CUDA Errors with YOLOv5 Object Detection CPAI 2.5.x / Blue Iris – Seeking Insights - Artificial Intelligence Discussion Boards

I'm so impressed with what your dev community has come up with so far for your open source AI.

Vehicle number plate recognition is truly awesome.

Think of the potential applications for that module alone in all kinds of applications from law enforcement to industry-use.

This AI has tremendous potential the more ideas and new modules are developed and added to it and could become extremely useful and valuable for so many real-world time-saving applications.

Massive Kudos to your core developer team and all developers contributing open source modules to further improve and expand the AI's capabilities.

How are you guys funded? Do you have any financial backing/grants or do you rely on donations? Or are you guys doing this without any funding?

I'm not a professional programmer myself but recognise the enormous future potential AI has to solving real-world problems and making difficult tasks easier. I'd love to help contribute in any way I can perhaps with funding if you need it.

Over the last 2-3 years I've been experimenting heavily with OpenAI ChatGPT which has become an invauable tool to me. One of the potential uses for it I experimented with last year is using ChatGPT for creating online teenage chat personas to simulate chats for catching child groomers & peadophiles. ChatGPT really nailed this. It was able to have very convincing chatroom text-chats in the manner of a 12-14 year old child after I gave it a fictional life-background story, a personality-style and taught it to adopt the typical text chat style of children online. It was hard to tell at times it was not a child and just an AI simulation! Imagine the potential AI has for bringing technological solutions to technological Internet-crime problems like that. With the ability to chat with thousands of child-predators at the same time, gather valuable intel and do risk assessment for law enforcement and child-protection agencies worldwide. That's just one application, imagine cybercrime/fraud solutions..or identifying and tracking lost/stolen property and pets from RFID chip tags etc..the possibilities are endlesses with a AI!

Keep up the great work guys and let's keep sharing new ideas!

Paul

Hi guys,

I have been playing with OpenAI ChatGPT over the last years and as a programmer and developer myself would love to get hold of an open source AI and run one locally so when I discovered your project I was astonished. Kudos to your development community for not only making such a brilliant AI and do many modules but doing it open source and letting people contribute. AI is now coming of age and we're going to see a big shift soon towards mainstream use with ever-more advanced AI's but most like Apple Siri,Google Asssistant, Microsoft Copilot and the most powerful OpenAI ChatGPT are closed source.

Your product looks really interesting with an incredible develop community potential. I'd like to install it on an old MacPro1,1 I have which despite being an early 1st gen MacPro (circa 2006) is best-spec and quite a beast: 2 x 4 core 3Ghz CPU, 32GB RAM, 20TB disk array, ATI Radeon HD 5770 1GB VRAM GPU. Would this be a suitable server platform for running your AI (either in MacOSX 10.11 most recent MacOS version it supports) or the latest 64-bit Ubuntu Linux? I'm thinking more about GPU support?

I'd love to build something like ChatGPT, but with Internet access. Basically something like Google Assistant & OpenAI ChatGPT. Would your open source AI be a suitable fit for something like or should I looking at something else?

Any suggestions would be appreciated! Although I'm familar with using all the above AI's and especially OpenAI ChatGPT I'm only getting started on installing one myself and the development side so it's all new to me.

Regards,

Paul

P.S. A few suggestion ideas for cool future module add-ons for your AI that would bring really useful real-world functionality in so many areas could be:

1. Ability to extract GPS coordinates from photos/video and place on maps in real-time. Plus video time-stamp recognition and ability to jump quickly through videos to specific time-stamp periods on videos like CCTV recordings.

2. Reverse-image searching for facial/object/location feature image recognition matching. As well as text/data pattern search matching.

3. TTS voice synthesis capability, like ChatGPT latest version has which is scarily so close to natural human sounding speech now!

4. RFID chip-scanning for RFID chip detection and reading embedded code from chips, plus QR barcode recognition and generation.

5. Cellular phone GPS tracking with real time map overlay and possible integration with Google Maps/Google Streetview.

6. Human-behaviour mimicking from chat interactions (which ChatGPT does extremely well with its 'mirroring psychology' subroutine) to sound ever more human-like..it's now even clearing making throat clearing noises and changing vocal intonations based on interactions).

