Optimization for Tensorflow Lite Recipe (i.MX6 - Machine Learning)

[ardtrkc]

Hello !

We are working on a project about Machine Learning with i.MX6 processor on Yocto environment. The recipe that we try to adapt is in the link below.

https://source.codeaurora.org/external/imx/meta-imx-machinelearning/tree/recipes-libraries?h=sumo

As we are working with i.MX6, we just want to be sure about no hardware optimizations were used in this imx-machinelearning recipe for TFlite.

Your guidance and clarifications are important for us.

Thank you in advance.

[Bio_TICFSL · Accepted solution]

Hello ardtrkc,

Tensorflow is not a neural network based application, it is a framework for neural networks. That said, you can definitely run a trained model based on TensorFlow or Caffe by using the DNN module inside the OpenCV library. Out of the box, it will run on the i.MX6Q and any other i.MX containing a CPU supporting Neon extensions. But The eIQ machine learning set is intendend to processor to have NPU like i.MX8 there is no support in i.MX6.

You can run machine learning and deep learning in iMX6 but this will be purely by software.

https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/Setting-up-a-Deep-learning-library-on-i-...

Regards