eIQ Toolkit: can't find available models

[sicScar1]

Hi there, 

I'm trying to become familiar with the eIQ Toolkit, Neutron converter, and all the available tools for AI/ML from NXP. I'm developing on an i. MX95 board, and it features a powerful NPU core for running models. 

Reading the EIQTUG guide, the latest release is 31 Jul 2025. I found (5.8.5 chapter) a list of supported models for the Neutron Software. This lists a lot of models, compatible with the iMX95 chip. 
Where can I find these models? How can I integrate and test them? How can I customize them to fit my use cases?

The actual version of the tool I'm using is v1.16.0
Thanks a lot, 

Simone

[Manuel_Salas]

Hello @sicScar1 

I hope you are doing very well.

Please take a look to the i.MX Machine Learning User's Guide in section 9.1 TensorFlow Lite Demos.

There is mentioned a file called download_models.py what is used to download the models.

Actually, you can find the file here.

$ python3 download_models.py

Best regards,

Salas.

[sicScar1]

Hi @Manuel_Salas , 

Thanks for the file link. 

As far as I can see, this is partially true. Yes, some models are downloaded, but not all of the ones listed: 

Why is that? Am I missing something? 
Regards,

Simone

[Manuel_Salas · Accepted solution]

Hello @sicScar1 

From internal team:

The models should be available here: NXP/eiq-model-zoo: A collection of machine learning models for vision optimized for NXP products. This repo should have the steps to create/download models that eIQ supports.

Best regards,

Salas.

[sicScar1]

Hi @Manuel_Salas , 

Thanks for the feedback.
Yes, I knew Model Zoo, but when I looked into it, it wasn't very clear, and the references were old. 
Coincidentally, yesterday there was a huge commit with A LOT of changes, just when you wrote me here. So I guess we need to give it a try again and check these modifications done in the model zoo. 

For the moment, I'm setting the latest post as a solution.
I will come back if there's something else to check again.

Best regards  

[Manuel_Salas]

Hello @sicScar1 

I asked internally.

I will let you know any update.

Best regards,

Salas.