|
Our installer is a fairly standard Windows installer and should clean out most of its debris. The only exception being (at this time) that it won't uninstall modules or ProgramData. This is to allow the update of the server to happen without requiring a full (and painfully slow, sometimes) install of the modules.
We're rewriting the entire installer to allow a full clean and hose-down.
For now, you can uninstall, then delete the C:\Program Files\CodeProject\AI folder, and also the C:\ProgramData\CodeProject\AI folders and you're clean.
After your reinstall, did the CodeProject.AI dashboard come up? In Blue Iris I would suggest disabling the stop/start AI service.
cheers
Chris Maunder
|
|
|
|
|
Chris I'm currently reinstalling CPAI and Coral Runtime, I'll let you know how it goes.
I noticed in the manual there's a section on a HIB error that blams the registry calls albeit it seems to be for Linux. I wonder if something similar is happening in windows.
This is what it says (from the Coral Website):
Get started with the M.2 or Mini PCIe Accelerator | Coral[^]
Quote: Troubleshooting on Linux
Here are some solutions to possible problems on Linux.
HIB error
If you are running on ARM64 platform and receive error messages such as the following when you run an inference...
HIB Error. hib_error_status = 0000000000002200, hib_first_error_status = 0000000000000200
... You should be able to solve it if you modify your kernel command line arguments to include gasket.dma_bit_mask=32.
For information about how to modify your kernel command line arguments, refer to your respective platform documentation. For bootloaders based on U-Boot, you can usually modify the arguments either by modifying the bootargs U-Boot environment variable or by setting othbootargs environment variable as follows:
=> setenv othbootargs gasket.dma_bit_mask=32
=> printenv othbootargs
othbootargs=gasket.dma_bit_mask=32
=> saveenv
If you make the above change and then receive errors such as, DMA: Out of SW-IOMMU space, then you need to increase the swiotlb buffer size by adding another kernel command line argument: swiotlb=65536.
As mentioned that is for Linux, but wonder if could also affect windows..
|
|
|
|
|
Back up and no blue screen now. Same issues though with Hib error.
Latest Logs:
Quote: cam setup guide and common issues pages.
20:43:12:Operating System: Windows (Microsoft Windows 10.0.19045)
20:43:12:CPUs: 11th Gen Intel(R) Core(TM) i5-11400 @ 2.60GHz (Intel)
20:43:12: 1 CPU x 6 cores. 12 logical processors (x64)
20:43:12:GPU: Intel(R) UHD Graphics 730 (1,024 MiB) (Intel Corporation)
20:43:12: Driver: 30.0.101.1273
20:43:12:System RAM: 16 GiB
20:43:12:Target: Windows
20:43:12:BuildConfig: Release
20:43:12:Execution Env: Native
20:43:12:Runtime Env: Production
20:43:12:.NET framework: .NET 7.0.5
20:43:12:App DataDir: C:\ProgramData\CodeProject\AI
20:43:12:Video adapter info:
20:43:12: Intel(R) UHD Graphics 730:
20:43:12: Driver Version 30.0.101.1273
20:43:12: Video Processor Intel(R) UHD Graphics Family
20:43:12:ROOT_PATH = C:\Program Files\CodeProject\AI
20:43:12:RUNTIMES_PATH = C:\Program Files\CodeProject\AI\runtimes
20:43:12:PREINSTALLED_MODULES_PATH = C:\Program Files\CodeProject\AI\preinstalled-modules
20:43:12:MODULES_PATH = C:\Program Files\CodeProject\AI\modules
20:43:12:PYTHON_PATH = \bin\windows\%PYTHON_RUNTIME%\venv\scripts\Python
20:43:12:Data Dir = C:\ProgramData\CodeProject\AI
20:43:12:Server version: 2.1.10-Beta
20:43:12:ModuleRunner Start
20:43:12:Starting Background AI Modules
20:43:14:Client request 'list-custom' in queue 'objectdetection_queue' (...e78435)
20:43:15:GetCommandByRuntime: Runtime=python37, Location=Local
