<![CDATA[Chili farming productivity is frequently compromised by the delayed diagnosis of plant pathogens. Conventional visual identification methods often lack the necessary precision and temporal efficiency for timely intervention. This study introduces LeafXpert, a mobile-based diagnostic framework leveraging the YOLOv8n deep learning architecture for real-time chili leaf disease identification. Utilizing the PlantVillage dataset, the system is trained to categorize five distinct states: Bacterial Spot, Cercospora, Leaf Blight, Leaf Curl, and Healthy. To ensure deployment feasibility on resource-constrained hardware, the model was converted to TensorFlow Lite (TFLite) for efficient on-device inference. Performance evaluation yielded superior results, achieving a Precision of 97.6%, Recall of 98.0%, and mAP50 of 98.5%. These findings demonstrate that integrating edge computing into Android applications provides a robust, field-operable solution for plant health monitoring, effectively eliminating the latency and connectivity requirements associated with server-based diagnostics.]]>
Copyrights © 2026
