Openvino error:expected: 'pin.oid < (int)netInputLayer.shapes.size()

[AyasaYasca]

I tried to make a program about FaceMask-Detection.Before,my programme could run successfully with the caffe model.However,When I turn caffe to ir model,it has problems as follows.

gdb debug:

i guess the problem is model'format.but i cant sovle it

pls help me.

[Wan_Intel]

Hi AyasaYasca,

Thanks for reaching out to us.

Could you please share the following details with us for further investigation?

·      Which version of the OpenVINO™ toolkit are you using?

·      Hardware specifications

·      Are you using custom demo application or Open Model Zoo Demo?

·      Are you using custom model or Open Model Zoo Demo?

Regards,

Wan

[AyasaYasca]

Hello wan:

Thank you for your prompt reply.

Run -environment:

rasberry3B+

The version of openvino:

l_openvino_toolkit_runtime_raspbian_p_2020.4.287.tgz

------------------------------------------------------------------

Both  caffe and ir  model are here.

If you want the source code later, I can provide it to you.

[Wan_Intel]

Hi AyasaYasca,

I’ve checked your IR model with OpenVINO™ Benchmark Python tool as shown as follows:

 

Did you create your application from scratch? You can share it here for further investigation.

Regards,

Wan

[AyasaYasca]

Hello wan:

Here is my program.

Thanks for your help. ovo

[Wan_Intel]

Hi AyasaYasca,

I noticed that you are using CPU plugin in your application (Line 52).

   this->net.setPreferableTarget(dnn::DNN_TARGET_CPU);

For your information, CPU plugin is not supported by OpenVINO™ toolkit for Raspbian Operating System. You must use Intel® Neural Compute Stick 2 with Raspberry Pi and use MYRIAD plugin instead of CPU plugin.

Please refer to Development and Target Platforms in Install OpenVINO™ toolkit for Raspbian OS and Supported Devices for more information.

Regards,

Wan

[AyasaYasca]

Hi Wan,

Thank for your patient reply.

Although I obey your order to change the error code ,it still has problems:

firstly, I changed the parament ,it still has error

 

secendly, I deleted the code shown above .But the else error occur

I guess the reason as follows:

the first one is the incorrect version of the openvino

another is the false ir format

Because English is not my first language. My expression may confuse you ..... 

Looking forward to your reply.

[Wan_Intel]

Hi AyasaYasca,

Do you use Intel® Neural Compute Stick 2 with Raspberry Pi when changing the plugin to MYRIAD plugin?

Regards,

Wan

[AyasaYasca]

Hello wan,

Here

  

[AyasaYasca]

add:

  

[Wan_Intel]

Hi AyasaYasca,

I noticed that the error you encountered:

“'inputShapeLimitation.size()' is 2 must be equal to 'blobShape.size()' is 4”

 was due to incompatible sizes used which related to the blob size. You may refer to here for more information.

I suggest that you use OpenVINO™ Runtime API to Implement Inference Pipeline to integrate OpenVINO™ with your application.

You may refer to the following inference engine samples:

·      Hello Classification C++ Sample

·      Image Classification Async C++ Sample

Regards,

Wan

[Wan_Intel]

Hi AyasaYasca,

Thanks for your question.

This thread will no longer be monitored since we have provided information. 

If you need any additional information from Intel, please submit a new question.

Best regards,

Wan