20:43:15:Command: C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral\bin\windows\python37\venv\scripts\Python
20:43:15:
20:43:15:Starting C:\Program Files...ws\python37\venv\scripts\Python "C:\Program Files...ectdetection_coral_adapter.py"
20:43:15:Attempting to start ObjectDetectionCoral with C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral\bin\windows\python37\venv\scripts\Python "C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral\objectdetection_coral_adapter.py"
20:43:15:
20:43:15:Module 'ObjectDetection (Coral)' (ID: ObjectDetectionCoral)
20:43:15:Module Path: C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral
20:43:15:AutoStart: True
20:43:15:Queue: objectdetection_queue
20:43:15:Platforms: windows,linux,linux-arm64,macos,macos-arm64
20:43:15:GPU: Support enabled
20:43:15:Parallelism: 1
20:43:15:Accelerator:
20:43:15:Half Precis.: enable
20:43:15:Runtime: python37
20:43:15:Runtime Loc: Local
20:43:15:FilePath: objectdetection_coral_adapter.py
20:43:15:Pre installed: False
20:43:15:Start pause: 1 sec
20:43:15:LogVerbosity:
20:43:15:Valid: True
20:43:15:Environment Variables
20:43:15:MODELS_DIR = %CURRENT_MODULE_PATH%/assets
20:43:15:MODEL_SIZE = Medium
20:43:15:
20:43:15:Started ObjectDetection (Coral) module
20:43:23:Client request 'detect' in queue 'objectdetection_queue' (...abbef9)
20:43:24:Client request 'detect' in queue 'objectdetection_queue' (...ddda08)
20:43:24:Client request 'detect' in queue 'objectdetection_queue' (...9e7279)
20:43:30:objectdetection_coral_adapter.py: E driver/mmio_driver.cc:254] HIB Error. hib_error_status = ffffffffffffffff, hib_first_error_status = ffffffffffffffff
20:43:43:objectdetection_coral_adapter.py: E driver/mmio_driver.cc:254] HIB Error. hib_error_status = ffffffffffffffff, hib_first_error_status = ffffffffffffffff
20:43:43:objectdetection_coral_adapter.py: CPAI_MODULE_REQUIRED_MB not found. Setting to default 0
20:43:43:objectdetection_coral_adapter.py: NUM_THREADS not found. Setting to default 1
20:43:43:objectdetection_coral_adapter.py: MIN_CONFIDENCE not found. Setting to default 0.5
20:43:43:objectdetection_coral_adapter.py: MODULE_PATH: C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral
20:43:43:objectdetection_coral_adapter.py: MODELS_DIR: C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral\assets
20:43:43:objectdetection_coral_adapter.py: MODEL_SIZE: medium
20:43:43:objectdetection_coral_adapter.py: Timeout connecting to the server
20:43:43:objectdetection_coral_adapter.py: ObjectDetection (Coral) started.ObjectDetection (Coral): ObjectDetection (Coral) started.
20:43:43:objectdetection_coral_adapter.py: CPU_MODEL_NAME: efficientdet_lite3_512_ptq.tflite
20:43:43:objectdetection_coral_adapter.py: TPU_MODEL_NAME: efficientdet_lite3_512_ptq_edgetpu.tflite
20:43:43:objectdetection_coral_adapter.py: Input details: {'name': 'serving_default_images ', 'index': 0, 'shape': array([ 1, 512, 512, 3]), 'shape_signature': array([ 1, 512, 512, 3]), 'dtype': , 'quantization': (0.0078125, 127), 'quantization_parameters': {'scales': array([0.0078125], dtype=float32), 'zero_points': array([127]), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
20:43:43:objectdetection_coral_adapter.py: Output details: {'name': 'StatefulPartitionedCall:31', 'index': 931, 'shape': array([ 1, 25, 4]), 'shape_signature': array([ 1, 25, 4]), 'dtype': , 'quantization': (0.0, 0), 'quantization_parameters': {'scales': array([], dtype=float32), 'zero_points': array([], dtype=int32), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
20:43:43:ObjectDetection (Coral): ObjectDetection (Coral) started.
