|
I know the YoloV5 6.2 training module has been around for a bit now, I'm wondering if a v8 version is being planned or will be coming at any point in the future?
modified 16-Apr-24 12:06pm.
|
|
|
|
|
Yes! At some point.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
After upgrading from v2.5.6 to 2.6.2 the docker container fails with following message:
Creating runtimes path '/app/runtimes'
Creating downloaded models path '/app/downloads/models'
Unable to start the server: An item with the same key has already been added. Key: ObjectDetectionCoral.
Check that another instance is not running on the same port.
Press Enter to close.
Does anybody know how to fix that issue?
Thanks a lot
Tbs
modified 15-Apr-24 13:01pm.
|
|
|
|
|
Can you please look in your /modules folder and list the folders there?
Could also be that you simply need to re-install the module. What platform are you running on?
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
This issue has been found, fixed and will be available in the next version.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
In future, Face Recognition can sync database?
or something I miss information for do that?
modified 16-Apr-24 12:05pm.
|
|
|
|
|
No plans at the moment, sorry.
cheers
Chris Maunder
|
|
|
|
|
Error message:
14:15:31:Started Object Detection (YOLOv5 6.2) module
14:15:31:detect_adapter.py: Traceback (most recent call last):
14:15:31:detect_adapter.py: File "C:\Program Files\CodeProject\AI\modules\ObjectDetectionYOLOv5-6.2\detect_adapter.py", line 13, in
14:15:31:detect_adapter.py: from module_runner import ModuleRunner
14:15:31:detect_adapter.py: File "../../SDK/Python\module_runner.py", line 30, in
14:15:31:detect_adapter.py: import aiohttp
14:15:31:detect_adapter.py: ModuleNotFoundError: No module named 'aiohttp'
modified 16-Apr-24 12:05pm.
|
|
|
|
|
It fails if you choose to only install Yolo 6.2, if you set it to install all the default options of Yolo 5 and face, it works.
|
|
|
|
|
|
Is there a way to prevent the TPU from automatically shutting down after 120 seconds of inactivity? It takes quite a while for an image to be recognized afterwards.
12:41:19:System: Docker (fa0761f44809)
12:41:19:Operating System: Linux (Ubuntu 22.04)
12:41:19:CPUs: 13th Gen Intel(R) Core(TM) i5-1340P (Intel)
12:41:19: 1 CPU x 12 cores. 16 logical processors (x64)
12:41:19:System RAM: 31 GiB
12:41:19:Platform: Linux
12:41:19:BuildConfig: Release
12:41:19:Execution Env: Docker
12:41:19:Runtime Env: Production
12:41:19:Runtimes installed:
12:41:19: .NET runtime: 7.0.17
12:41:19: .NET SDK: Not found
12:41:19: Default Python: 3.10.12
12:41:19: Go: Not found
12:41:19: NodeJS: Not found
12:41:19: Rust: Not found
12:41:19:App DataDir: /etc/codeproject/ai
12:41:19:Video adapter info:
12:41:19:STARTING CODEPROJECT.AI SERVER
12:41:19:RUNTIMES_PATH = /app/runtimes
12:41:19:PREINSTALLED_MODULES_PATH = /app/preinstalled-modules
12:41:19:DEMO_MODULES_PATH = /app/demos/modules
12:41:19:MODULES_PATH = /app/modules
12:41:19:PYTHON_PATH = /bin/linux/%PYTHON_NAME%/venv/bin/python3
12:41:19:Data Dir = /etc/codeproject/ai
12:41:19:Server version: 2.6.2
12:45:04:Started Object Detection (Coral) module
12:45:08:objectdetection_coral_adapter.py: TPU detected
12:45:08:objectdetection_coral_adapter.py: Attempting multi-TPU initialisation
12:45:08:objectdetection_coral_adapter.py: Supporting multiple Edge TPUs
12:45:48:Response rec'd from Object Detection (Coral) command 'detect' (...8c51a5) ['Found car, car, car...'] took 245ms
12:45:56:Response rec'd from Object Detection (Coral) command 'detect' (...3a38f2) ['Found car, car, car...'] took 75ms
12:45:57:Response rec'd from Object Detection (Coral) command 'detect' (...1c094c) ['Found car, car, car...'] took 80ms
12:45:57:Response rec'd from Object Detection (Coral) command 'detect' (...cb4a86) ['Found car, car, car...'] took 74ms
12:45:58:Response rec'd from Object Detection (Coral) command 'detect' (...08b14a) ['Found car, car, car...'] took 75ms
12:45:58:Response rec'd from Object Detection (Coral) command 'detect' (...ec2923) ['Found car, car, car...'] took 78ms
12:46:45:Response rec'd from Object Detection (Coral) command 'custom' (...229416) ['Found car, car, car...'] took 80ms
12:46:46:Response rec'd from Object Detection (Coral) command 'detect' (...d7ac0a) ['Found car, car, car...'] took 74ms
12:46:47:Response rec'd from Object Detection (Coral) command 'custom' (...883f24) ['Found car, car, car...'] took 74ms
12:48:50:objectdetection_coral_adapter.py: WARNING:root:No work in 120.0 seconds, watchdog shutting down TPUs.
