DAHONRX: Plant Leaf Disease Identification And Detection Using Efficientnetb4 Architecture

Abstract

Background: Plant diseases significantly threaten global crop productivity, yet traditional manual inspections remain time-consuming and frequently inaccessible to remote farmers. Objectives: To address this challenge, this study introduces DahonRx, an AI-powered mobile application engineered for the precise, early identification of plant leaf diseases in corn and peanut crops. Method: The system is built on the EfficientNetB4 architecture, specifically selected for its optimal balance of high diagnostic accuracy and the computational efficiency required for mobile deployment. Developed using the TensorFlow and Keras frameworks, the model processes 160 × 160 input images and is optimized into a TensorFlow Lite format for on-device inference. The user-facing mobile interface, developed with Flutter, allows farmers to capture or upload leaf images to receive instant disease predictions, confidence scores, and historical scan logs. Results: Following rigorous training and strategic fine-tuning, the model achieved a final test accuracy of 98.96%. Conclusion: By integrating sophisticated deep learning into an accessible interface, DahonRx demonstrates a scalable solution for modern agriculture. Contribution: This provides farmers with a reliable tool to significantly enhance crop management and productivity in real-world environments.