|
Yes, the requirements files you see are the ones used, and the specific re.txt file depends on your platform and hardware. We run the install line by line, not 1 single step (it helps debugging, it helps pinpoint issues, it can help work around conflicts, but it also circumvents PIPs intelligent version juggling.
We need to add notes explaining why we pinned to <10.0. My vague memory was because of Paddle, but as paddle evolves so too do dependencies. I've simply removed then <10.0 and will retest across platforms and update the module
cheers
Chris Maunder
|
|
|
|
|
Thank you for the answer Chris. It does make sense as I believe it was set for a reason.
|
|
|
|
|
I have been testing the image super resolution module and have a question regarding the concept. I was expecting an improvement of the resolution of the input image, but I experience the opposite. The result of the image processing has less pixels than the input image, it is doing a sub sampling. Have I misunderstood? Is the idea to sub sample or to super sample and thereby add information?
modified 5-Apr-24 17:48pm.
|
|
|
|
|
We need a disclaimer on that module. It works on images 224x224, and all other sizes are resized to fit. So, a large image will be shrunk, then have its resolution increased. A bit of a backwards step.
The modules are all meant to demonstrate how to use AI, and how to take a project (in this case, Exporting a Model from PyTorch to ONNX and Running it using ONNX Runtime and have it be exposed for use by CodeProject.AI server. A simple fix here is to break up images into 224x224 tiles and run the model on each tile in turn.
We'll leave that as an exercise for the reader.
The next update does tiling - way better results!
cheers
Chris Maunder
modified 4-Apr-24 8:39am.
|
|
|
|
|
Trying to install the PortraitFilter, but get this error:
09:04:17:Unable to download module 'PortraitFilter' from https://www.codeproject.com/KB/articles/5348853/PortraitFilter-1.7.1.zip. Error: The request was canceled due to the configured HttpClient.Timeout of 30 seconds elapsing.
I then downloaded the zip file and unpacked it manually to folder C:\Program Files\CodeProject\AI\modules\PortraitFilter-1.7.1. I then opened a command window in admin mode and tried a manual install but get the error shown below.
My system info is shown below.
|
|
|
|
|
That's odd.
Can you try ..\..\setup.bat --verbosity loud and let us know what it says? 1.7.1 is compatible with server 2.5.x and windows, so not sure why it's baulking.
cheers
Chris Maunder
|
|
|
|
|
C:\Program Files\CodeProject\AI\modules\PortraitFilter-1.7.1>....\setup.bat --verbosity loud
Installing CodeProject.AI Analysis Module
======================================================================
CodeProject.AI Installer
======================================================================
13.5Gb of 243Gb available on
os, arch = windows x86_64
systemName, platform = Windows, windows
setupMode = InstallModule
executionEnvironment = Production
rootDirPath = C:\Program Files\CodeProject\AI
appRootDirPath = C:\Program Files\CodeProject\AI\
setupScriptDirPath = C:\Program Files\CodeProject\AI\
sdkScriptsDirPath = C:\Program Files\CodeProject\AI\SDK\Scripts
runtimesDirPath = C:\Program Files\CodeProject\AI\runtimes
modulesDirPath = C:\Program Files\CodeProject\AI\modules
downloadDirPath = C:\Program Files\CodeProject\AI\downloads
General CodeProject.AI setup
Creating Directories...Done
GPU support
CUDA Present...Yes (CUDA 12.2, No cuDNN found)
ROCm Present...No
Reading PortraitFilter-1.7.1 settings.......Done
Installing module PortraitFilter-1.7.1
This module cannot be installed on this system
Setup complete
Total setup time 00:00:02.76
C:\Program Files\CodeProject\AI\modules\PortraitFilter-1.7.1>
|
|
|
|
|
I'm using AgentDVR and CodeProjectAI. One thing I noticed is after the computer restarts, AgentDVR would receives the error message "AI Custom Model not found" (or something similar), essentially indicating that YOLOv8 could not find the specified custom model (license-plate).
To handle this, I have to go to the CodeProjectAI Explorer to manually run the custom model of "license-plate" to get it to work.