7. Scaleability: support for clustered low-powered CPU microprocessors/GPU's and distributed network/cloud data storage arrays for improved data storage and greatly improved processing power for potentially developing a super-computer AI for more demanding applications.

Hi all I have had number plate recognition working for quite some time then over the past few months it has stopped working and getting the error shown below:

14:30:00:Object Detection (YOLOv5 6.2): Detecting using license-plate
14:30:00:Response rec'd from Object Detection (YOLOv5 6.2) command 'custom' (...7f9def) ['Found DayPlate']  took 67ms
14:30:00:License Plate Reader:  [AttributeError] : Traceback (most recent call last):
  File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR_adapter.py", line 57, in process
    result = await detect_platenumber(self, self.opts, image)
  File "C:\Program Files\CodeProject\AI\modules\ALPR\ALPR.py", line 132, in detect_platenumber
    bounding_box_result = ocr.ocr(numpy_plate, rec=False, cls=False)
AttributeError: 'NoneType' object has no attribute 'ocr'

I cant figure it out or find a solution so wondered if anyone here has an idea please? Starting to loose my mind Frown | :-(

Thanks

You should be asking this question here[^]. This is the forum for CodeProjec AI questions.

Advanced TypeScript Programming Projects

Thanks have moved it

Advice for object detection for LEO person detection and LEO vehicle detection, speed detection with blue iris setup. RTX4070 8.9, 6-4k cams, Street facing cams

I'm working on my thesis project, and want to make an intelligent search assistant that understands context and, of course, processes and repsonds in natural languaje. The data I want to train this model on is from the Virtual Observatory and a python library that can be used to retrieve data from it.

I thought on Open AI's GPT-3 API, but the knowledge to which it has access to is outdated. The I thought about IBM's Wattson Discovery, but I feel that using their solution would limit the response type or the training process very much.

What ML model(s) would work better in my case? Or what software/solution would be useful?

I'm reaching out to share a perplexing issue I've encountered with the integration CPAI with my BI setup, hoping to find if anyone else has experienced something similar or could offer any insights. The problem first manifested around 01:45 am on 14/02/2024, and despite troubleshooting efforts, it recurred this morning, indicating a persistent underlying issue.

Initially, the system logs from 14th February showed an error related to CUDA, specifically mentioning "an illegal memory access was encountered". This issue caused a loop of errors until a system reboot was performed at 9:06 am.

Here is the exact log entry for reference:

Terminal

<pre lang="Terminal">2024-02-14 01:39:57: Object Detection (YOLOv5 6.2): Retrieved objectdetection_queue command 'custom' in Object Detection (YOLOv5 6.2)
2024-02-14 01:39:57: Object Detection (YOLOv5 6.2): Detecting using ipcam-combined in Object Detection (YOLOv5 6.2)
2024-02-14 01:39:57: Response received (#reqid 85bde494-89d3-429d-a21b-c10b9430c5a8 for command custom)
2024-02-14 01:39:57: Object Detection (YOLOv5 6.2):  [RuntimeError] : Traceback (most recent call last):
  File "C:\Program Files\CodeProject\AI\modules\ObjectDetectionYOLOv5-6.2\detect.py", line 141, in do_detection
    det                  = detector(img, size=640)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\nn\modules\module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\models\common.py", line 669, in forward
    with dt[0]:
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\utils\general.py", line 158, in __enter__
    self.start = self.time()
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\utils\general.py", line 167, in time
    torch.cuda.synchronize()
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\cuda\__init__.py", line 566, in synchronize
    return torch._C._cuda_synchronize()
RuntimeError: CUDA error: an illegal memory access was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
 in Object Detection (YOLOv5 6.2)

I upgraded yesterday from CPAI v2.5.1 to v2.5.4, hoping the update would resolve the issue which took place on the 14th. However, this morning, the same CUDA error reappeared, this time indicating "an illegal instruction was encountered". The error persisted until a reboot was done just after 9 am. Below is the log excerpt from today's occurrence:

