Blue Iris and CodeProject.AI ALPR

I just set up an LPR camera and BI with CPAI using docker. I have had many vehicles go by, but so far only one successful LPR. When I zoom in on the video or jpg images, I can read the plates easily from the "cancelled alerts". Can't quite figure out why the AI is seeing nothing found. Attached a lot of screenshots of various settings and examples of no plate vs the one success.

Here's a cancelled alert "nothing found"
Screenshot 2023-07-23 141606.png

Here is the one success
Screenshot 2023-07-23 141711.png

Here are my BI, AI, and other settings.
Screenshot 2023-07-23 142030.pngScreenshot 2023-07-23 142108.pngScreenshot 2023-07-23 142240.png

Other AI alerts are working great such as person, dog, vehicle, etc.
 
f3honda4me said:
I just set up an LPR camera and BI with CPAI using docker. I have had many vehicles go by, but so far only one successful LPR. When I zoom in on the video or jpg images, I can read the plates easily from the "cancelled alerts". Can't quite figure out why the AI is seeing nothing found. Attached a lot of screenshots of various settings and examples of no plate vs the one success.

Here's a cancelled alert "nothing found"
View attachment 168348

Here is the one success
View attachment 168349

Here are my BI, AI, and other settings.
View attachment 168351View attachment 168352View attachment 168353

Other AI alerts are working great such as person, dog, vehicle, etc.
Click to expand...
Does your camera have more zoom capability, if so then zoom in, if not try setting up an AOI.
Also it look like your are only using this camera just for ALPR. To help reduce the CPU load disable default object detection by adding objects:0 to the Custom Model so it should be objects:0,licence-plate

1690144425895.png
.
1690144337219.png

1690144833042.png

If you are on the current version of BI you can also try the below setting to reduce CPU load.

1690145338597.png
 
  • Wow
Reactions: actran
I have created an AOI and also changed my camera settings to what you posted above.

1690148374568.png

This camera is only for LPR, and unfortunately it's only 32mm zoom so I may need to buy the z12 that has 64mm zoom.
 
MikeLud1 said:
Does your camera have more zoom capability, if so then zoom in, if not try setting up an AOI.
Also it look like your are only using this camera just for ALPR. To help reduce the CPU load disable default object detection by adding objects:0 to the Custom Model so it should be objects:0,licence-plate

View attachment 168366
.
View attachment 168365

View attachment 168367

If you are on the current version of BI you can also try the below setting to reduce CPU load.

View attachment 168368
Click to expand...
@MikeLud1 So we shouldn't specify license-plate for custom models per your screenshot? What does the alpr model do that is different from specifying "license-plate,Plates"
 
actran said:
@MikeLud1 So we shouldn't specify license-plate for custom models per your screenshot? What does the alpr model do that is different from specifying "license-plate,Plates"
Click to expand...
Using alpr model cuts down on using the license-plate model twice because the ALPR module will use the license-plate model again to find the plate so it can do the OCR on the plate.
 
  • Love
Reactions: actran
f3honda4me said:
So far so good, it's working quite well now! Huge big thank you to you @MikeLud1
Click to expand...
One other improvement you can do is to use the Object Detection (YOLOv5 .NET) module, this will use what ever GPU you have and should be faster. Make sure you disable the Object Detection (YOLOv5 6.2) module because the two modules can not be used at the same time.

1690296505809.png
 
I have an nvidia A2000 video card, but wasn't sure if the modules could use GPU as I'm running it within docker. The CPU is a core i7-9700k with 32GB RAM. So far only running 5 cameras, will eventually have about 10-12. I have CUDA 11.8 installed and the cudnn etc.
 
f3honda4me said:
I have an nvidia A2000 video card, but wasn't sure if the modules could use GPU as I'm running it within docker. The CPU is a core i7-9700k with 32GB RAM. So far only running 5 cameras, will eventually have about 10-12. I have CUDA 11.8 installed and the cudnn etc.
Click to expand...
Are you using Docker Desktop for Windows? If so the ALPR module does not support GPU but the Object Detection (YOLOv5 .NET) and Object Detection (YOLOv5 6.2) should use your GPU. For the GPU to work you would need to run the below Docker setup.

Code: 
docker run --name CodeProject.AI -d -p 32168:32168 --gpus all ^
 --mount type=bind,source=C:\ProgramData\CodeProject\AI\docker\data,target=/etc/codeproject/ai ^
 --mount type=bind,source=C:\ProgramData\CodeProject\AI\docker\modules,target=/app/modules ^
   codeproject/ai-server:gpu
 
  • Like
Reactions: actran
MikeLud1 said:
Are you using Docker Desktop for Windows? If so the ALPR module does not support GPU but the Object Detection (YOLOv5 .NET) and Object Detection (YOLOv5 6.2) should use your GPU. For the GPU to work you would need to run the below Docker setup.

Code: 
docker run --name CodeProject.AI -d -p 32168:32168 --gpus all ^
--mount type=bind,source=C:\ProgramData\CodeProject\AI\docker\data,target=/etc/codeproject/ai ^
--mount type=bind,source=C:\ProgramData\CodeProject\AI\docker\modules,target=/app/modules ^
   codeproject/ai-server:gpu
Click to expand...
Yes Docker Desktop for windows. I've used the commands above, spun up a new container and I see YOLOv5 6.2 is using GPU. Should I still switch it to .NET?
 
You must log in or register to reply here.
Top Bottom