20:43:43:Request 'list-custom' dequeued from 'objectdetection_queue' (...e78435)
20:43:43:Request 'detect' dequeued from 'objectdetection_queue' (...abbef9)
20:43:43:Request 'detect' dequeued from 'objectdetection_queue' (...ddda08)
20:43:43:Request 'detect' dequeued from 'objectdetection_queue' (...9e7279)
20:43:43:ObjectDetection (Coral): Retrieved objectdetection_queue command
20:43:43:ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'list-custom' (...e78435) took 1ms
20:43:43:Response received (...e78435)
20:43:43:ObjectDetection (Coral): Retrieved objectdetection_queue command
20:43:43:ObjectDetection (Coral): Retrieved objectdetection_queue command
20:43:43:ObjectDetection (Coral): Retrieved objectdetection_queue command
20:43:44:ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (...abbef9) took 1027ms
20:43:44:Response received (...abbef9): The interpreter is in use. Please try again later
20:43:44:ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (...ddda08) took 1041ms
20:43:44:Response received (...ddda08): The interpreter is in use. Please try again later
20:44:17:ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (...9e7279) took 33493ms
|
|
|
|
|
No solutions yet?
Sorry not been around for a few days, been in hospital.
|
|
|
|
|
Hello,
i have an M.2 Coral installed in my Server running Proxmox.
I created a VM with the Coral PCI Device attached.
Installed all Coral Drivers and Modules inside the VM.
Installed Docker and attached the device to the Docker Container.
Installed ObjectDetection (Coral)
The M.2 Coral is attached to the VM according lspci inside the Container:
00:10.0 System peripheral: Global Unichip Corp. Coral Edge TPU
Starting the Coral ObjectDetection tells me "Edge TPU detected" :
08:20:18:Running init for ObjectDetection (Coral)
08:20:19:objectdetection_coral_adapter.py: CPAI_MODULE_REQUIRED_MB not found. Setting to default 0
08:20:19:objectdetection_coral_adapter.py: NUM_THREADS not found. Setting to default 1
08:20:19:objectdetection_coral_adapter.py: MIN_CONFIDENCE not found. Setting to default 0.5
08:20:19:objectdetection_coral_adapter.py: MODULE_PATH: /app/modules/ObjectDetectionCoral
08:20:19:objectdetection_coral_adapter.py: MODELS_DIR: /app/modules/ObjectDetectionCoral/assets
08:20:19:objectdetection_coral_adapter.py: MODEL_SIZE: medium
08:20:19:objectdetection_coral_adapter.py: CPU_MODEL_NAME: efficientdet_lite3_512_ptq.tflite
08:20:19:objectdetection_coral_adapter.py: TPU_MODEL_NAME: efficientdet_lite3_512_ptq_edgetpu.tflite
08:20:19:objectdetection_coral_adapter.py: Edge TPU detected
08:20:19:objectdetection_coral_adapter.py: Input details: {'name': 'serving_default_images:0', 'index': 0, 'shape': array([ 1, 512, 512, 3], dtype=int32), 'shape_signature': array([ 1, 512, 512, 3], dtype=int32), 'dtype': , 'quantization': (0.0078125, 127), 'quantization_parameters': {'scales': array([0.0078125], dtype=float32), 'zero_points': array([127], dtype=int32), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
08:20:19:objectdetection_coral_adapter.py: Output details: {'name': 'StatefulPartitionedCall:31', 'index': 23, 'shape': array([ 1, 25, 4], dtype=int32), 'shape_signature': array([ 1, 25, 4], dtype=int32), 'dtype': , 'quantization': (0.0, 0), 'quantization_parameters': {'scales': array([], dtype=float32), 'zero_points': array([], dtype=int32), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
08:20:19:ObjectDetection (Coral): ObjectDetection (Coral) started.
ObjectDetection (Coral) is Starting but using CPU
Everything seems to work properly but detecting does nothing.