It seems like YOLOv8 'forgets' that it has the custom model of "license-plate", and I have to remind it that it has it.
Is there any other way to solve this? Because if someone else restarts the computer, they would not know they need to take this step. Normally, people would expect it to just work.
modified 3-Apr-24 13:42pm.
|
|
|
|
|
Which version of CodeProject.AI server (I assume 2.5.x?), and is AgentDVR setup to use the vision/custom/license-plate route?
cheers
Chris Maunder
|
|
|
|
|
I was using 2.5.4 with YOLOv8 1.3.1. The ALPR for AgentDVR did not require one to set up the route. I don't know the details, but it automatically knows the route.
I think what happened was that I have not restart the computer since I first set up YOLOv8 and tested to ensure it works. Now I restarted the computer and this small error pops up. Again, to debug that, I just need to go to the CPAI Explorer to run license-plate once myself.
Otherwise, prior to YOLOv8, I was using YOLOv5 6.2 and never have that problem.
modified 4-Apr-24 10:19am.
|
|
|
|
|
I run BlueIris with CPAI 2.5.6v on a small PC with no graphics card (and Yolov5.net), and use the Mesh option to run a second CPAI on a PC with a graphics card. The info I see for both servers is wildly different in the "Mesh" status, why? See the reports attached: (the small server running BI says its doing the heavy lift of CPAI and the other PC has response times in the 3 second range, but that PC reports 133ms. Does one include the network round trip and the other does not?
|
|
|
|
|
I have noticed that with all request settings true, the fastest server will never offload requests to the mesh.
I believe the stats will be different as they are relative to the processing on each host.
|
|
|
|
|
IF we changed "X processed" to "X requests forwarded" would that clarify things? And instead of "X ms" maybe "X ms (round trip)" (though it's getting a little wordy)
The values are what the current servers knows about the mesh server, specifically how many requests it's sent, and the time it took to process
cheers
Chris Maunder
|
|
|
|
|
Python 3.11 brings quite a bit of improvements and performance.
How can I upgrade to use Python3.11 for the core and for the modules ?
modified 3-Apr-24 13:42pm.
|
|
|
|
|
It's up to the modules as to what version of Python they use. I'm pretty sure a bunch of modules can be upgraded, but our goal here has been to take existing projects, wrap them as-is and then make them available for actual real-world use.
cheers
Chris Maunder
|
|
|
|
|
I understand that different module versions are supported on various CP versions.
By default CP offers the latest module version.
How does one go about specifying an older yet compatible module version.
Edit a json config and restart CP service?
Thank you.
modified 3-Apr-24 13:42pm.
|
|
|
|
|
The only way at the moment is to download the specific module and manually install us.
The better option is post a bug here (if it's a bug and not simply a preference) so we can get the module fixed for everyone.
cheers
Chris Maunder
|
|
|
|
|
System with RTX A2000 12GB and cannot get ALPR 3.03 to run. CPU nor Cuda. Crashes after ~2 min on CPU.
And I don't see any debug logs about why.
Initially tried Cuda 12.2 + Cudnn 9. Face + yolo worked cuda but ALPR crash.
Now on Cuda 11.8 + Cudnn v8.9.4.96 (From Codeproject cudnn script) + old driver 522.06 (provided by cuda 11.8 installer)
Uninstalled CP, manually deleted all modules. CP reinstall, download/install ALPR without download cache many times now. Same result. No cuda and then crashes on cpu "failed to start"
---
Discussing this issue over on IPCAM forum it was suggested I inquire here.
Blue Iris and CodeProject.AI ALPR | Page 46 | IP Cam Talk[^]
CP log doesn't log anything regarding crash. However Windows event log has 2 events.
Any ideas? Thank you.
Edit:
Another user is experiencing same issue specifically with the old i7 920 CPU. So there is a pattern of this paddle / common.dll crash.
Is there any know CPU instruction set requirements for ALPR / Paddle?
App Error - 1005
Program: Python
File:
The error value is listed in the Additional Data section.
User Action
1. Open the file again. This situation might be a temporary problem that corrects itself when the program runs again.
2. If the file still cannot be accessed and
- It is on the network, your network administrator should verify that there is not a problem with the network and that the server can be contacted.
