The algorithms being used for face detection and recognition are using a custom inference engine/architecture we internally call "Oasis-Lite" which is distinct from popular open-source engines like TensorFlow Lite. When a customer purchases the SLN-VIZNAS-IOT they are given access to the source code for the firmware being run on the chip, but the "oasis-Lite" inference engine is provided as a runtime library that cannot be modified by the customer. That being said, we at NXP are continually upgrading and improving the performance and feature set of our engine so that faces can be detected and recognized even faster, and new features can be added.
