|
i installed python 3.9.19 and got the following errors on YOLOv5 6.2.
Infor Started Object Detection (YOLOv5 6.2) module
Error detect_adapter.py: Traceback (most recent call last):
Error detect_adapter.py: File "/usr/bin/codeproject.ai-server-2.6.2/modules/ObjectDetectionYOLOv5-6.2/detect_adapter.py", line 20, in <module>
Error detect_adapter.py: from detect import do_detection
Error detect_adapter.py: File "/usr/bin/codeproject.ai-server-2.6.2/modules/ObjectDetectionYOLOv5-6.2/detect.py", line 8, in <module>
Error detect_adapter.py: from yolov5.models.common import DetectMultiBackend, AutoShape
Error detect_adapter.py: ModuleNotFoundError: No module named 'yolov5'
Infor ** Module ObjectDetectionYOLOv5-6.2 has shutdown
Infor detect_adapter.py: has exited
|
|
|
|
|
We've just updated the Ubuntu version to 2.6.4. Curious to know if 2.6.4 works better for you.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Hi everyone.
CodeProject environment:
-hw:raspberry PI5 8G+CoralUsb
-latest pi os lite 64Bit
-latest docker engine
Downloaded and installed the rpi64 2.5.6.1 docker image.
Disabled all modules and activated only the coral module (disable GPU, EfficientDetLite, Tiny, No MultiTPU).
Configured automatic boot.
Everything works fine but after a few hours the system no longer responds at blueiris call.
I have to restore it by restarting the coral module from the web interface (the log appears blocked) or with a system reboot.
Do you have any suggestions on what to do? What can I check?
Thank you
Mike
modified 2-Apr-24 13:18pm.
|
|
|
|
|
I have been running it pretty flawlessly. Did you install the coral driver on the pi5?
I think you might have to compile gasket-dkms from source from what I remember
|
|
|
|
|
thank you for the message. if it were a problem with the coral usb driver I think the system would never work. the raspberry responds correctly to blueiris even for many hours but then unexpectedly freezes even if the operating system and the codeproject web interface continue to work. no information on the codeproject interface. just restart the raspberry and everything will be perfect again for another few hours
|
|
|
|
|
I have found the exact same issue both on my Pi4 and Pi5 on 2.5.6.1 with the USB Coral. I noticed that if I have the Model Size on Tiny it takes longer before the crash/timeout.
I purchased the Pineberry HATAI! and a Google Coral A+E Key. I have installed latest Ubuntu 23.10 server and after a bit of messing with PCIe Apex & Gasket drivers I finally got it up and running on 2.5.6.1 and it has been working for a few days now and Blue Iris alerts are working perfectly.
I definitely think there is an issue with the USB side of things as the m.2 bottom hat seems to be much more stable.
|
|
|
|
|
The USB is unreliable. I had one and had random freezes with it. I switched over to their physical cards on all my systems and they never stop working.
I do run the pi hat on my pi5 and its rock solid. The other systems I have use the dual edge tpu with the pci card from makerfabs
|
|
|
|
|
It’s fixed quickly when CPAI is restarted, so even if the USB is unreliable, it should be trivial to handle. CPAI should be monitoring it and restart itself (or the Docker container) when it stops responding.
|
|
|
|
|
I have the same issue. It just silently dies.
|
|
|
|
|
Hi, I have the same configuration without freezing but with a strange behaviour:
After the first connexion or after some time working, delay is considerably increased :
08:52:30:Started Object Detection (Coral) module
08:52:33:objectdetection_coral_adapter.py: Using Edge TPU
08:53:48:Response rec'd from Object Detection (Coral) command 'detect' (...87bb02) ['Found person, person, person...'] took 2697ms
08:53:55:Response rec'd from Object Detection (Coral) command 'detect' (...f8a055) ['Found person, person, person...'] took 21ms
08:56:01:Response rec'd from Object Detection (Coral) command 'detect' (...d82031) ['Found person, person, person...'] took 22ms
Just after it becomes OK.
After a long delay I get these messages in dmesg:
usb 2-1: reset SuperSpeed USB device number 3 using xhci-hcd
usb 2-1: LPM exit latency is zeroed, disabling LPM.
