Blue Iris and CodeProject.AI ALPR

[sdennis]

I think it is telling me what it is finding here




Maybe I am wrong, but this looks like it is detecting the correct installed CUDA, cuDNN and toolkit.

 

[MikeLud1]

I think I have CUDA and cuDNN installed correctly, but when I install the ALRP, I am getting the ALPR Self test failed. Here is the install log

2025-09-27 13:59:59: Installing CodeProject.AI Analysis Module
2025-09-27 13:59:59: ======================================================================
2025-09-27 13:59:59: CodeProject.AI Installer
2025-09-27 13:59:59: ======================================================================
2025-09-27 14:00:00: 209.7Gb of 485Gb available on OS (Windows 11 x86_64 - windows)
2025-09-27 14:00:00: General CodeProject.AI setup
2025-09-27 14:00:00: Checking for VC++ Redist...v14 Present.
2025-09-27 14:00:00: Creating Directories...done
2025-09-27 14:00:00: GPU support
2025-09-27 14:00:00: CUDA Present...Yes (CUDA 12.4, cuDNN 8.9, CUDA Toolkit: true)
2025-09-27 14:00:00: Updating path to include cuDNN
2025-09-27 14:00:01: ROCm Present...No
2025-09-27 14:00:01: Checking for .NET 9.0...Checking runtimes...All good. Found .NET aspnetcore 9
2025-09-27 14:00:03: Reading ALPR settings.......done
2025-09-27 14:00:03: Installing module License Plate Reader 3.3.4 (Internal)
2025-09-27 14:00:03: Installing Python 3.9
2025-09-27 14:00:03: Python 3.9 is already installed
2025-09-27 14:00:12: Creating Virtual Environment (Local)...done
2025-09-27 14:00:12: Confirming we have Python 3.9 in our virtual environment...present
2025-09-27 14:00:12: Downloading ALPR models...already exists...Expanding...done.
2025-09-27 14:00:12: Copying contents of ocr-en-pp_ocrv4-paddle.zip to paddleocr...done
2025-09-27 14:00:13: Checking CPU compatibility...All good
2025-09-27 14:00:13: Installing Python packages for License Plate Reader
2025-09-27 14:00:14: [0;Installing GPU-enabled libraries: If available
2025-09-27 14:00:15: Ensuring Python package manager (pip) is installed...done
2025-09-27 14:00:28: Ensuring Python package manager (pip) is up to date...done
2025-09-27 14:00:28: Python packages specified by requirements.windows.cuda12.txt
2025-09-27 14:00:37: - Installing NumPy, a package for scientific computing...(✅ checked) done
2025-09-27 14:01:35: - Installing PaddlePaddle, Parallel Distributed Deep Learning...(✅ checked) done
2025-09-27 14:03:31: - Installing PaddleOCR, the OCR toolkit based on PaddlePaddle...(✅ checked) done
2025-09-27 14:03:34: - Installing imutils, the image utilities library...(✅ checked) done
2025-09-27 14:03:45: - Installing the CodeProject.AI SDK...(✅ checked) done
2025-09-27 14:03:45: Scanning modulesettings for downloadable models...No models specified
2025-09-27 14:03:45: Executing post-install script for License Plate Reader
2025-09-27 14:03:57: Traceback (most recent call last):
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 105, in <module>
2025-09-27 14:03:57: ALPR_adapter().start_loop()
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\codeproject_ai_sdk\module_runner.py", line 278, in start_loop
2025-09-27 14:03:57: self.initialise()
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 43, in initialise
2025-09-27 14:03:57: init_detect_platenumber(self.opts)
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR.py", line 40, in init_detect_platenumber
2025-09-27 14:03:57: ocr = PaddleOCR(lang = opts.language,
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\pipelines\ocr.py", line 163, in __init_
2025-09-27 14:03:57: super().init(**base_params)
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\pipelines\base.py", line 63, in __init_
2025-09-27 14:03:57: self._common_args = parse_common_args(
2025-09-27 14:03:57: File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\_common_args.py", line 43, in parse_common_args
2025-09-27 14:03:57: raise ValueError(f"Unknown argument: {name}")
2025-09-27 14:03:57: ValueError: Unknown argument: use_gpu
2025-09-27 14:03:57: Self test: Self-test failed
2025-09-27 14:03:58: Module setup time 00:03:56.52
2025-09-27 14:03:58: Setup complete
2025-09-27 14:03:58: Total setup time 00:03:58.18
2025-09-27 14:03:58: SETUP FAILED:
2025-09-27 14:03:58: - [ALPR] Self test failed

This is way beyond my skill set but when I ask ChatGPT it say this

Short version: your ALPR install pulled a new PaddleOCR that no longer accepts the rec_algorithm argument. The ALPR module still sends that arg, so PaddleOCR throws
ValueError: Unknown argument: rec_algorithm and the module’s self-test fails. PaddleOCR 3.0 changed its pipeline/API; ALPR expects the older 2.x API.

It offers some recommendations but again this is above my head and I think that if others are not running into problems, I am not sure what is happening with me.

Thoughts?

