A seasoned gardener can diagnose plant stress by visual inspection. Plants tend to get a little bit droopy as they lose water throughout the day and get thirsty.
What if we could train a machine learning model to detect this "droopiness" to trigger a watering event?
What it does
Project "Droop, there it is" aims to reduce water consumption by triggering irrigation events only when drought stress is identified. A battery-powered Arduino Nano 33 BLE sense running an image classifier to identify plant "droopiness" can be deployed in fields without power to signal nearby irrigation equipment.
How we built it
This tinyML use case features Knowledge Distillation to reduce a powerful image classifier for deployment on limited capacity devices.