Is that intentionnal ? or may be usb malfunctionning ?
|
|
|
|
|
I'm testing ocr via codeproject/explorer.
When I place an image with a resolution of 4032x3024, it does not read the characters that appear in the center of the image, but when I cut it and save it in another file with a resolution of 966x747, it returns successfully.
What can it be? I'm new to using codeproject...
Files: OCR - Google Drive[^]
|
|
|
|
|
Thanks very much for your message. When CodeProject.AI Server processes an image, it gets scaled down. So generally, if you can't read the text in an image, CodeProject.AI Server probably won't be able to read the text in the image.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Thanks for the feedback. Isn't there some configuration in the ocr module files to avoid this reduction? Because in the file I placed on Google Drive, for example, you can read both images. I put the results of the OCR scan on Google Drive.
modified 22-Mar-24 20:28pm.
|
|
|
|
|
Hello Everyone,
I am not a tech person. Blue Iris is perfect for my home security and I love that I can get answers to my questions using Code Project when I switched from Deep Stack. I am facing problems with downloading the module for for License Plate reading. I am so confused with having to to get paddle ocr from github on the backend. Is there any easy way from a kid from the 1980's to just download the software? I did use Rekor Scout and that was amazing but costly.
|
|
|
|
|
Thanks very much for your message. Could you please share your System Info tab from your CodeProject.AI Server dashboard? I can help better if I know what your setup is.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Asking because both started up and seem to be working. Is this bad? Good? Pointless? More / less accurate detections?
modified 22-Mar-24 12:59pm.
|
|
|
|
|
The two modules will fight over requests. You are essentially splitting the work between the newer and older model architecture so you'll get slightly different (possibly hard to notice) results at double the memory cost.
cheers
Chris Maunder
|
|
|
|
|
I use CPAI to do object validation for BlueIris. I use MobileSSD model with tiny model size.
Everything runs smoothly at about 60ms but times to times (once every 2-3 hours) I get a random > 2600ms detection time.
Sometimes it seems to be related to "Refreshing the Tensorflow Interpreter" but most of the time not. Any idea what could be the source ? Anything I can do to avoid that ?
07:37:25:Response rec'd from Object Detection (Coral) command 'detect' (...3f8019) ['No objects found'] took 52ms
07:38:32:Response rec'd from Object Detection (Coral) command 'detect' (...8e2903) ['No objects found'] took 53ms
07:39:24:Response rec'd from Object Detection (Coral) command 'detect' (...375f07) ['No objects found'] took 53ms
07:40:31:Response rec'd from Object Detection (Coral) command 'detect' (...7d9bf1) ['No objects found'] took 55ms
07:41:23:Response rec'd from Object Detection (Coral) command 'detect' (...a206c9) ['No objects found'] took 16ms
07:42:30:Response rec'd from Object Detection (Coral) command 'detect' (...ce5724) ['No objects found'] took 17ms
07:43:22:Response rec'd from Object Detection (Coral) command 'detect' (...3a592f) ['No objects found'] took 52ms
07:44:29:Response rec'd from Object Detection (Coral) command 'detect' (...9b146c) ['No objects found'] took 52ms
07:45:24:Response rec'd from Object Detection (Coral) command 'detect' (...aa6089) ['No objects found'] took 2656ms
07:46:28:Response rec'd from Object Detection (Coral) command 'detect' (...9c9570) ['No objects found'] took 52ms
07:47:20:Response rec'd from Object Detection (Coral) command 'detect' (...ccf584) ['No objects found'] took 17ms
07:48:27:Response rec'd from Object Detection (Coral) command 'detect' (...96f7cd) ['No objects found'] took 16ms
07:49:19:Response rec'd from Object Detection (Coral) command 'detect' (...6813a1) ['No objects found'] took 17ms
07:50:27:Response rec'd from Object Detection (Coral) command 'detect' (...e8d687) ['No objects found'] took 53ms
