Can't run "modelrunner"

[fzfhw]

Hello, I want to deploy a model to my navqplus board.
To do this i read that I have to run the modelrunner on the board.
When I try to start it there is always the error: "modelrunner: command not found"

Before I tried it I copied the "lib" and "bin" data from the "usr" - folder in the eIQ installation dir to the "/usr/bin" and "/usr/lib" folder on my navqplus.

I read the "i.MX Linux User's Guide" and "eIQ Toolkit User Guide" and didn't find nothing specific for the modelrunner.
Is there any alternative to run my "Deepview RT" or "Tensorflow Lite" model on this companion computer?

[brian14]

Hi @fzfhw, 

Thank you for contacting NXP Support.

You can try directly running your TF Lite model on the target using the i.MX Machine Learning Chapter 2.7 Running image classification example.

i.MX Machine Learning User's Guide (nxp.com)

[fzfhw]

Hi @brian14 ,

Thanks for your help.

I tried to run my tflite model with this Python-Script.

When I'm doing this I get the error "Traceback (most recent call last):
File "label_image.py", line 133, in <module>
print('{:08.6f}: {}'.format(float(results[i]), labels[i]))
TypeError: only size-1 arrays can be converted to Python scalars"

I created the model with the eIQ-Toolkit.
You can see the results array on the screenshot.

[brian14]

Hi @fzfhw, 

First, I would like to ask about your model. Do you train the model with eiq? Do you select the image classification model?

The label_image.py is a script for image classification models.

If it's possible, could you please share your model to reproduce the issue on my side using the i.MX8MP EVK?

Have a great day!