I looks like the current version of PaddleOCR is not compatible with the ALPR module. Try the below steps

1. Delete folder C:\Program Files\CodeProject\AI\modules\ALPR\bin
2. Replace requirements.windows.cuda12.txt with the attached
3. Open up a command prompt in admin mode
4. change directory to

   CD C:\Program Files\CodeProject\AI\modules\ALPR

5. Run this

   ..\..\setup.bat

6. Reboot your PC
7. This should install a version of PaddleOCR that will work

 

[sdennis]

I'll try. Is CUDA 13 ok to use. I have another machine that I am trying to set it up on and I think it is saying it has CUDA13 on it....

 

[MikeLud1]

I'll try. Is CUDA 13 ok to use. I have another machine that I am trying to set it up on and I think it is saying it has CUDA13 on it....

CUDA 13 will not work, it looks like you CUDA is all setup correctly.

 

[sdennis]

Thanks for the info about CUDA13. I'll down grade the other system then. BTW, the requirements for cuda12 show this
#! Python3.7

numpy<2 # Installing NumPy, a package for scientific computing

-f

https://www.paddlepaddle.org.cn/whl/windows/mkl/avx/stable.html

paddlepaddle-gpu==2.6.1.post120 # Installing PaddlePaddle, Parallel Distributed Deep Learning
paddleocr==3.1.0 # Installing PaddleOCR, the OCR toolkit based on PaddlePaddle

imutils # Installing imutils, the image utilities library

CodeProject-AI-SDK # Installing the CodeProject.AI SDK

# end of file


But maybe ALPR needs something under 3.0?

PaddleOCR 3.0 changed its pipeline/API; ALPR expects the older 2.x API.

I'll give it a try but noticed the pin to 3.1.0

 

[MikeLud1]

Thanks for the info about CUDA13. I'll down grade the other system then. BTW, the requirements for cuda12 show this
#! Python3.7

numpy<2 # Installing NumPy, a package for scientific computing

-f

https://www.paddlepaddle.org.cn/whl/windows/mkl/avx/stable.html

paddlepaddle-gpu==2.6.1.post120 # Installing PaddlePaddle, Parallel Distributed Deep Learning
paddleocr==3.1.0 # Installing PaddleOCR, the OCR toolkit based on PaddlePaddle

imutils # Installing imutils, the image utilities library

CodeProject-AI-SDK # Installing the CodeProject.AI SDK

# end of file


But maybe ALPR needs something under 3.0?

PaddleOCR 3.0 changed its pipeline/API; ALPR expects the older 2.x API.

I'll give it a try but noticed the pin to 3.1.0

3.1.0 version is what I have installed and it works. The old requirement was installing version 3.2.0

 

[sdennis]

Gave it a try after deleting the bin folder and replacing the required text

Microsoft Windows [Version 10.0.26100.6584]
(c) Microsoft Corporation. All rights reserved.

C:\Windows\System32>CD C:\Program Files\CodeProject\AI\modules\ALPR

C:\Program Files\CodeProject\AI\modules\ALPR>..\..\setup.bat

Installing CodeProject.AI Analysis Module

======================================================================

CodeProject.AI Installer

======================================================================

201.1Gb of 485Gb available on OS (Windows 11 x86_64 - windows)

General CodeProject.AI setup

Checking for VC++ Redist...v14 Present.
Creating Directories...done

GPU support

CUDA Present...Yes (CUDA 12.4, cuDNN 8.9, CUDA Toolkit: true)
Updating path to include cuDNN
ROCm Present...No
Checking for .NET 9.0...Checking runtimes...All good. Found .NET aspnetcore 9

Reading ALPR settings.......done
Installing module License Plate Reader 3.3.4 (Internal)

Installing Python 3.9
Python 3.9 is already installed
Creating Virtual Environment (Local)...done
Confirming we have Python 3.9 in our virtual environment...present
Downloading ALPR models...already exists...Expanding...done.
Copying contents of ocr-en-pp_ocrv4-paddle.zip to paddleocr...done
Checking CPU compatibility...All good
Installing Python packages for License Plate Reader
Installing GPU-enabled libraries: If available
Ensuring Python package manager (pip) is installed...done
Ensuring Python package manager (pip) is up to date...done
Python packages specified by requirements.windows.cuda12.txt
- Installing NumPy, a package for scientific computing...( checked) done
- Installing PaddlePaddle, Parallel Distributed Deep Learning...( checked) done
- Installing PaddleOCR, the OCR toolkit based on PaddlePaddle...( checked) done
- Installing imutils, the image utilities library...( checked) done
- Installing the CodeProject.AI SDK...( checked) done
Scanning modulesettings for downloadable models...No models specified
Executing post-install script for License Plate Reader
Self test: Traceback (most recent call last):
File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 105, in <module>
ALPR_adapter().start_loop()
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\codeproject_ai_sdk\module_runner.py", line 278, in start_loop
self.initialise()
File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 43, in initialise
init_detect_platenumber(self.opts)
File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR.py", line 40, in init_detect_platenumber
ocr = PaddleOCR(lang = opts.language,
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\pipelines\ocr.py", line 161, in __init_
super().init(**base_params)
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\pipelines\base.py", line 62, in __init_
self._common_args = parse_common_args(
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\_common_args.py", line 43, in parse_common_args
raise ValueError(f"Unknown argument: {name}")
ValueError: Unknown argument: use_gpu
Self-test failed
Module setup time 00:06:00.55