Are im missing something here?
Systeminfo:
Server version: 2.1.11-Beta
Operating System: Linux (Linux 5.10.0-25-amd64 #1 SMP Debian 5.10.191-1 (2023-08-16))
CPUs: QEMU Virtual CPU version 2.5+
1 CPU x 1 core. 1 logical processors (x64)
System RAM: 4 GiB
Target: Linux
BuildConfig: Release
Execution Env: Docker
Runtime Env: Production
.NET framework: .NET 7.0.10
System GPU info:
GPU 3D Usage 0%
GPU RAM Usage 0
Video adapter info:
Global Environment variables:
CPAI_APPROOTPATH = /app
CPAI_PORT = 32168
Serverlogs:
24:27:36:Operating System: Linux (Linux 5.10.0-25-amd64 #1 SMP Debian 5.10.191-1 (2023-08-16))
24:27:36:CPUs: QEMU Virtual CPU version 2.5+
24:27:36: 1 CPU x 1 core. 1 logical processors (x64)
24:27:36:System RAM: 4 GiB
24:27:36:Target: Linux
24:27:36:BuildConfig: Release
24:27:36:Execution Env: Docker
24:27:36:Runtime Env: Production
24:27:36:.NET framework: .NET 7.0.10
24:27:36:App DataDir: /etc/codeproject/ai
24:27:36:Video adapter info:
24:27:36:ROOT_PATH = /app
24:27:36:RUNTIMES_PATH = /app/runtimes
24:27:36:PREINSTALLED_MODULES_PATH = /app/preinstalled-modules
24:27:36:MODULES_PATH = /app/modules
24:27:36:PYTHON_PATH = /bin/linux/%PYTHON_RUNTIME%/venv/bin/python3
24:27:36:Data Dir = /etc/codeproject/ai
24:27:36:Server version: 2.1.11-Beta
24:27:36:ModuleRunner Start
24:27:36:Overriding address(es) 'http://+:32168, http://+:5000'. Binding to endpoints defined via IConfiguration and/or UseKestrel() instead.
24:27:36:Starting Background AI Modules
24:27:39:GetCommandByRuntime: Runtime=python39, Location=Local
24:27:39:Command: /app/modules/ObjectDetectionCoral/bin/linux/python39/venv/bin/python3
24:27:39:
24:27:39:Attempting to start ObjectDetectionCoral with /app/modules/ObjectDetectionCoral/bin/linux/python39/venv/bin/python3 "/app/modules/ObjectDetectionCoral/objectdetection_coral_adapter.py"
24:27:39:Starting /app...onCoral/bin/linux/python39/venv/bin/python3 "/app...ionCoral/objectdetection_coral_adapter.py"
24:27:39:
24:27:39:Module 'ObjectDetection (Coral)' (ID: ObjectDetectionCoral)
24:27:39:Module Path: /app/modules/ObjectDetectionCoral
24:27:39:AutoStart: True
24:27:39:Queue: objectdetection_queue
24:27:39:Platforms: windows,linux,linux-arm64,macos,macos-arm64
24:27:39:GPU: Support enabled
24:27:39:Parallelism: 1
24:27:39:Accelerator:
24:27:39:Half Precis.: enable
24:27:39:Runtime: python39
24:27:39:Runtime Loc: Local
24:27:39:FilePath: objectdetection_coral_adapter.py
24:27:39:Pre installed: False
24:27:39:Start pause: 1 sec
24:27:39:LogVerbosity:
24:27:39:Valid: True
24:27:39:Environment Variables
24:27:39:MODELS_DIR = %CURRENT_MODULE_PATH%/assets
24:27:39:MODEL_SIZE = Medium
24:27:39:
24:27:39:Started ObjectDetection (Coral) module
24:27:41:Current Version is 2.1.11-Beta
24:27:41:Server: This is the latest version
24:27:42:Running init for ObjectDetection (Coral)
24:27:42:objectdetection_coral_adapter.py: CPAI_MODULE_REQUIRED_MB not found. Setting to default 0
24:27:42:objectdetection_coral_adapter.py: NUM_THREADS not found. Setting to default 1
24:27:42:objectdetection_coral_adapter.py: MIN_CONFIDENCE not found. Setting to default 0.5
24:27:42:objectdetection_coral_adapter.py: MODULE_PATH: /app/modules/ObjectDetectionCoral
24:27:42:objectdetection_coral_adapter.py: MODELS_DIR: /app/modules/ObjectDetectionCoral/assets
