Check out his post if you haven't yet. This post links to a few additional pieces of documentation and code references.
The code for our project lives in a few places:
This video documents when I first got the Coral Edge facial detection model running with Processing streaming Pi Camera images to Python:
This video shows our custom facial recognition algorithm classifying faces from image crops determined from the result of the aforementioned facial detection model:
As Brent mentioned in his post, there is an interesting custom ML pipeline here which emphasizes data ownership and control of the automated system. We hope to iterate further on physical output from this system to better communicate the critical discussion we set out to achieve with the project.