~/inference_engine_samples_build/intel64/Release/classification_sample_async -i $CLT/pics/dog.bmp -m $CLWS/cl_ext_cosh/model.ckpt.xml -d CPU
When I run the above command, I got the following error:
Supported primitive descriptors list is empty for node: ModCosh/cosh/Cosh
Segmentation fault (core dumped)
Can anybody tell me what's wrong?
Thanks for reaching out.
First of all, I recommend you to use the latest version of the OpenVINO™ toolkit, please check The OpenVINO workshop content here.
If not works, would it be possible to share your model (IR files) to test from our end?
Let me know if you would like to share the files privately and I can send you a private message.
Intel Customer Support Technician
A Contingent Worker at Intel
Thank you for your response.
I understand your position to update the OpenVino version. Are you tried posting your question on the slack community to see if anyone else is using Windows?
Thanks Randall for your reply. I navigated to openvino directory of Udacity workspace and found that
the version is 2019.3.376. So, the version is not an issue. Have you tested my IR files? What's the result?
I will try on my local workspace with the links you provided, as I need to continue working after the course
The command given in my initial inquiry is not correct. Please change it with the following command:
~/inference_engine_samples_build/intel64/Release/classification_sample_async -i $CLT/pics/dog.bmp -m $CLWS/cl_ext_cosh/model.ckpt.xml -d CPU -l $CLWS/cl_cosh/user_ie_extensions/cpu/build/libcosh_cpu_extension.so
After testing my IR file, please let me know the result.
Thank you for your response.
Yes, the version is not an issue, I think you need to modify some files for good functionality, you can check this guide for executing a Custom Layer Implementation for Linux.
We tested your files from my end and all works fine:
Top 2 results:
[ INFO ] Execution successful"
The the link you provided is the same github repo, which our customer layers exercise adapted from.
However, in the repo there is no procedure for building samples in the inference_engine. In our exercise,
the code is: cd /opt/intel/openvino/deployment_tools/inference_engine/samples/./build_samples.sh.
I run the following code without sample building step, and it didn't work.
~/inference_engine_samples_build/intel64/Release/classification_sample_async -i $CLT/pics/dog.bmp
-m $CLWS/cl_ext_cosh/model.ckpt.xml -d CPU -l $CLWS/cl_cosh/user_ie_extensions/cpu/build/libcosh_cpu_extension.so
The openvino version in our workspace is 2019.r3 while the tutorial is based on version 2019.r2. This might somewhat
affect in implementing the tutorial. However, there are many people who got the success result (got the probability).
I don't know why some of us (4 people) didn't get the expected result, even though we took the exactly same
steps. A guy in New York went through every step with me (in South Korea) on Zoom (video conference), but I am
still getting the error. I am successful until generating IR files (You got the expected result with my IR files) and
compiling extension library. There might be some technical issues in sample building procedure. For your reference,
I attached the excel sheet showing the whole procedure I have taken.
At the beginning of the exercise, the instructor said:
"Custom layers is an advanced topic, and one that is not expected to be used often (if at all) in most use cases
of the OpenVINO toolkit. This exercise is meant to introduce you to the concept, but you won't need to use
it again in the rest of this course."
So, I decided not to care this issue any more. However, I've learnt a lot in digging the problem.
Anyway, thanks for your help and looking forward to interacting with you in the future.
Thank you for your reply, I really appreciate the information and procedure given about your issue.
Sure, if you have more questions, we will be glad to assist you.