Smallholder farmers in developing regions often face rapidly changing soil conditions due to climate instability, but lack access to reliable diagnostic tools. To address this challenge, the team trained a TinyML model on 100,000 soil samples from maize fields in Rwanda to interpret subtle soil variations without relying on internet connectivity, grid power, or costly…

This team applied Tiny Machine Learning, or TinyML, to enhance the SolarSENSE off-grid, solar-powered soil sensor. The system measures temperature, moisture, phosphate and salinity, with an ESP32 microcontroller processing the data locally using TinyML. Results are delivered to a smartphone via offline Wi-Fi, eliminating the need for internet access. Designed for small-scale farms, this solution…