AI-Driven Identification of Rice Crop Disorders Using Multi-Model Classification Framework

Abstract

Cultivated in many nations across the globe, rice is a staple food of great importance. Rice leaf diseases can severely affect crop cultivation, resulting in low crop yields and financial losses. In an older method of identifying leaf diseases, they are classified based on their color, morphology, texture, and shape. Fully automated instructional systems can quickly identify diseased leaves with minimal human assistance. Most of the earlier research on identifying leaf diseases in rice crops used machine learning and feature extraction methods. Characteristics like its shade, surface, patterns of veins, and lesion extent were retrieved from photos of sick leaves. Stated differently, machine learning identifies the illness by extracting characteristics. Instead, machine learning-based feature vector extraction is not totally superior because it involves retraining and missing one dimension. The proposed hybrid model predicts the diseased leaves of rice crops with 97% accuracy and minimum training and validation losses of 0.80 and 1.25, respectively.

Plant phenotyping relevance

イネ葉の病害状態を画像から自動推定する分類手法が研究の中心であり、植物の病徴・病害状態を直接評価しているため。

Code and data availability

The paper uses a publicly available Kaggle rice leaf disease image dataset (3158 RGB images, five classes), but no dataset URL, identifier, or authors' code/model release is provided in the supplied blocks, and no allowed URL can be matched to an actionable asset.

No evidence-backed public reproduction asset is currently recorded.