Internet of Things (IoT) devices are becoming more advanced through additional sensors, reduced size and increased computational power. In particular, this increase in computational power allows one to run previously-trained machine learning algorithms natively on an IoT device.

This presents an exciting opportunity: IoT devices often feature a variety of onboard sensors which can be used as inputs into a machine learning algorithm.

This talk will use the presenter's boxing training as a practical example of applying sensor data to a machine learning algorithm. In particular, this talk will demonstrate using motion sensor data obtained on an Arduino Nano 33 BLE Sense configured with TensorFlow Lite. This talk will discuss the entire analytical process from problem and data analysis through to algorithm training and deployment. It will also discuss boxing concepts and how these concepts are modelled in an IoT context.

The links to code are provided below:
- https://github.com/ajosephau/boxing_tracker_nano_ble_sense
- https://github.com/ajosephau/boxing_tracker_wio_terminal

The slide deck is available here:
- https://www.dropbox.com/s/7448hf8q9umnnse/Wearable%20Tech%20boxing%20technique.pdf?dl=0