Terminal

2024-02-18 04:52:40: Object Detection (YOLOv5 6.2): Detecting using ipcam-combined in Object Detection (YOLOv5 6.2)
2024-02-18 04:52:40: Response rec'd from Object Detection (YOLOv5 6.2) command 'custom' (#reqid ad29496b-5caf-4d8b-b02f-7fcc6c7ab605) ['No objects found']  took 22ms
2024-02-18 04:52:40: Client request 'custom' in queue 'objectdetection_queue' (#reqid 7e1b52dd-b880-4c35-b7c6-0f076127faab)
2024-02-18 04:52:40: Request 'custom' dequeued from 'objectdetection_queue' (#reqid 7e1b52dd-b880-4c35-b7c6-0f076127faab)
2024-02-18 04:52:40: Object Detection (YOLOv5 6.2): Retrieved objectdetection_queue command 'custom' in Object Detection (YOLOv5 6.2)
2024-02-18 04:52:40: Object Detection (YOLOv5 6.2): Detecting using ipcam-combined in Object Detection (YOLOv5 6.2)
2024-02-18 04:52:40: Response rec'd from Object Detection (YOLOv5 6.2) command 'custom' (#reqid 7e1b52dd-b880-4c35-b7c6-0f076127faab)
2024-02-18 04:52:40: Object Detection (YOLOv5 6.2):  [RuntimeError] : Traceback (most recent call last):
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\models\common.py", line 715, in forward
    max_det=self.max_det)  # NMS
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\utils\general.py", line 920, in non_max_suppression
    x = torch.cat((box, conf, j.float(), mask), 1)[conf.view(-1) > conf_thres]
RuntimeError: CUDA error: an illegal instruction was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\Program Files\CodeProject\AI\modules\ObjectDetectionYOLOv5-6.2\detect.py", line 141, in do_detection
    det                  = detector(img, size=640)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\nn\modules\module.py", line 1190, in _call_impl
    return forward_call(*input, **kwargs)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\models\common.py", line 717, in forward
    scale_boxes(shape1, y[i][:, :4], shape0[i])
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\utils\general.py", line 162, in __exit__
    self.dt = self.time() - self.start  # delta-time
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\yolov5\utils\general.py", line 167, in time
    torch.cuda.synchronize()
  File "C:\Program Files\CodeProject\AI\runtimes\bin\windows\python37\venv\lib\site-packages\torch\cuda\__init__.py", line 566, in synchronize
    return torch._C._cuda_synchronize()
RuntimeError: CUDA error: an illegal instruction was encountered
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.
 in Object Detection (YOLOv5 6.2)

The repeating nature of this error is particularly concerning as it undermines the reliability of object detection capabilities, which are crucial for the functionality I rely on. It's disconcerting to see the system fail in such a manner, especially considering the otherwise commendable performance improvements in the AI aspects of the software.

Has anyone else encountered similar issues, particularly with CUDA errors causing system instability? Any advice on troubleshooting or resolving this would be immensely appreciated. I've also posted this on the Blue Iris forum to cast a wider net for potential solutions.

Thank you in advance for your time and assistance.

As already suggested, please post your question in the CodeProject.AI Discussions[^] forum.

I've posted in the CodeProject.AI thank you, how do I delete this post I cannot see any option allowing me to delete?

That's OK, you can leave this post here so other people can see where to go for help.

How does the CodeProject.AI Discussions forum relate to "ordinary" Artificial Intelligence forum in the General Programming group of CP? Which topics or kinds of questions should go in each of them?

Religious freedom is the freedom to say that two plus two make five.

CodeProject.AI is a specific project run by the CodeProject staff. So any question related to this project and its usage should be posted in its own discussion forum. Mainly because that is where the project team go to look for questions. The general AI forum is for general, i.e. non-CodeProject.AI, questions. If you look on the CodeProject home page at the picture at the top, it has a link to the correct forum.

Please post your question in the CodeProject.AI Discussions[^] forum.

Hello everyone,

I'm new to this forum and I'm currently in the midst of a decision-making process regarding the optimal GPU usage for running CPAI on Blue Iris, particularly for a single-camera setup at my home. The primary purpose is to detect human presence while effectively filtering out false triggers due to weather conditions.

At present, I own a RTX 3080, which I'm considering repurposing for this task, especially since I'm contemplating an upgrade to one of the 4000 series super cards in the near future. However, I'm deliberating whether the 3080 might be overkill for my specific requirements.

