The MobileNetV2-Driven XAI Framework for Hydroponic Spinach Disease Diagnosis

Being one of the best-organized, sustainable, and nature-friendly methods in urban agriculture, hydroponics is a state-of-the-art approach. Among hydroponic leafy crops, spinach, a commonly chosen plant due to its short cultivation cycle, high nutritional benefits, and steady consumer demands, is vulnerable to various diseases caused by pests, fungi, and bacteria. This leads to significant losses and quality issues, even when it is grown under controlled conditions. Taking into account these constraints, this study presents an automated, flexible, and reliable disease detection solution, specifically developed for hydroponic spinach farming. The proposed method works with spinach leaf image data and integrates advanced machine learning techniques. MobileNetV2, a conventional deep learning classifier known for effective feature extraction, is the core of the proposed system architecture. To maintain the classifier robustness and prevent overfitting, particularly in the case when data are limited, image augmentation and K-fold cross-validation are incorporated. Key metrics such as accuracy, precision, recall, F1-score, and computational complexity are used to evaluate the performance of the system. The results show that an accuracy of approximately 90% can be achieved using moderate computational resources. Furthermore, Explainable Artificial Intelligence (XAI) is also applied to understand model predictions in a better way and validate disease-specific feature learning. This ensures that the proposed method can identify plant diseases correctly without the need for complex infrastructure.