Cassava Disease Detection and Classification – A Comparative study of Fuzzy Logic and Traditional Machine Learning

Aim/Background: Plant diseases are a significant threat to global food security, affecting crop yields and economic stability. Early detection and intervention are crucial to mitigating these losses. Currently, AI-driven plant disease detection largely depends on Convolutional Neural Networks (CNNs). However, the necessity for large datasets and substantial computational resources limits their application in resource-constrained environments. This research investigates the potential of fuzzy logic as an alternative approach for detecting and classifying cassava diseases, aiming to address the inherent uncertainty and imprecision in plant disease expression. Methods: We developed a fuzzy image processing framework that integrates fuzzy logic-based algorithms to handle variable disease manifestations. By conducting comparative analyses with conventional machine learning techniques, we evaluated the efficacy of the fuzzy logic and hybrid models in terms of accuracy, precision, and computational efficiency. Results: The hybrid system achieved an overall accuracy of approximately 92%, noticeably outperforming the non-fuzzy CNN model. The fuzzy logic system demonstrated higher precision and recall in detecting diseases with subtle or variable symptoms, showcasing its ability to effectively manage uncertainty. Overall, the fuzzy logic-integrated system excelled in both accuracy and robustness compared to the non-fuzzy CNN. Conclusion: This study led to the development of a prototype application that provides farmers with real-time disease detection capabilities. Our work validates the application of fuzzy logic in agriculture, offering a practical tool for improving cassava disease management.