07:51:19:Response rec'd from Object Detection (Coral) command 'detect' (...df9d12) ['No objects found'] took 53ms
07:52:26:Response rec'd from Object Detection (Coral) command 'detect' (...05a38c) ['No objects found'] took 52ms
07:53:18:Response rec'd from Object Detection (Coral) command 'detect' (...a65898) ['No objects found'] took 52ms">
19:56:30:Response rec'd from Object Detection (Coral) command 'detect' (...b1904b) ['No objects found'] took 54ms
19:58:02:Response rec'd from Object Detection (Coral) command 'detect' (...e64121) ['No objects found'] took 53ms
19:58:29:Response rec'd from Object Detection (Coral) command 'detect' (...5b20a3) ['No objects found'] took 58ms
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:objectdetection_coral_adapter.py: Refreshing the Tensorflow Interpreter
20:51:05:Response rec'd from Object Detection (Coral) command 'detect' (...4383f3) ['No objects found'] took 2675ms
20:51:06:Response rec'd from Object Detection (Coral) command 'detect' (...404675) ['No objects found'] took 61ms
20:51:06:Response rec'd from Object Detection (Coral) command 'detect' (...27deff) ['No objects found'] took 55ms
Module 'Object Detection (Coral)' 2.1.5 (ID: ObjectDetectionCoral)
Valid: True
Module Path: <root>/modules/ObjectDetectionCoral
AutoStart: True
Queue: objectdetection_queue
Runtime: python3.9
Runtime Loc: Local
FilePath: objectdetection_coral_adapter.py
Pre installed: False
Start pause: 1 sec
Parallelism: 1
LogVerbosity:
Platforms: all
GPU Libraries: installed if available
GPU Enabled: enabled
Accelerator:
Half Precis.: enable
Environment Variables
CPAI_CORAL_MODEL_NAME = MobileNet SSD
CPAI_CORAL_MULTI_TPU = False
MODELS_DIR = <root>/modules/ObjectDetectionCoral/assets
MODEL_SIZE = tiny
Status Data: {
"inferenceDevice": "TPU",
"inferenceLibrary": "TF-Lite",
"canUseGPU": "false",
"successfulInferences": 6577,
"failedInferences": 0,
"numInferences": 6577,
"averageInferenceMs": 10.651056712786986
}
Started: 15 Mar 2024 5:22:06 PM Coordinated Universal Time
LastSeen: 20 Mar 2024 12:21:35 PM Coordinated Universal Time
Status: Started
Requests: 10913 (includes status calls)
modified 20-Mar-24 11:23am.
|
|
|
|
|
Thanks very much for your report. Could you please also share your System Info tab from your CodeProject.AI Server dashboard?
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
I'm using Proxmox / LXC container / Docker / "codeproject/ai-server"
Server version: 2.5.6
System: Docker
Operating System: Linux (Ubuntu 22.04)
CPUs: AMD Ryzen 7 PRO 4750G with Radeon Graphics (AMD)
1 CPU x 8 cores. 16 logical processors (x64)
System RAM: 62 GiB
Platform: Linux
BuildConfig: Release
Execution Env: Docker
Runtime Env: Production
Runtimes installed:
.NET runtime: 7.0.16
.NET SDK: Not found
Default Python: Not found
Go: Not found
NodeJS: Not found
Video adapter info:
System GPU info:
GPU 3D Usage 0%
GPU RAM Usage 0
Global Environment variables:
CPAI_APPROOTPATH = <root>
CPAI_PORT = 32168
|
|
|
|
|
What you're seeing is Coral restarting the interpreter every hour.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Thanks,
Is that expected or can it be tuned ?
|
|
|
|
|
Theoretically it is something we could code in on our end, but we have no plans to do so at this time.
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
I constantly get this error caused by various errors:
01:28:20:objectdetection_coral_adapter.py: F driver/usb/usb_driver.cc:815] [0-0] bulk out for header failed. Abort. Unavailable: USB error -4 [AsyncBulkOutTransfer]
01:28:20:Module ObjectDetectionCoral has shutdown
02:31:33:objectdetection_coral_adapter.py: F driver/usb/usb_driver.cc:1148] HandleQueuedBulkIn transfer in failed. Not found: USB transfer error 5 [LibUsbDataInCallback]
02:31:34:Module ObjectDetectionCoral has shutdown
It starts working again if I restart the Docker container.
The container should detect this kind of thing and restart itself, instead of silently failing.
modified 20-Mar-24 11:12am.
|
|
|
|
|
This is on our TODO list
cheers
Chris Maunder
|
|
|
|