Deep Learning–Driven Image Classification Framework for Accurate Detection of Rice Plant Diseases

Abstract

Rice production is increasingly under threat by a serious fungal disease in the Chidambaram region of Cuddalore district, especially false smut, sheath blight, and brown spot, which are becoming more severe under global climate change. Usually, farmer do their inspections at a later stage, which causes critical damage to the rice crops. This manual inspection is error-prone, time-consuming, and subjective. In these situations, AI-enabled tools and methods are essential for accurate and timely rice disease prediction. This research introduces a novel approach using deep learning–driven image classification framework for accurate detection of rice plant diseases (DLDICF-ADRPD). The DLDICF-ADRPD undergoes three different stages, namely data collection, data preprocessing, feature extraction, detection and classification of diseases. This combination leads to an efficient and robust disease classification system. The series of experiments was conducted to assess the proposed DLDICF-ADRPD performance using large dataset of rice leaf images from different disease types and growth phases, obtained from the publicly accessible Kaggle datasets. When compared to other existing disease prediction models, our DLDICF-ADRPD model performs better. Overall, the suggested DLDICF-ADRPD design greatly increases the reliability and accuracy of disease recognition, supporting global food security and sustainable agriculture.

Plant phenotyping relevance

イネ葉画像から病害状態を推定する深層学習による画像分類手法の開発・性能評価が研究の中心であり、植物病害フェノタイピングに該当する。

Code and data availability

植物フェノタイピング解析を再現する公開資産であることを、入力本文と直接リンクから確認できなかったため保留しました。

No evidence-backed public reproduction asset is currently recorded.