24:27:42:objectdetection_coral_adapter.py: MODEL_SIZE: medium
24:27:42:objectdetection_coral_adapter.py: CPU_MODEL_NAME: efficientdet_lite3_512_ptq.tflite
24:27:42:objectdetection_coral_adapter.py: TPU_MODEL_NAME: efficientdet_lite3_512_ptq_edgetpu.tflite
24:27:42:objectdetection_coral_adapter.py: Edge TPU detected
24:27:42:objectdetection_coral_adapter.py: Input details: {'name': 'serving_default_images:0', 'index': 0, 'shape': array([ 1, 512, 512, 3], dtype=int32), 'shape_signature': array([ 1, 512, 512, 3], dtype=int32), 'dtype': , 'quantization': (0.0078125, 127), 'quantization_parameters': {'scales': array([0.0078125], dtype=float32), 'zero_points': array([127], dtype=int32), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
24:27:42:objectdetection_coral_adapter.py: Output details: {'name': 'StatefulPartitionedCall:31', 'index': 23, 'shape': array([ 1, 25, 4], dtype=int32), 'shape_signature': array([ 1, 25, 4], dtype=int32), 'dtype': , 'quantization': (0.0, 0), 'quantization_parameters': {'scales': array([], dtype=float32), 'zero_points': array([], dtype=int32), 'quantized_dimension': 0}, 'sparsity_parameters': {}}
24:27:42:ObjectDetection (Coral): ObjectDetection (Coral) started.
24:28:03:Client request 'list-custom' in queue 'objectdetection_queue' (...bd74c6)
24:28:03:Client request 'list-custom' in queue 'objectdetection_queue' (...6c8e5b)
24:28:09:Client request 'detect' in queue 'objectdetection_queue' (...83649e)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...bd1107)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...e3cb2f)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...4683a6)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...ac0487)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...b764db)
24:28:54:Client request 'detect' in queue 'objectdetection_queue' (...8a0ade)
08:02:39:Sending shutdown request to python3/ObjectDetectionCoral
08:02:39:Client request 'Quit' in queue 'objectdetection_queue' (...005acd)
08:03:12:Forcing shutdown of python3/ObjectDetectionCoral
08:03:12:Waiting for ObjectDetectionCoral to end.
08:03:12:Module ObjectDetectionCoral has shutdown
|
|
|
|
|
Unfortunately I can't help with proxmox isues, but have you tried increasing the amount of RAM from 4 to 8Gb? Just a thought.
cheers
Chris Maunder
|
|
|
|
|
It's not a proxmox problem. I can use the coral tpu inside the same vm. I attached the tpu test wise with a Frigate NVR Docker Container.
I can try adding more ram.
|
|
|
|
|
I had a rough time trying to get the Coral to work. Sometimes everything would report as being OK, but things like loading the delegates from the edgetpu libraries would just hang or crash the module. And then some days it would just...work.
Things that seemed to help for me was plugging devices in before starting the module, and if there were issues, shutting down, unplugging, then plugging into a different USB port (I'm using the USB Coral) or using a different cable. Not solutions for you, unfortunately, but I definitely did find it fiddly.
Except on the Raspberry Pi. It's always been rock solid on the Pi. So weird.
cheers
Chris Maunder
|
|
|
|
|
Sorry for this late response. My Family got the flue one by one...
So Adding more RAM did finally do the Trick.
I did build my own Image with libedgetpu1-max and some drivers for M.2.
