How to get new ll_aton library

[sandeepchawla74]

I am using STM32N6570-DK board.

I wanted to use resnet50 model for Image Classification 

I generated files using stedgeai using the command : 

stedgeai generate --model resnet50_v2_224_int8.tflite --target stm32n6 --st-neural-art default@user_neuralart_STM32N6570-DK.json --input-data-type uint8

It generated the correct files.

When I tried to compile the generated files , the version of aton library is a new one 1.1.1.14 where the code of ImageClassification contains 1.1.0.31 due to which the code is not getting compiled

From which location can I get the latest files for ll_aton lib

Thanks & Regards,

Sandeep

[Julian E.]

Hello @sandeepchawla74,

The getting started image classification was updated yesterday to support the st edge ai core 2.2

GitHub - STMicroelectronics/STM32N6-GettingStarted-ImageClassification: An AI software application package demonstrating simple implementation of image classification use case on STM32N6 product.​

I think this should solve the issue you are describing.

Else, you can use the previous version of the core, but I do not recommend it.

Have a good day,

Julian

[sandeepchawla74]

Hello Julian , 

I could not find the information , I was looking for . 

Let me rephrase my question .

Is there a separate git where I can find different versions of 

Middlewares/AI_Runtime : Placeholder for AI runtime library

I am looking for version 1.1.1.14.

Regards,

Sandeep

[Julian E. · Accepted solution]

Hello @sandeepchawla74,

Thank you for the clarification.

There is no such git, but you can find the AI runtime corresponding to each st edge ai core version in their local install.

The AI_Runtime version 1.1.1-14 is for the version 2.2 of the st edge ai core.

So you at least need to install the core 2.2 and then in the local folder. 

For example on windows, by defulat it should be here: C:\ST\STEdgeAI\2.2\Middlewares\ST\AI\

You can verify the version in Npu\ll_aton\ll_aton_version.h

Based on the getting started, I think you don't need everything but just the /Inc, /Src and /Npu for most applications.

The documentation can be found here:

https://stedgeai-dc.st.com/assets/embedded-docs/index.html 

Have a good day,

Julian