From space, AI can finally see how Kenya's small farms really grow their maize

Many Kenyan farmers grow maize mixed in with other crops in small, patchwork fields — a method called intercropping. It feeds a lot of people, but satellite maps and official statistics usually miss it completely, because the mixed fields are hard to read from above and there is too little labelled field data to train computers on. Now researchers at the Chinese Academy of Sciences and partners have built a mapping method that learns from the rhythm of the growing season itself. Using Sentinel-2 satellite images, it produced Kenya's first 10-metre national maps separating single-crop maize from intercropped maize, for both the 2023 long rains and the 2023–2024 short rains, with accuracy of 88.7% and 85.8%. Because it needs far fewer costly field surveys, the same approach could be adapted for other small-farm regions where the food picture has been blurry.

SDG 2 · Zero hunger — Mechanism: tech innovation