After substantial research and discussions on the Blue Iris forum, the consensus appears to be in favour of the GTX 1650. It's been recommended as a more than adequate solution for CPAI, offering sufficient processing speed while maintaining lower power consumption.

My current setup relies solely on CPU processing, resulting in a latency of about 120-200ms for image processing. In contrast, I came across a post on the Blue Iris forum indicating that the GTX 1650 could potentially reduce this latency to around 30+ms. This substantial improvement naturally piques my interest.

However, I can't help but wonder about the potential benefits of deploying my GTX 3080 for this purpose. Would there be a significant advantage in terms of processing speed or efficiency? I've noticed that a fellow member, MikeLud, is conducting tests with a GTX 4090, which adds another layer of curiosity regarding the performance spectrum of these GPUs.

While I'm currently leaning towards the GTX 1650, primarily due to its power efficiency and seemingly adequate capabilities for my needs, I'm eager to hear your thoughts and experiences. Has anyone here used a GTX 3080/3080TI for a similar setup? If so, what were your observations regarding latency, power consumption, and overall performance?

Your insights and any additional information you can provide would be greatly appreciated. I’m looking to make the most informed decision to ensure efficient and effective surveillance at my home.

Thank you in advance for your valuable input!

Best regards
Wink | ;)

Hello, I'm trying to add Super Resolution to my CodeProject.AI_ServerGPU Docker container running on UNRAID. I went into the container and clicked Install Modules and clicked Install beside Super Resolution. I get the below log within the Server logs tab in the container. I also tried letting the container run as Privileged with no luck.

Is there a different way I need to follow to install additional Modules within this container when using docker than using the Install Modules button?

My environment details:
OS: UNRAID 6.12.6
Docker container: CodeProject.AI_ServerGPU version 2.2.4-Beta
GPU: NVIDIA GeForce GTX 1050 Ti on driver latest: v545.29.06
CUDA Version 12.3

17:34:12:Preparing to install module 'SuperResolution'
17:34:12:Downloading module 'SuperResolution'
17:34:13:Installing module 'SuperResolution'
17:34:13:SuperResolution: Hi Docker! We will disable shared python installs for downloaded modules
17:34:13:SuperResolution: No schemas installed
17:34:13:SuperResolution: (No schemas means: we can't detect if you're in light or dark mode)
17:34:13:SuperResolution: sh: 1: lsmod: not found
17:34:13:SuperResolution:              Installing CodeProject.AI Analysis Module                
17:34:13:SuperResolution: ======================================================================
17:34:13:SuperResolution:                    CodeProject.AI Installer                           
17:34:13:SuperResolution: ======================================================================
17:34:13:SuperResolution: 66.02 GiB  available
17:34:13:SuperResolution: Installing curl...
17:34:13:SuperResolution: WARNING: apt does not have a stable CLI interface. Use with caution in scripts.
17:34:14:SuperResolution: E: Failed to fetch http://security.ubuntu.com/ubuntu/pool/main/c/curl/libcurl4_7.81.0-1ubuntu1.13_amd64.deb  404  Not Found [IP: 91.189.91.83 80]
17:34:14:SuperResolution: E: Failed to fetch http://security.ubuntu.com/ubuntu/pool/main/c/curl/curl_7.81.0-1ubuntu1.13_amd64.deb  404  Not Found [IP: 91.189.91.83 80]
17:34:14:SuperResolution: E: Unable to fetch some archives, maybe run apt-get update or try with --fix-missing?
17:34:14:Module SuperResolution installed successfully.
17:34:14:
17:34:14:Module 'Super Resolution' 1.6 (ID: SuperResolution)
17:34:14:Installer exited with code 10
17:34:14:Module Path:   /app/modules/SuperResolution
17:34:14:AutoStart:     True
17:34:14:Queue:         superresolution_queue
17:34:14:Platforms:     windows,linux,linux-arm64,macos,macos-arm64
17:34:14:GPU:           Support disabled
17:34:14:Parallelism:   1
17:34:14:Accelerator:
17:34:14:Half Precis.:  enable
17:34:14:Runtime:       python38
17:34:14:Runtime Loc:   Local
17:34:14:FilePath:      superres_adapter.py
17:34:14:Pre installed: False
17:34:14:Start pause:   0 sec
17:34:14:LogVerbosity:
17:34:14:Valid:         True
17:34:14:Environment Variables
17:34:14:PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION = python
17:34:14:
17:34:14:Error trying to start Super Resolution (superres_adapter.py)
17:34:14:Module SuperResolution started successfully.
17:34:14:An error occurred trying to start process '/app/modules/SuperResolution/bin/linux/python38/venv/bin/python3' with working directory '/app/modules/SuperResolution'. No such file or directory
17:34:14:   at System.Diagnostics.Process.ForkAndExecProcess(ProcessStartInfo startInfo, String resolvedFilename, String[] argv, String[] envp, String cwd, Boolean setCredentials, UInt32 userId, UInt32 groupId, UInt32[] groups, Int32& stdinFd, Int32& stdoutFd, Int32& stderrFd, Boolean usesTerminal, Boolean throwOnNoExec)
   at System.Diagnostics.Process.StartCore(ProcessStartInfo startInfo)
   at CodeProject.AI.Server.Modules.ModuleProcessServices.StartProcess(ModuleConfig module)
17:34:14:Please check the CodeProject.AI installation completed successfully
17:34:15:Call to Install on module SuperResolution has completed.

