AI model achieves 98% accuracy on cassava disease diagnosis from photos

Researchers built a hybrid AI system combining two types of neural networks to identify cassava plant diseases from leaf images, reaching 98% accuracy on a dataset of 31,000 photos. This matters because cassava is a critical food crop in Africa and Asia — faster, more reliable disease detection could help farmers catch infections early and prevent crop losses before they spread.

This is the first documented case of AI-based cassava disease diagnosis crossing 98% accuracy in a real agricultural dataset, which is the threshold where automated field diagnosis becomes competitive with expert agronomists and could actually be deployed at scale in regions without reliable extension services.