NCS2 in RPi 3B+ - RuntimeError: Cannot create ShapeOf layer fc7_mbox_priorbox/0_port id:192

[ananya_nrt]

I am trying to run a face recognition python script which is loading four other models for complete execution.

I am using Raspberry Pi 3B+ (Raspbian Stretch), NCS2 stick with OpenVINO 2020 R3. 

The main script is calling and loading the other models on execution using the NCS2 stick.

I am getting the error:

RuntimeError: Cannot create ShapeOf layer fc7_mbox_priorbox/0_port id:192

I am using IE Network and Plugin instead of IE Core. Can that be an issue?

Thanks in advance.

[IntelSupport]

Hi Ananya,

Thank you for reaching out to us. We would suggest our customers use IE Core instead of using IE Plugin and IE Network. This is due to IE Plugin has been removed in the 2021.1 version. For future releases, IE Network will be removed as well. As additional information, you can refer to the following link to view the deprecated list:

https://docs.openvinotoolkit.org/2020.4/deprecated.html

Having said that, could you install the OpenVINO toolkit 2021.1 version on the raspberry pi and try to execute the program? Please refer to the following link to download the latest version of the OpenVINO toolkit for Raspbian OS:

https://download.01.org/opencv/2021/openvinotoolkit/2021.1/

If it doesn’t work, we need your cooperation to share additional information regarding:

• Models used in the program
• Python script used in the program

Regards,

Adli

[ananya_nrt]

Hi Adli,

I am in the process of migrating from IE Plugin and IE Network to IE Core already. However, for the OpenVINO toolkit version, I have a compulsion to use OpenVINO toolkit 2020.x, so that is a dead-end for me.

Also, the models that I am using are from the OpenVINO toolkit pre-trained models. I am using:

• face-detection-adas-0001
• face-detection-retail-0004
• face-reidentitfication-retail-0095
• head-pose-estimation-adas-0001
• landmarks-regression-retail-0009

My python script is loading all the models onto the NCS2 stick and using the Pi Camera feed as the input for these models.

Please let me know if this information will suffice.

Thanks.

[ananya_nrt]

Spoiler

We would suggest our customers use IE Core instead of using IE Plugin and IE Network. This is due to IE Plugin has been removed in the 2021.1 version. 

Also, since support for IE Plugin has been pulled off from OpenVINO toolkit 2020.1 version, so IE Plugin should still be running for OpenVINO toolkit 2020.x version?

[IntelSupport · Accepted solution]

Hi Ananya,

IE Plugin has been deprecated starting from the OpenVINO toolkit 2021.1 version onwards. For Raspbian OS, we suggest you use the 2020.4 version of the OpenVINO toolkit. Please refer to the following link to download the 2020.4 release:

https://download.01.org/opencv/2020/openvinotoolkit/2020.4/

Regards,

Adli

[ananya_nrt]

Hi Adli,

I think I will take this up, switching to OpenVINO toolkit v2020.4. 

Thanks a lot!

[IntelSupport]

Hi Ananya,

This thread will no longer be monitored since this issue has been resolved. If you need any additional information from Intel, please submit a new question.

Best Regards,

Adli