12:56:18:Response rec'd from Object Detection (Coral) command 'detect' (...0b333f) ['Found car, person, car...'] took 3430ms
13:04:01:Response rec'd from Object Detection (Coral) command 'custom' (...14da1e) ['Found car, car, car...'] took 54ms
modified 5-Apr-24 17:46pm.
|
|
|
|
|
Thanks for pointing this out. I'm not sure why it'd take so long to start up. (For example, it looks like your first iteration took 245ms.) The intent is if something is stalled for whatever reason in the processing pipeline, the watchdog will reset things in the background so CPAI doesn't sit stalled for hours.
I've added some code that should make the recycle process faster and I've increased the watchdog timeout so it takes longer to trigger. Likely in the next CPAI release. Hopefully that'll fix the problem for you.
|
|
|
|
|
I also experience this, with the first detection after the TPU has shutdown often taking > 2 seconds.. which seems to have the knock on effect of making a TPU equipped host report a very high detection time to mesh hosts, and thus never receiving any mesh offload work unless I run a load of detections through manually to get the time down
|
|
|
|
|
With object detection yolov5.net 1.10.1 i just update and i cant get it to use the gpu ( i have a P1000) i select it in the menu it restarts and go back to CPU (DirectML) is this a bug ?
What I have tried:
ReInstall object detection yolov5.net 1.10.1 , pc restart , reinstalling CPAI
modified 5-Apr-24 17:46pm.
|
|
|
|
|
Thanks very much for your report. Could you please share your System Info tab from your CodeProject.AI Server dashboard?
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
|
Sean
It looks like a code typo mine is showing CPU, I tested it and it is definitely using the GPU.
|
|
|
|
|
Mike,
Could you please send my your System Info tab contents.
Thanks,
"Mistakes are prevented by Experience. Experience is gained by making mistakes."
|
|
|
|
|
Both of my systems are showing CPU
Server version: 2.6.2
System: Windows
Operating System: Windows (Microsoft Windows 10.0.19045)
CPUs: AMD Ryzen 9 5950X 16-Core Processor (AMD)
1 CPU x 16 cores. 32 logical processors (x64)
GPU (Primary): NVIDIA GeForce RTX 4090 (24 GiB) (NVIDIA)
Driver: 551.86, CUDA: 12.4 (up to: 12.4), Compute: 8.9, cuDNN: 8.9
System RAM: 128 GiB
Platform: Windows
BuildConfig: Release
Execution Env: Native
Runtime Env: Production
Runtimes installed:
.NET runtime: 7.0.17
.NET SDK: 7.0.203
Default Python: 3.11.0
Go: Not found
NodeJS: 20.8.1
Rust: Not found
Video adapter info:
NVIDIA GeForce RTX 4090:
Driver Version 31.0.15.5186
Video Processor NVIDIA GeForce RTX 4090
System GPU info:
GPU 3D Usage 1%
GPU RAM Usage 6 GiB
Global Environment variables:
CPAI_APPROOTPATH = <root>
CPAI_PORT = 32168
Server version: 2.6.2
System: Windows
Operating System: Windows (Microsoft Windows 10.0.19045)
CPUs: AMD Ryzen 9 3950X 16-Core Processor (AMD)
1 CPU x 16 cores. 32 logical processors (x64)
GPU (Primary): NVIDIA GeForce RTX 3060 (12 GiB) (NVIDIA)
Driver: 551.86, CUDA: 12.4 (up to: 12.4), Compute: 8.6, cuDNN: 8.9
System RAM: 64 GiB
Platform: Windows
BuildConfig: Release
Execution Env: Native
Runtime Env: Production
Runtimes installed:
.NET runtime: 7.0.5
.NET SDK: Not found
Default Python: 3.10.1
Go: Not found
NodeJS: Not found
Rust: Not found
Video adapter info:
NVIDIA GeForce RTX 3060:
Driver Version 31.0.15.5186
Video Processor NVIDIA GeForce RTX 3060
System GPU info:
GPU 3D Usage 3%
GPU RAM Usage 6.8 GiB
Global Environment variables:
CPAI_APPROOTPATH = <root>
CPAI_PORT = 32168
|
|
|
|
|
Mike, you are correct, it must be a type. DirectML will not be displayed without whatever graphics driver installed on the Windows system being used.