It runs with about 100 - 120 ms Roundtrip.
|
|
|
|
|
If you can share your working code I'll integrate this into the module for all to enjoy
Well done! (and sorry to hear about your family. Ah, flu season...)
cheers
Chris Maunder
|
|
|
|
|
Are you sharing the TPU with Frigate? I'm unable to get the Coral to work with both Frigate and CPAI at the same time. As soon as I shut down Frigate, it came right up in CPAI.
|
|
|
|
|
No! That's in my Knowledge technically impossible beacause you can map /dev/apex only once.
I did just test the Docker Setup with a Frigate Container to verify that the M.2 Coral Hardware could be used in Docker)
|
|
|
|
|
Well, you can normally map devices to more than one container. That’s one of reasons to use Docker over the more traditional virtual machine route.
For example, it’s extremely common to map the GPU to multiple containers. You can’t do that with a VM unless the host and GPU support SR-IOV.
|
|
|
|
|
Uninstall Coral Objectdetection, execute a shell in the docker container, install python3.9-venv, and reinstall Coral Objectdetection
apt-get install python3.9-venv
|
|
|
|
|
Hi, I cant start Coral is come up this, is there anyway I can fix it? thanks
02:35:48:Started ObjectDetection (Coral) module
02:35:48:objectdetection_coral_adapter.py: Traceback (most recent call last):
02:35:48:objectdetection_coral_adapter.py: File "C:\Program Files\CodeProject\AI\modules\ObjectDetectionCoral\objectdetection_coral_adapter.py", line 9, in
02:35:48:objectdetection_coral_adapter.py: from request_data import RequestData
02:35:48:objectdetection_coral_adapter.py: File "../../SDK/Python\request_data.py", line 8, in
02:35:48:objectdetection_coral_adapter.py: from PIL import Image
02:35:48:objectdetection_coral_adapter.py: ImportError: cannot import name 'Image' from 'PIL' (unknown location)
02:35:48:Module ObjectDetectionCoral has shutdown
02:35:48:objectdetection_coral_adapter.py: has exited
|
|
|
|
|
It looks like the installation failed. I would suggest uninstalling and then reinstalling the coral module
cheers
Chris Maunder
|
|
|
|
|
I have a new Installation of BlueIris 5.x on Windows 10, with CPAI v2.1.10 and a USB Coral TPU running as a Windows service. Object detections were averaging about 103ms consistently.
I upgraded to v2.1.11, and immediately the detection time jumped 10x to about 1030ms! Many entries like this:
17 13:20:03: ObjectDetection (Coral): Retrieved objectdetection_queue command in ObjectDetection (Coral)
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid 484224d3-0f27-4544-addf-4efd3867333f) took 1110ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid 007c37e5-bb77-4b6e-99a8-0972236a7b69) took 1131ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid e708f751-36df-460e-a981-58a8601e4db2) took 1128ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: Response received (#reqid 007c37e5-bb77-4b6e-99a8-0972236a7b69): The interpreter is in use. Please try again later
2023-08-17 13:20:04: Response received (#reqid 484224d3-0f27-4544-addf-4efd3867333f): The interpreter is in use. Please try again later
2023-08-17 13:20:04: Response received (#reqid e708f751-36df-460e-a981-58a8601e4db2): The interpreter is in use. Please try again later
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid bdf9ed8d-f857-4c53-8b07-97d9e8e46ac2) took 1122ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid d522b233-b83f-4ae1-a552-315570d56094) took 1145ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: Response received (#reqid c3df76c7-4642-424a-a32e-9c4db9d2d816): The interpreter is in use. Please try again later
2023-08-17 13:20:04: Response received (#reqid 13d53023-1b7e-4ff9-8917-85800572fb83): The interpreter is in use. Please try again later
2023-08-17 13:20:04: ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid 13d53023-1b7e-4ff9-8917-85800572fb83) took 1088ms (command timing) in ObjectDetection (Coral)
2023-08-17 13:20:04: Response received (#reqid bdf9ed8d-f857-4c53-8b07-97d9e8e46ac2): The interpreter is in use. Please try again later
2023-08-17 13:20:04: Response received (#reqid d522b233-b83f-4ae1-a552-315570d56094): The interpreter is in use. Please try again later
I of course uninstalled 2.1.11 and reinstalled 2.1.10 - but the slow detection times and the 'interpreter is in use' messages persist.
