Banana disease diagnosis and productivity advisory application (banana assist)

Abstract

Banana farming is a key economic activity in Uganda, but it faces severe threats from diseases that devastate crops. Smallholder farmers, who rely on bananas for food and income, often lack the resources for early detection or effective management, identifying banana varieties especially at the rooting stage making management uncertain. This leaves them vulnerable to market and climate fluctuations. Current solutions for disease detection and advisory services are largely inaccessible or insufficient. This project develops an AI-driven mobile application to support banana farmers by detecting some of the common banana diseases such as Panama Disease (Fusarium Wilt), Black and Yellow Sigatoka, Bract Mosaic Virus, Moko Disease, and pest infestations, identifying five banana varieties (Gonja, Kibuzi, Mbwazirume, Mpologoma, Musaka), providing banana planting management and offering expert farming advice. Using surveys, interviews, and high-quality image datasets, AI models were trained to recognize disease symptoms and classify banana varieties. An integrated chatbot provided personalized guidance for disease management and best practices. The study demonstrated the potential of AI in agriculture, with the disease detection model effectively identifying symptoms and the advisory chatbot receiving positive feedback from farmers. However, challenges such as limited smartphone and internet access as well as digital literacy hinder broader adoption. This project highlights the transformative potential of AI-powered tools to enhance productivity, resilience, and food security among smallholder farmers, offering scalable solutions tailored to Uganda’s agricultural challenges.