MIX - Hybrid Convolutional Neural Network Framework with Explainable Artificial Intelligence for Fig Leaves Disease Detection

Early detection of fig leaf diseases is crucial for maintaining the health and productivity of fig trees.Traditional diagnostic methods are time-consuming, labor-intensive, and dependent on expert knowledge, highlighting the need for efficient and accurate solutions.This paper proposes a novel hybrid framework, called the MIX model, which combines feature fusion from three pre-trained Convolutional Neural Network (CNN) architectures-MobileNetV2, InceptionV3, and Xception-with machine learning classifiers to classify fig leaves as healthy or infected.Using a dataset of 2,321 high-resolution images captured under diverse environmental conditions, the model employs advanced data augmentation techniques and transfer learning to address challenges of data scarcity and variability.The feature fusion process leverages the strengths of the CNNs, while classifiers such as Support Vector Machine (SVM) enhance prediction accuracy, achieving a maximum accuracy of 93.51%.The incorporation of Grad-CAM provides interpretability by visualizing the key regions influencing the model's decisions.Experimental results demonstrate that the MIX model outperforms individual CNNs, showcasing its robustness and reliability.This framework offers a scalable, transparent, and effective solution for early fig leaf disease detection, contributing significantly to precision agriculture and plant health monitoring.