Summary4 points
- The AI app Nuru diagnoses cassava disease from a phone photo; no internet is required.
- Farmers identify cassava disease correctly on their own only about a fifth to a third of the time.
- A Benin study found actual adoption of the app sat at just fourteen per cent, even where it was available.
- Cassava mosaic disease can cut yields by up to seventy per cent when it spreads unmanaged.
Josephine Arisat farms cassava in Busia County, Kenya. A few years ago, she used a phone app called Nuru to check her crop for disease before harvest. Instead of finding infected plants, the app confirmed her cassava was healthy, so she replanted an entire acre from that harvest without a single diseased stem in it.
That kind of confidence used to be out of reach for most smallholder farmers. Left to identify cassava disease by eye, farmers and even trained extension workers get it right only eighteen to thirty-one per cent of the time, according to researchers who study the crop. Cassava feeds more than half a billion people across Africa every day, so a wrong guess in the field is not a small mistake.
What Nuru Actually Does
Nuru, meaning light in Swahili, was built by a team led by David Hughes at Penn State University, working with the International Institute of Tropical Agriculture and partners including the FAO and CIMMYT. A farmer points a phone camera at a cassava leaf, and the app identifies cassava mosaic disease, cassava brown streak disease, or damage from green mites, all without needing an internet connection.
The app runs on a machine learning model trained on well over one hundred thousand annotated leaf images. Researchers testing it in East Africa found Nuru matched or beat trained agricultural extension workers at spotting disease symptoms correctly, a meaningful result in regions where a single extension officer might be responsible for thousands of farmers spread across a wide area.
Since its first cassava release in 2018, the tool has expanded to diagnose potato diseases and the invasive fall armyworm on maize, crops that between them support millions of additional farming households across East and West Africa.
The Gap Between Built and Used
Building an accurate tool turned out to be the easier half of the problem. A study of cassava farmers in Benin found that even where Nuru was available, actual adoption sat at just fourteen per cent. The farmers most likely to use it were the ones who owned an Android phone, had already been through disease training, and had some formal education, meaning the tool’s benefits still cluster around farmers who already had more resources to begin with.
That pattern echoes what researchers see across African agritech more broadly. A good tool sitting unused because of a missing phone, a missing skill, or a missing training session helps no one, no matter how accurate its diagnosis is.
Why This Still Matters For Farmers
For a smallholder without daily access to an agricultural expert, a free, offline tool that performs close to expert level is still a meaningful shift, even with adoption lagging behind availability. In Cote d’Ivoire, cassava mosaic disease alone can cut yields by up to seventy per cent when it goes unmanaged. Catching that early, even through a basic smartphone camera, is the difference between a harvest and a lost season.
PlantVillage has continued expanding Nuru into Tanzania, Uganda, and parts of West Africa, working alongside national research institutes rather than distributing the app on its own. That partnership model, tech built by researchers and deployed through the same extension systems farmers already trust, looks like a more realistic path to real use than the app succeeding purely on its own.
