Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6430 Discussions

does old version(2021.3) of the openvino toolkit support dynamic shape model converting?

cmkk
Beginner
702 Views

I recently take over an project which use 2021.3(one of the older version) openvino as its inference engine, and now I have a dynamic shape onnx model that need to be converted into an ir model so it can be used in the project.

After I check the documatation, I found out that for latest version, you can simply set --input_shape with  -1 for those unknow or uncertain dims, but before the 2021.3version, it seemed that you can not do it in this way, it will throw a "not fully defined" error for input shape. with its documentation says as such:  https://docs.openvino.ai/2021.3/openvino_docs_MO_DG_prepare_model_convert_model_Converting_Model_General.html#:~:text=The%20shape%20should%20not%20contain%20undefined

My question is: does the 2021.3 version of openvino support dynamic shape model converting? it does, how to do it with "python mo.py", mean what extra params should I be setting?

 

thanks in advance for any suggestion or insights.

0 Kudos
2 Replies
Peh_Intel
Moderator
660 Views

Hi cmkk,


Thanks for reaching out to us.


Yes, you are right. Directly specifying input_shape -1 for the dynamic dimension is only available starting from Model Optimizer 2022.1.


The previous version of Model Optimizer does not have this additional feature.



Regards,

Peh


0 Kudos
Peh_Intel
Moderator
641 Views

Hi cmkk,


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



Regards,

Peh


0 Kudos
Reply