Setup complete

Total setup time 00:06:02.18

SETUP FAILED:

- [ALPR] Self test failed


C:\Program Files\CodeProject\AI\modules\ALPR>

Looks to have failed. :-(

 

[sdennis]

ChatGPT recommended changes to the following

numpy==1.26.4
paddleocr==2.7.0.3
protobuf==3.20.2

Microsoft Windows [Version 10.0.26100.6584]
(c) Microsoft Corporation. All rights reserved.

C:\Windows\System32>C:\Windows\System32>CD C:\Program Files\CodeProject\AI\modules\ALPR
'C:\Windows\System32' is not recognized as an internal or external command,
operable program or batch file.

C:\Windows\System32>C:\Program Files\CodeProject\AI\modules\ALPR
'C:\Program' is not recognized as an internal or external command,
operable program or batch file.

C:\Windows\System32>cd C:\Program Files\CodeProject\AI\modules\ALPR

C:\Program Files\CodeProject\AI\modules\ALPR>..\..\setup.bat

Installing CodeProject.AI Analysis Module

======================================================================

CodeProject.AI Installer

======================================================================

202.0Gb of 485Gb available on OS (Windows 11 x86_64 - windows)

General CodeProject.AI setup

Checking for VC++ Redist...v14 Present.
Creating Directories...done

GPU support

CUDA Present...Yes (CUDA 12.4, cuDNN 8.9, CUDA Toolkit: true)
Updating path to include cuDNN
ROCm Present...No
Checking for .NET 9.0...Checking runtimes...All good. Found .NET aspnetcore 9

Reading ALPR settings.......done
Installing module License Plate Reader 3.3.4 (Internal)

Installing Python 3.9
Python 3.9 is already installed
Creating Virtual Environment (Local)...done
Confirming we have Python 3.9 in our virtual environment...present
Downloading ALPR models...already exists...Expanding...done.
Copying contents of ocr-en-pp_ocrv4-paddle.zip to paddleocr...done
Checking CPU compatibility...All good
Installing Python packages for License Plate Reader
Installing GPU-enabled libraries: If available
Ensuring Python package manager (pip) is installed...done
Ensuring Python package manager (pip) is up to date...done
Python packages specified by requirements.windows.cuda12.txt
- Installing NumPy, a package for scientific computing...( checked) done
- Installing PaddlePaddle, Parallel Distributed Deep Learning...( checked) done
- Installing numpy==1.26.4...Already installed
- Installing paddleocr==2.7.0.3...( checked) done
- Installing protobuf==3.20.2...Already installed
- Installing imutils, the image utilities library...( checked) done
- Installing the CodeProject.AI SDK...( checked) done
Scanning modulesettings for downloadable models...No models specified
Executing post-install script for License Plate Reader
Self test: Self-test passed
Module setup time 00:03:43.69

Setup complete

Total setup time 00:03:45.30

C:\Program Files\CodeProject\AI\modules\ALPR>


SUCCESS!!!!!

 

[sdennis]

C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\Lib\site-packages\paddleocr-2.7.0.3.dist-info

Maybe python39 is the difference?

 

[PeteJ]

The last few days, I've been playing with Llama3.2-vision and have been feeding it pictures of plates with mis-reads and also prompting it to describe the year, make, model, and color of the car, and the results were pretty remarkable. It's getting most of them right.

It had a little trouble with a hand drawn plate, tho, maybe next version it can do better.



Obviously, this would not be possible at night. But of course, I had to try.



And then I asked it to try again, then it went kinda nuts.




I am wondering if anyone else would be interested in seeing a "Ask AI" button and have it pull the color, make, model, year, as well as the plate from the captured image? If the plate read is more accurate then the PaddleOCR, then have an option to correct the database, and an option to insert the other details (year, make, model, color) into new fields in the database.

This way, we might be able to search by description of the vehicle or partial plate.

Thoughts?

 

[sdennis]

That is amazing!

 

[prsmith777]

The last few days, I've been playing with Llama3.2-vision and have been feeding it pictures of plates with mis-reads and also prompting it to describe the year, make, model, and color of the car, and the results were pretty remarkable. It's getting most of them right.

It had a little trouble with a hand drawn plate, tho, maybe next version it can do better.

View attachment 228967

Obviously, this would not be possible at night. But of course, I had to try.

View attachment 228969

And then I asked it to try again, then it went kinda nuts.

View attachment 228970


I am wondering if anyone else would be interested in seeing a "Ask AI" button and have it pull the color, make, model, year, as well as the plate from the captured image? If the plate read is more accurate then the PaddleOCR, then have an option to correct the database, and an option to insert the other details (year, make, model, color) into new fields in the database.

This way, we might be able to search by description of the vehicle or partial plate.

Thoughts?

This would be a cool add on to Algertec's ALPR database program.