Error in loding IR which has dynamic_axis in original onnx model

[TarunM]

I successfully converted my ONNX model to IR, but I am now getting error while reading the network in Inference Engine.

    dec_net = ie.read_network(model=dec_xml, weights=dec_bin)
  File "ie_api.pyx", line 293, in openvino.inference_engine.ie_api.IECore.read_network
  File "ie_api.pyx", line 315, in openvino.inference_engine.ie_api.IECore.read_network
RuntimeError: low_alpha has zero dimension which is not allowed

The orginal onnx model has dynamic_axes  at index 1for inputs low_alpha and high_alpha. 

I need dynamic axes because input shapes are variable for these two parametes.

I have tested the model works perfectly fine using ONNX model and onnuxruntime. So there is something wrong with openvino or I am doing something wron. Please help.

Attached are the IR and onnx

[IntelSupport]

Hi TarunM,

Thanks for reaching out. I have tested your model on Benchmark application and encountered the same issue. We are still investigating this and will get back to you with the information soon.

Regards,

Aznie

[IntelSupport]

Hi TarunM,

Thank you for your patience. Unfortunately, we have confirmed with the developer that dynamic shapes/axes using OpenVINO are yet to be supported in the current version.

Regards,

Aznie

[TarunM]

I just saw the newer version of OpenVino is out (2021.4) . How to solve this problem using that?

[IntelSupport]

Hi TarunM,

We apologize to you. These dynamic shapes/axes features are not yet supported in the 2021.4 release and targeted to be released in OpenVINO 2022.1.

Regards,

Aznie

[IntelSupport]

Hi TarunM,

This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question.

Regards,

Aznie