An AI-Powered Smartphone-Based Solution for Rice Leaf Disease Detection and a Smart Electronic Reporting System Integrating Deep Convolutional Neural Network and the Internet of Things

Authors:
  • SSShadman Shakib
    Shadman Shakib
  • KZKamaru Zzaman
    Kamaru Zzaman
  • MTM. S. Arefin Taibin
    M. S. Arefin Taibin
  • EHEman Hossen
    Eman Hossen
  • EAEyakub Ali
    Eyakub Ali
  • RKRobin Khisa
    Robin Khisa

Abstract
This paper presents an integrated system for the early detection and management of rice leaf diseases that leverages a Deep Convolutional Neural Network (Deep-CNN) and an Android application supported by a multi-layer IoT architecture. Numerous high-resolution images of four rice leaf diseases (Bacterial Blight, Blast, Brownspot, and Tungro) were used to train the DeepCNN model. The model, implemented using TensorFlow and Keras, achieved a training accuracy of 99.5% and a validation accuracy of 99.75%. Key metrics, such as precision, recall, and F1 score, demonstrated the model's robust classification performance. The Deep-CNN model was optimized for mobile deployment with a user-friendly interface for rice leaf disease detection and complaint management. The multi-layer IoT system ensures efficient data flow and processing by utilizing modern software frameworks and cloud technologies. The test results indicated a high model accuracy of 98.53% with minimal loss, as supported by the ROC curve analysis. The Android application provides a practical tool for farmers to enhance agricultural practices through timely and accurate disease management. This approach highlights the potential of data-driven technologies to promote sustainable crop cultivation.

This publication has 24 references indexed in Scilit: