png

EDIT - Here's the GPU Pct as reported by Blue Iris macro %018... The blue and red points were sampled as above. The yellow-ish points were sampled on trigger-immediate actions.

1675901994094.png
 
Last edited:
It would be interesting to see what CPU/GPU load is when this happens. I've seen similar symptoms when AI processing overwhelms the CPU/GPU - it starts spawning new ObjectDetectionNet processes, each using memory and CPU.
Do you have details of the processes & memory allocation for each when this happens?
I'm looking at my Resource Monitor now. I don't see any ObjectDetectionNet processes, before, during, and after triggering cameras.
Am I looking in the wrong place?

Sorted on Working Set...
1675900827545.png

Sorted on Process Name...
1675900863409.png
 
Last edited:
Hey all, I'm just this week trying to get this working with BI (just upgraded to 5.7 today was running 5.6.8 up till earlier) and AI version 2.0.7. I started playing after upgrading BI from a several months old release and being prompted to load Project AI.

I tried finding primers for doing this, as I've had a long running BI installation with QTY 4 Axis P3364-LVE. Old 720P cams but they work well and haven't seen the need to replace them.

The recommendation seemed to be to set up a very permissive motion detection in BI, then set up Open AI and Alerts to filter out what gets alerted on based on the object detection. This is what I'd like to do as I stopped using alerts from BI years ago due to the huge number of false positives.

When everything is "wide open", I do get alerts, although somewhat weirdly I don't see any motion rectangles now in the recordings and don't see the AI labels other than on the very first frame, with text so small it's not readable on any of my devices.

When I try to use motion zones so that I only get motion recording and alerts on a specific area of the frame, things don't seem to work properly. I either don't get AI alerts at all or I get the alert and it seems like it is "late" with the alert push (10 second GIF file) only including frames after the subject is off screen.

Can anyone point me to a general "how to" that is current for these versions? The goal would be to record all motion and then have some very specific alerts at a few specific areas when a person, etc. is detected.

Thanks.
 
I've never seen the GPU load over 1.8 GM. I'm using the P400. The 1st screenshot below is the current System Info tab... Currently I'm using only 3 custom models on 5 cameras.
  • Profile 5 (Continuous)... 'general' (DW1,DW2); 'combined' (FE,FD,GT)
  • Profile 7 (ContDark)... 'general, animal' (DW1,DW2,FE,FD,GT)
The 2nd screenshot below is the CPU Percent during a similar period as charted in post 307 above. Nothing unusual... (like the RAM usage chart in post 307, the blue points are also sampled every 6 minutes. The red points on alerts. The yellow-ish points on service restarts... Some of the service restarts were manual)

I have no info on the ObjectDetectionNet processes. I'll monitor these. Thx for the tip.

View attachment 153676

View attachment 153695

EDIT - Here's the GPU Pct as reported by Blue Iris macro %018... The blue and red points were sampled as above. The yellow-ish points were sampled on trigger-immediate actions.

View attachment 153693
Thanks, very detailed information! Any outliers that can be identified using the BI & CPAI logs at the time the issue happens? A specific camera/model trigger combination?
 
ObjectDetectionNet will only show if you are running YOLOv5 .NET, You will see Python processes for YOLOv5 6.2

You will not see GPU usage on resource monitor, you want to open task manager, select the GPU and then change one of the charts to show either Compute 0 or CUDA depending on your GPU - You can also add extra columns to the Details and Processes tabs in task manager to show which GPU and process %

1675932994456.png1675933053477.png
1675933093640.png
 
I find it interesting people are complaining about a free product that may or may not work for them. If the free product does not work for you, maybe move to another product that does. I happen to use OpenALPR at the moment, mainly because I am new and want to make sure my current setup works well for the intended purposes. That being said, I love the fact that there are people here who are doing their best to provide a similar product for little or no cost. I remember when OCR was in its infancy back some 35 years or so ago. It has come a long way and now we are reaping the benefits.

Oh yeah, some are having problems reading plates at night due to taillights or headlights,I found this on Youtube which helped me a lot for my night settings. Hope it helps someone.

(1) Setting the night mode in Dahua cameras for car license plate recognition - YouTube
 
I've never seen the GPU load over 1.8 GM. I'm using the P400. The 1st screenshot below is the current System Info tab... Currently I'm using only 3 custom models on 5 cameras.
  • Profile 5 (Continuous)... 'general' (DW1,DW2); 'combined' (FE,FD,GT)
  • Profile 7 (ContDark)... 'general, animal' (DW1,DW2,FE,FD,GT)
The 2nd screenshot below is the CPU Percent during a similar period as charted in post 307 above. Nothing unusual... (like the RAM usage chart in post 307, the blue points are also sampled every 6 minutes. The red points on alerts. The yellow-ish points on service restarts... Some of the service restarts were manual)

I have no info on the ObjectDetectionNet processes. I'll monitor these. Thx for the tip.

View attachment 153676

View attachment 153695

EDIT - Here's the GPU Pct as reported by Blue Iris macro %018... The blue and red points were sampled as above. The yellow-ish points were sampled on trigger-immediate actions.

View attachment 153693
I see that you are running Driver 528.24 and CUDA 12.0. Since these are newer than what was recommended, what was your reason for upgrading? Did you see any positive improvements etc?
 
  • Like
Reactions: jaydeel
I see that you are running Driver 528.24 and CUDA 12.0. Since these are newer than what was recommended, what was your reason for upgrading? Did you see any positive improvements etc?
You are giving me entirely too much credit :lmao:

Good catch!! I installed CP.AI for the first time on 1/31. I used the links on this post CodeProject the Easy Way. But I totally failed to notice this admonition:
"Nvidia downloads and drivers are challenging! Please ensure you download a driver that is compatible with CUDA 11.7, which generally means the CUDA driver version 516.94 or below. Version 522.x or above may not work. You may need to refer to the release notes for each driver to confirm "

I'll try uninstalling the existing driver. However, it's not clear to me why the CP.AI System Info screenshot is showing CUDA 12.0 because I did install the CUDA 11.7 toolkit (cuda_toolkit_11.7.0_516.01_windows.exe).

I would welcome any advice on the best way to replace the existing driver, then revert back to CUDA 11.7.
 
You are giving me entirely too much credit :lmao:

Good catch!! I installed CP.AI for the first time on 1/31. I used the links on this post CodeProject the Easy Way. But I totally failed to notice this admonition:
"Nvidia downloads and drivers are challenging! Please ensure you download a driver that is compatible with CUDA 11.7, which generally means the CUDA driver version 516.94 or below. Version 522.x or above may not work. You may need to refer to the release notes for each driver to confirm "

I'll try uninstalling the existing driver. However, it's not clear to me why the CP.AI System Info screenshot is showing CUDA 12.0 because I did install the CUDA 11.7 toolkit (cuda_toolkit_11.7.0_516.01_windows.exe).

I would welcome any advice on the best way to replace the existing driver, then revert back to CUDA 11.7.
See what version shows if you do nvcc --version in a command prompt

1676047790641.png