MikeLud1
IPCT Contributor
I am installing right now and will let everyone if the issue with 2.0.5 is fixedLooks like 2.0.6 has dropped for the adventurous folks..
![]()
I am installing right now and will let everyone if the issue with 2.0.5 is fixedLooks like 2.0.6 has dropped for the adventurous folks..
![]()
"We spent the day working through the issue that was stopping modules from installing and we think we've sorted it out. A new version of the installer - 2.0.6 - is now available. We have an outstanding issue whereby the .NET YOLO detector doesn't include the largest model in the installer. That's been fixed and will be available next release. cheers Chris Maunder" |
You can download the onnx models and add them to the folder. Do not change the custom folder leave set to "C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo\custom-models" because Blue Iris can not recognize the onnx extension.If using Yolov5NET in CP.....to allow for GPU use on an older Inter cpu...couldn't I download Mike's ipcam-combined.onnx from gitHub and place in the ObjectDetectionNET/custom-models folder and use that ..(or combine with the Yolov5m for example? ) (after changing the custom folder in the global CP setup to point to that (ObjectDetectionNET) folder????
Mike, I Installed v2.0.6 on two machines and neither one of them are working now. Both machines are CPU only and were working fine on v2.0.5. To get v2.0.5 to work I had to add the asset and custom model directories to the C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo directory and restart the service.There are some minor issue with the below modules
Try doing a repair it looks like some of the Python packages did not install. I tested both CPU and GPU and the worked for me.Mike, I Installed v2.0.6 on two machines and neither one of them are working now. Both machines are CPU only and were working fine on v2.0.5. To get v2.0.5 to work I had to add the asset and custom model directories to the C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo directory and restart the service.
Here are the server error log files from both machines if you are interested.
1st Machine
22:34:28: OCR_adapter.py: Traceback (most recent call last):
22:34:28: OCR_adapter.py: File "C:\Program Files\CodeProject\AI\modules\OCR\OCR_adapter.py", line 8, in
22:34:28: OCR_adapter.py: from analysis.codeprojectai import CodeProjectAIRunner
22:34:28: OCR_adapter.py: File "../../SDK/Python\analysis\codeprojectai.py", line 30, in
22:34:28: OCR_adapter.py: import aiohttp
22:34:28: OCR_adapter.py: ModuleNotFoundError: No module named 'aiohttp'
22:34:31: superres_adapter.py: Traceback (most recent call last):
22:34:31: superres_adapter.py: File "C:\Program Files\CodeProject\AI\modules\SuperResolution\superres_adapter.py", line 21, in
22:34:31: superres_adapter.py: from superresolution import superresolution, load_pretrained_weights
22:34:31: superres_adapter.py: File "C:\Program Files\CodeProject\AI\modules\SuperResolution\superresolution.py", line 10, in
22:34:31: superres_adapter.py: from torch import nn
22:34:31: superres_adapter.py: ModuleNotFoundError: No module named 'torch'
2nd Machine
22:18:18: detect_adapter.py: Traceback (most recent call last):
22:18:18: detect_adapter.py: File "C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo\detect_adapter.py", line 16, in
22:18:18: detect_adapter.py: from analysis.codeprojectai import CodeProjectAIRunner
22:18:18: detect_adapter.py: File "../../SDK/Python\analysis\codeprojectai.py", line 36, in
22:18:18: detect_adapter.py: from analysis.requestdata import AIRequestData
22:18:18: detect_adapter.py: File "../../SDK/Python\analysis\requestdata.py", line 8, in
22:18:18: detect_adapter.py: from PIL import Image
22:18:18: detect_adapter.py: ImportError: cannot import name 'Image' from 'PIL' (unknown location)
Does that mean we can't use Yolov5x.NET version of CPAI? or will it see the .pt file versions of the custom models?You can download the onnx models and add them to the folder. Do not change the custom folder leave set to "C:\Program Files\CodeProject\AI\AnalysisLayer\ObjectDetectionYolo\custom-models" because Blue Iris can not recognize the onnx extension.
The Yolov5 v 6.2 version works ok in cpai 2.0.6 but the Yolov5.NET version does not work for me using any models I can try...how would one "repair", just reinstall cpai?Try doing a repair it looks like some of the Python packages did not install. I tested both CPU and GPU and the worked for me.
Running a repair did not work. I have spent a few hours trying everything I can think of and still can't get either machine work reliably. I've also tried uninstalling, reinstalling and even rolling back to previous versions to no avail. I've got AI to trigger a few times, but the processing times are extremely high.Try doing a repair it looks like some of the Python packages did not install.
Strange, I updated from 2.04 to 2.06 and everything runs same ??Running a repair did not work. I have spent a few hours trying everything I can think of and still can't get either machine work reliably. I've also tried uninstalling, reinstalling and even rolling back to previous versions to no avail. I've got AI to trigger a few times, but the processing times are extremely high.
View attachment 151245
It is strange. Maybe I shouldn't have installed 2.05.Strange, I updated from 2.04 to 2.06 and everything runs same ??
I'm still at 2.0.2 after successfully installing 2.0.3 and 2.0.4. Versions newer than 2.0.2 would not start the service so the server was not responding.It is strange. Maybe I shouldn't have installed 2.05.
Thanks Mike. I will try this tonight. If I have a GPU, what is the current recommended setup running v2.0.6? I use your General and Animal models. Should I use YOLOv5.NET (GPU) or YOLOv5.6.2? Depending on that selection, which version of each of your models should be used? I just want to make sure I'm using the correct setup once I get to v2.0.6. Really appreciate your input!I am see several users are having issues with v2.0.X. The current version is 2.0.6 and recommend uninstall any previous version and delete the two directors show below and then install version 2.0.6
- C:\Program Files\CodeProject
- C:\ProgramData\CodeProject
If you have a Nvidia GPU then use Object Detection (YOLOv5 6.2)Thanks Mike. I will try this tonight. If I have a GPU, what is the current recommended setup running v2.0.6? I use your General and Animal models. Should I use YOLOv5.NET (GPU) or YOLOv5.6.2? Depending on that selection, which version of each of your models should be used? I just want to make sure I'm using the correct setup once I get to v2.0.6. Really appreciate your input!
Do I still need to install the Cudnn file after installing 2.0.x?If you have a Nvidia GPU then use Object Detection (YOLOv5 6.2)
What GPU do you have ?Do I still need to install the Cudnn file after installing 2.0.x?
A 3060. I installed the CUDA driver before CodeProject.AI. YOLOv5 already says GPU before I installed cuDNN.What GPU do you have ?
You need to install CUDA first then cuDNN before installing CodeProject.AI 2.0.6
Does it work, if so the you must have cuDNN installedA 3060. I installed the CUDA driver before CodeProject.AI. YOLOv5 already says GPU before I installed cuDNN.