Suggestions?
Thanks,
JDC
|
|
|
|
|
I encountered the same issue when setting this up for the first time yesterday, BI 5.x on Windows 10, CPAI v2.1.10 with Coral TPU but running on a RPi4B 2Gb via the docker image.
No answer I'm afraid as I'm in the midst of troubleshooting. I've currently throttled it back from 4 cameras using AI to 1 camera to see if it's the volume of requests causing the issue.
My only observation so far is that I'm seeing a lot more CPAI requests from BI than expected even when there's no motion. These requests are originating from the Static Object analysis which appears to be stuck on the more frequent analysis after a scene change (20 sec by default) rather than the primary analysis frequency (120 sec by default), I've got an open question on the BI forum regarding this.
|
|
|
|
|
I just got this on Docker on my x86 server after about 32 hours of runtime. So it's across platforms. Restarting the Docker container was enough to solve it (temporarily) for me.
Also, when it's doing those 1000ms analyses, it's not actually doing any analysis. Unfortunately, the API doesn't seem to recognize these as failures and simply returns a result as if the submitted image was really analyzed and nothing was found.
codeprojectai | 2023-08-18T14:27:29.547644909Z Trace Client request 'detect' in queue 'objectdetection_queue' (#reqid 0dc53ec2-f14f-4445-a0da-221b3739c00d)
codeprojectai | 2023-08-18T14:27:29.547663423Z Trace Request 'detect' dequeued from 'objectdetection_queue' (#reqid 0dc53ec2-f14f-4445-a0da-221b3739c00d)
codeprojectai | 2023-08-18T14:27:29.549420690Z Debug ObjectDetection (Coral): Retrieved objectdetection_queue command
codeprojectai | 2023-08-18T14:27:30.451766346Z Infor ObjectDetection (Coral): Queue request for ObjectDetection (Coral) command 'detect' (#reqid 68e21248-d17a-47f3-887a-0339a2dd22f0) took 1014ms
codeprojectai | 2023-08-18T14:27:30.452620714Z Trace Response received (#reqid 68e21248-d17a-47f3-887a-0339a2dd22f0): The interpreter is in use. Please try again later
modified 18-Aug-23 15:09pm.
|
|
|
|
|
Thanks very much for the report. When you upgraded, did you just update the server from 2.1.10 to 2.1.11 or did you uninstall and reinstall the Coral module as well?
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
I did not change anything related to the TPU, just ran the update to 2.1.11. When I encountered the issue, I uninstalled 2.1.11 and then reinstalled 2.1.10.
Thanks,
JDC
|
|
|
|
|
Does the problem persist now that you're back on 2.1.10 or is it still slow?
Also, have you tried changing the model size in the settings (“… -> Model size -> try ’small’)
Thanks,
Sean Ewington
CodeProject
|
|
|
|
|
Yes the speed remains about 10x slower even on 2.1.10, after reverting from 2.1.11.
Model size was already set to small, I also tried tiny - same response times.
JDC
|
|
|
|
|
I just tried stopping Blue Iris and Codeproject .AI, then uninstalling the edgeTPU runtime package using the uninstall.bat in the 20221024 package. I then ran the install.bat and started everything... same slow response times.
KDC
|
|
|
|
|
I'm not an expert on Blue Iris but there is a setting that allows you to choose the size of the images that are passed to CodeProject.AI. As an experiment would it be possible to try passing different sized images to CodeProject.AI using the Explorer (blue link in the CodeProject.AI dashboard at the top) and see if that affects inference speed.
It would also be interesting to see (if you experiment with the explorer) if the image size affects total processing speed and inference or just total processing (the values will appear in the results window in the explorer)
cheers
Chris Maunder
|
|
|
|
|