Tokyo University and Kubota develop drone method to predict potato yield before harvest

Key takeaway

• Researchers at the University of Tokyo and Kubota Corporation have developed a drone-based system that predicts underground potato yield before harvest by combining drone imagery, machine learning, and a mathematical growth model.

• Field trials in 2023 and 2024 achieved high accuracy (0.8+ correlation for biomass, 0.7+ for yield), offering farmers a non-destructive alternative to traditional sampling methods and supporting better harvest timing decisions.

3 Key Points

FAQ

How does the potato yield prediction method work?

Fields are photographed periodically with drones equipped with RGB and multispectral cameras. Image features such as plant cover ratio, canopy height, color indices, and vegetation indices are extracted. A machine-learning model, trained on the relationship between these features and measured underground biomass, then estimates tuber biomass for unharvested plots. The team applies time-series data to a Gompertz growth curve—an S-shaped mathematical model of biological growth—to predict yield at harvest.

What were the accuracy results from field trials?

The team conducted trials in 2023 and 2024 at the University of Tokyo Field Science Center in Nishi-Tokyo City across multiple treatment plots with varying planting density and seed tuber conditions. They achieved a correlation coefficient of 0.8 or higher for tuber biomass estimation and 0.7 or higher for yield prediction using the growth curve.