A workshop on using EdgeML ProtoNN to create a gesture recognising device on constrained hardware. Based on the Microsoft Research EdgeML project (https://github.com/microsoft/EdgeML) this trimmed down repository presents a simple workshop on building a basic Gesture Pod using the Adafruit Trinket M0 instead of the more expensive Arduino MKR1000. This basic Gesture Pod indicates gestures recognised by using the RGB LED on the Adafruit Trinket M0, rather than via Bluetooth in the full Gesture Pod implementation.
The workshop documents how to assemble the basic Gesture Pod and how to create a custom gesture model.
Required hardware: Adafruit Trinket M0 MPU-6050 3-axis accelerometer/gyro Mini breadboard Jumper wire
A device can be built for as little as £15 (2019 pricing)
I would like to recognise the work and support of Shishir Patil and Harsha Vardhan Simhadri from the EdgeML team in the creation of this workshop.