How running deep neural networks on the microcontroller NXP i.MX 8M Plus (avec une démonstration en live sur une carte de Phytec)

On the device we have a demo which runs a machine learning algorithm for facial recognition in combination with a deep neural network for facial feature comparison. We will explain how the demo was created and how it is composed, followed by an explanation on how to convert a model to make it ready for inference on the embedded system. Here we will explain more on best practice on training/implementing a deep neural network and what complications can be expected when focusing on inference on embedded hardware.