Rice Disease Detection and Classification Using Deep Learning with Regularization Technique

Read full text Summarise with VIRAT
Authors: Abdulmalik Abdulsalam, Akintayo Michael Ayoade, Grace Ojochenemi Emmanuelanorue, Umar Muhammad Bello, Paul Joseph Agada, Harisu Aliyu, Abdullahi Lawal Rukuna, Alabi Abdultoyyib Adebayo, Muhammad Sirajo, Muhammad Buhari Suleiman
Cited by: 0
References: 37

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Rice production plays a critical role in global food security, particularly in Asia and Africa, where it serves as a staple food for billions of people. However, rice cultivation is highly susceptible to foliar diseases such as bacterial leaf blight, leaf blast, brown spot, and tungro, which significantly reduce crop yield and grain quality. Traditional disease diagnosis relies on manual visual inspection, which is time-consuming, subjective, and often inaccurate, particularly in resourceconstrained farming environments. To address these limitations, this study proposes RiceNet-D169, a deep learning-based rice leaf disease detection system built on a modified DenseNet-169 convolutional neural network architecture. An experimental research design was adopted to develop and evaluate the proposed model. A publicly available Kaggle dataset consisting of 5,932 rice leaf images across four disease classes was utilized. The images were preprocessed through resizing and normalization, followed by systematic data augmentation techniques, including rotation, flipping, zooming, and brightness adjustment, to enhance dataset diversity and improve model generalization. Transfer learning was employed by fine-tuning a pre-trained DenseNet-169 model, while regularization techniques dropout, L2 weight decay, and batch normalization were incorporated to reduce overfitting. The dataset was split into training, testing and validation, and the model was trained. Experimental results demonstrate that the DenseNet-169 achieved an overall test accuracy of 99.83. The model was compared with state-of-the-art approaches, a fine-tuned CNN and ResNet50 transfer learning model implemented under the same experimental conditions. The results show that DenseNet-169 outperformed the comparison models in terms of accuracy, and stability.

Fields of Study & Subjects

Smart Agriculture and AI Advanced Neural Network Applications Plant Virus Research Studies

Publication details

Journal / source
International Journal of Innovative Science and Research Technology (IJISRT)
Pages
1704–1704
Publication date
21 March 2026
Language
EN