Cambridge AI tool maps smallholder crops in Senegal with 84% accuracy

A new University of Cambridge AI tool, Tessera, has proven more accurate at mapping smallholder crops in Senegal than existing methods, potentially bringing advanced agricultural monitoring to the Global South. The open-source model achieved 84% accuracy in tests, outperforming alternatives by up to 28% while using fewer resources. The study, published on 29 September in Environmental Research: Food Systems, compared Tessera against two standard satellite mapping methods and Google DeepMind's AlphaEarth. Researchers found Tessera performed best when trained on one year's data and applied to another, suggesting it could provide reliable crop statistics between costly ground surveys. The technology could help governments and food security organisations in regions where ground data is difficult to gather, particularly given this year's record El Niño disrupting West African rainfall. Lead author Madeline Lisaius emphasised Tessera's accessibility, calling it "a step towards greater geospatial data democratisation."