"Mistakes are prevented by Experience. Experience is gained by making mistakes."
|
|
|
|
|
Does object detection yolov5.net is it normal to take 20% of a i7 6700 cpu even when it set to use GPU
|
|
|
|
|
I have confirmed it is a typo in the code. Not the one you would expect, but it still causes the error.
I have fixed the code and it will be in the next release.
"Mistakes are prevented by Experience. Experience is gained by making mistakes."
|
|
|
|
|
|
What would this be?
08:59:27:
08:59:27:Started Object Detection (YOLOv8) module
09:00:19:Object Detection (YOLOv8): [AttributeError] : Traceback (most recent call last):
File "/app/modules/ObjectDetectionYOLOv8/detect.py", line 147, in do_detection
results = detector.predict(img, imgsz=resolution, half=use_half_precision,
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/ultralytics/engine/model.py", line 266, in predict
self.predictor.setup_model(model=self.model, verbose=is_cli)
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/ultralytics/engine/predictor.py", line 341, in setup_model
self.model = AutoBackend(
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/ultralytics/nn/autobackend.py", line 141, in __init__
model = model.fuse(verbose=verbose) if fuse else model
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/ultralytics/nn/tasks.py", line 176, in fuse
delattr(m, "bn") # remove batchnorm
File "/app/modules/ObjectDetectionYOLOv8/bin/linux/python38/venv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1266, in __delattr__
object.__delattr__(self, name)
AttributeError: bn
09:00:19:Response rec'd from Object Detection (YOLOv8) command 'detect' (...8d4e22)
Everything installed successfully without error, but the YOLOv8 didn't automattically start, I had to start it.
Here is my SystemInfo:
Server version: 2.6.2
System: Docker (Linux 5.10.102.1-microsoft-standard-WSL2 #1 SMP Wed Mar 2 00:30:59 UTC 2022)
Operating System: Linux (Ubuntu 22.04)
CPUs: AMD Ryzen Threadripper 3960X 24-Core Processor (AMD)
1 CPU x 24 cores. 48 logical processors (x64)
GPU (Primary): NVIDIA GeForce RTX 3070 (8 GiB) (NVIDIA)
Driver: 551.23, CUDA: 12.4 (up to: 12.4), Compute: 8.6, cuDNN: 8.9.6
System RAM: 63 GiB
Platform: Linux
BuildConfig: Release
Execution Env: Docker
Runtime Env: Production
Runtimes installed:
.NET runtime: 7.0.17
.NET SDK: Not found
Default Python: 3.10.12
Go: Not found
NodeJS: Not found
Rust: Not found
Video adapter info:
System GPU info:
GPU 3D Usage 25%
GPU RAM Usage 7.5 GiB
Global Environment variables:
CPAI_APPROOTPATH = <root>
CPAI_PORT = 32168
modified 5-Apr-24 17:46pm.
|
|
|
|
|
Thanks very much for your message. The YOLOv8 module isn't set to start automatically. What should happen is, once you start the module once, it should auto-start next time.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|