ALLAN LLOYD FRANCIS V. JUSI TUPI NATIONAL HIGH SCHOOL
Banana leaf diseases significantly reduce crop yield and are difficult to monitor manually. This study developed Musa-AI, an offline AI-based system integrating drone imagery and a ResNet18 neural network for banana disease detection. Using public datasets and UAV-captured images, the system achieved 97.3% accuracy in classifying Cordana leaf spot, Pestalotiopsis, Sigatoka, and healthy leaves. The system demonstrated robustness under varying conditions and scalability for large plantations. Musa-AI provides farmers with an accessible, practical solution for early disease detection and sustainable precision agriculture without requiring internet connectivity or high-end hardware.