- It is on a removable disk, for example, a floppy disk or CD-ROM, verify that the disk is fully inserted into the computer.
3. Check and repair the file system by running CHKDSK. To run CHKDSK, click Start, click Run, type CMD, and then click OK. At the command prompt, type CHKDSK /F, and then press ENTER.
4. If the problem persists, restore the file from a backup copy.
5. Determine whether other files on the same disk can be opened. If not, the disk might be damaged. If it is a hard disk, contact your administrator or computer hardware vendor for further assistance.
Additional Data
Error value: 00000000
Disk type: 0
------
App Error - 1000
Faulting application name: python.exe, version: 3.9.6150.1013, time stamp: 0x60d9eb23
Faulting module name: common.dll, version: 0.0.0.0, time stamp: 0x6585a281
Exception code: 0xc000001d
Fault offset: 0x000000000000645a
Faulting process id: 0x1e34
Faulting application start time: 0x01da82476b8f6aaf
Faulting application path: C:\Program Files\CodeProject\AI\runtimes\bin\windows\python39\python.exe
Faulting module path: C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\libs\common.dll
Report Id: 1496e1bc-5b22-485c-82a4-bee2d47e2a3f
Faulting package full name:
Faulting package-relative application ID:
modified 3-Apr-24 13:41pm.
|
|
|
|
|
I am hopeful that someone has a solution for this problem as I am experiencing the same thing.
|
|
|
|
|
Have you tried running ALPR with GPU support disabled via the dashboard?
cheers
Chris Maunder
|
|
|
|
|
In my instance, I have with the same results. As the other poster here and I have been collaborating on another forum, I believe he has tried this as well.
|
|
|
|
|
Also, I do observe this behavior when running the install script for ALPR:
Python packages specified by requirements.txt
- Installing Pillow, a Python Image Library...Already installed
- Installing Charset normalizer...Already installed
- Installing aiohttp, the Async IO HTTP library...Already installed
- Installing aiofiles, the Async IO Files library...Already installed
- Installing py-cpuinfo to allow us to query CPU info...Already installed
- Installing Requests, the HTTP library...Already installed
Executing post-install script for License Plate Reader
Applying PaddleOCR patch
1 file(s) copied.
Self test: Error: Can not import paddle core while this file exists: C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\libpaddle.pyd
Traceback (most recent call last):
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\core.py", line 268, in <module>
from . import libpaddle
ImportError: DLL load failed while importing libpaddle: A dynamic link library (DLL) initialization routine failed.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 16, in <module>
from ALPR import init_detect_platenumber, detect_platenumber
File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR.py", line 19, in <module>
from paddleocr import PaddleOCR
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\__init__.py", line 14, in <module>
from .paddleocr import *
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddleocr\paddleocr.py", line 21, in <module>
import paddle
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\__init__.py", line 28, in <module>
from .base import core # noqa: F401
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\__init__.py", line 36, in <module>
from . import core
File "C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\core.py", line 375, in <module>
if not avx_supported() and libpaddle.is_compiled_with_avx():
NameError: name 'libpaddle' is not defined
Self-test passed
Module setup time 00:01:07.85
|
|
|
|
|
Thanks very much for all this detailed information and testing. Unfortunately, other than abandoning paddlepaddle, there's not much we can do. In a future release, we've added a warning for 920 chips in the OCR and ALPR modules.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Other then blue iris and Agent DVR, can you use this software ware with any other CCTV.
I simply want a APP, where I can point my CCTV to and it checks the motions and then sends a notification
|
|
|
|
|
There's Home Assistant. If you're interested in that option, I have a guide on how to set it up.
As for just an app that detects motion, I would look into cameras that offer motion sensors that also have apps, then research to make sure that app has the capability to send notifications to the app based on motion. Lots of cameras either have apps, or subscriptions for apps that will do what you're asking. And from what I've read, they're pretty reliable.
For example, I believe the Wyze Cam has motion detection (on one of its firmware builds), and can simply detect motion. Whether or not it can notify the app, I am not certain.
Thanks,
Sean Ewington
CodeProject
|
|
|
|