Thank you! Yeah, I run a WebRTC stack and gstreamer+nnstreamer application to stream video from the Pi. I'm running a TensorFlow Lite model on-device for objec

Thank you! Yeah, I run a WebRTC stack and gstreamer+nnstreamer application to stream video from the Pi. I'm running a TensorFlow Lite model on-device for object detection. Getting that stack to run on a 32-bit SoC without an operating system would be tough. That said, there are some incredibly talented people in the world advancing the field of "TinyML" https://www.tinyml.org/ https://www.tinyml.org/ - it's not like you're suggesting something ridiculous! There are already specialized camera + NPU (Neural Processing Units) boards on the market for specialized edge CV tasks. PrintNanny's workflow/automation requires general computing capability though, which rules out a lot of specialized CV boards.