The New Frontier in Plant Pathology: Predicting Bergamot Diseases with CNN and Random Forest

The research paper describes an innovative model devised for the predictive analysis of diverse Bergamot diseases. The model employs both the robustness of Convolutional Neural Networks and the efficacy of the Random Forest algorithm. The prediction model demonstrates superior performance by classifying and predicting the incidence of various Bergamot diseases, such as Citrus Canker, Phytophthora Root Rot, Greasy Spot, Bergamot Witches’ Broom, Alternaria Brown Spot, and Tristeza. Extensive evaluations of the developed model revealed elevated Precision, Recall, and F1-Score values for each Bergamot disease classification. The model’s Precision, Recall, and F1-Score averaged 95.02 percent across all classes, demonstrating its exceptional ability to accurately predict each class with few false positives and negatives. In addition, the model demonstrated a high Accuracy ranging from 98% to 99% for individual disease classes, confirming its robustness and dependability for use in agricultural health diagnostics. The results demonstrate the value of combining CNNs and Random Forest for disease prediction, which could revolutionize disease management practices in Bergamot cultivation.