Unable to run conversion for a tensorflow "saved_model" format for inferance

[Petrica]

I have a problem running when trying to use the convertion script from the OpenVINO toolkit.

I am running the script with the following parameters:

mo_tf.py --saved_model_dir ./test/saved_model --transformations_config /opt/intel/openvino_2021/deployment_tools/model_optimizer/extensions/front/tf/ssd_support_api_v2.4.json --tensorflow_object_detection_api_pipeline_config ./test/pipeline.config --data_type FP16

End it ends up with an error :

"Exception occurred during running replacer "ObjectDetectionAPIOutputReplacement" (<class 'extensions.front.tf.ObjectDetectionAPI.ObjectDetectionAPIOutputReplacement'>): Graph contains 1 node after executing <class 'extensions.front.tf.ObjectDetectionAPI.ObjectDetectionAPIOutputReplacement'>. It considered as error because resulting IR will be empty which is not usual"

I want to specify that I've used in the past the script with success with the same pipeline config file.

I will attached the pipeline config and a screenshot of the error.

[Wan_Intel]

Hi Petrica,

Thanks for reaching out to us.

May I know which version of the OpenVINO™ toolkit are you using? For your information, OpenVINO™ Development Tools version 2022.1 is available now. You may refer to here to install the latest OpenVINO™ Development Tools and convert your model again.

On another note, if you still facing issue when using the latest OpenVINO™ Development Tools to convert your model, could you please share your model with us for further investigation?

Regards,

Wan

[Petrica]

I am still facing with the same issue Iven after installing the newer version. I will attach the saved model folder in this section, Thank you.

[Wan_Intel]

Hi Petrica,

Thanks for your information.

We are now investigating this issue, and we will update you as soon as possible.

Regards,

Wan

[Wan_Intel · Accepted solution]

Hi Petrica,

Thanks for your patience.

I have successfully converted the model into IR with the following command:

mo --saved_model_dir=saved_model --tensorflow_object_detection_api_pipeline_config=pipeline.config --framework=tf --reverse_input_channels

Could you please try it and see if you are able to convert the model into IR?

Regards,

Wan

[Wan_Intel]

Hi Petrica,

Thanks for your question.

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.

Regards,

Wan