Error with Intel® Neural Compute Stick, USB port stops work

[idata]

Dears,

We are developing a solution of analytic video with Intel® Neural Compute Stick on an upboard 2 box with ubuntu 16.04 LTS, but eventually we have a USB port blocked with this error:

cv2.error: OpenCV(4.0.1-openvino) /home/jenkins/workspace/OpenCV/OpenVINO/build/opencv/modules/dnn/src/op_inf_engine.cpp:553: error: (-215:Assertion failed) Failed to initialize Inference Engine backend: Can not init USB device: NC_DEVICE_NOT_FOUND in function 'initPlugin'`

After that, movidius is unavailable and we need to do a physical reboot.

We try to restart usb port, but unsuccessfully. Somebody had same problem?

Thank you in advanced.

Alonso.

[idata]

Hi @abgenis

Do you see this error when you run other examples as well? Are you using a USB powered hub or plugging directly in?

You might need to reset the NCS device within OpenVINO. There is a parameter KEY_VPU_FORCE_RESET that resets any stalled devices on plugin initialization, it must be used with SetConfig method. Here are some links to help guide in the right direction:

Myriad Plugin, Supported Configuration Parameters

https://docs.openvinotoolkit.org/2018_R5/_docs_IE_DG_supported_plugins_MYRIAD.htmlVPU Plugins, VPU Common Configuration Parametershttps://docs.openvinotoolkit.org/2018_R5/_docs_IE_DG_supported_plugins_VPU.html

Please let me know if this is helpful.

Best Regards,

Sahira

[idata]

Hi @Sahira_at_Intel

I'm working together with @abgenis.

About your questions, it follows:

Do you see this error when you run other examples as well?

Yes, if an example runs in a stressful manner for a certain period, this problem occurs.

Are you using a USB powered hub or plugging directly in?

No, we connected the NCS directly to the USB 3.0 port of an UpBoard 2

On redefining the device with KEY_VPU_FORCE_RESET we did this via python, follow code snippet in python3.5

from openvino.inference_engine import IENetwork, IEPlugin

plugin = IEPlugin (device = "MIRYAD")plugin.set_config ({"VPU_FORCE_RESET": "NO"})

NOTE: In python the KEY_VPU_FORCE_RESET parametyro does not exist, then we use VPU_FORCE_RESET

We do this at the beginning of the code, but we do not succeed. The rest of the instructions are made through opencv 4.0-1-openvino. Follow the passage

import cv2

netMovidius = cv2.dnn.readNet ('my_model.bin', 'my_model.xml')netMovidius.setPreferableTarget (cv2.dnn.DNN_TARGET_MYRIAD)frame = cv2.imread ('test_images / image1.jpg'blob = cv2.dnn.blobFromImage (frame, size = (300, 300))netMovidius.setInput (blob)out = netMovidius.forward ()

After starting the inference in the image, we get the error that @abgenis reported above

[idata]

Hi @RodolfoPlanetfone, @abgenis

Which version of OpenVINO are you using? I'd like to try to reproduce this issue, can you tell me which example you were running and for how long when you ran into the error.

Sincerely,

Sahira

[idata]

Hi @Sahira_at_Intel

We are using openVINO version computer_vision_sdk_2018.5.455.

Our example is based on converting a pre-trained model to use in openVino in multiple threads for detection in several cameras. We control the frames that we send through a queue, so there is no overhead in the movidius.

Our script usually lasts 7 to 14 days, after that time, we can no longer connect to the movidius inside the operating system, this requires that we turn off the system or change the USB port movidius.

[idata]

Hi @RodolfoPlanetfone

Please upgrade to the latest version of OpenVINO 2019 r1.1 here and try again.

Best Regards,

Sahira

[idata]

Hi, @Sahira_at_Intel,

Thanks for the answer! We will run tests with this version and return with a feedback.