modified 20-Jan-24 18:54pm.

Please post your question in the CodeProject.AI Discussions[^] forum.

Thanks, done and will remove this post!

CodeProject.AI Discussions - UNRAID installing Super Resolution Module fails[^]

Verify Super Resolution module compatibility with CodeProject.AI_ServerGPU version 2.2.4-Beta, your GPU, and CUDA version.
Inspect Docker container logs for error messages related to the module installation failure.
Try manually installing the module via the container's shell using specific scripts or commands.
Ensure your Docker container is up to date to avoid issues resolved in newer versions.
If problems persist, seek support from the CodeProject.AI community or support forums with detailed error logs. <pre><a href="capcuttemplates" rel="dofollow">https://capcuttemplates.co/</a>

hi
i have all work fine. but not continuous like i want

i want that blue iris will triggered almost in continuous time with face recognition

i use it like door locked
so lot people walk into the door
so i want when it detects face that is register it will confirm and send http commands

all work except the part of the continuity

its work every few minutes of break

i also play with the break time, trigger time, send pics per ms

i didn’t get it work

can it been done ?

thanks
yossi

If your issue is with the CodeProject AI server then please post your question in the CodeProject.AI Discussions[^] forum.

I had object detection .net working with the integrated gpu previously, so i switched to a 1650 video card. I removed .net and went with the 6.2 one.

I also ran the cuda installer and the batch file installer per the codeproject ai download page.

I've restarted the pc and i cant get it to go to GPU mode

any thoughts out there?

I had moved this thread to the other forum, but for some reason i cant find it, but here was the solution that worked for me:

Uninstall code project
Uninstall cuda 11.7
Uninstall gfx card drivers

Reboot

Install 516.94 gfx drivers
Install cuda 11.7.0
Install Code Project 2.0.8
Install cuDNN bat file (https://www.codeproject.com/ai/docs/faq/gpu.html)

modified 27-Sep-23 9:36am.

One suggestion, but first a question; does anything else on your box use the GPU in this expected manner? If this is a testing-of-the-water type of scenario wouldn't it be better to get tried-and-true stuff running first?

I have no piers here on CP and see myself as somewhat of a pariah when I post but I'm sure somewhere in the expanse of time between using the CPU and using the GPU, in realtime computer use (1995-present), sticking to allowing a CPU to enslave the GPU, is in the questor's best interests.

well, i can use the 1650 as the primary gpu without issue. In this situation i'm using the built in one for the lcd.

Both drivers show up and show up in device manger, i had no issues with the two cuda installers either

Perhaps ask your original question here then:
CodeProject.AI Discussions[^]
I've observed reposting on cp will result in boiled tempers so, because you're moving to a better venue, perhaps adding some keywords to better explain AND keeping your grammar right (like capitalizing your first-person pronouns) will cause someone there to come to the rescue.