Cacao Ripeness Detection System

Clients

Techstack

Purpose

Judging when a cacao pod is ripe is normally an eyeball call that varies from one picker to the next, and mistimed harvests cost growers in yield and quality. The university wanted an objective way to make that call so farmers could time harvests with confidence.

Description

We built a prototype that classifies cacao ripeness from measured pod characteristics using a machine-learning model trained on data the team gathered themselves. We validated the classifier against standard accuracy measures before deploying it to low-cost hardware, where it grades new readings on the spot. Taking a subjective field judgment and turning it into a repeatable, on-device decision is the kind of applied machine learning we take on.