Machine learning-based framework for plant disease identification and nutrient deficiency severity assessment using leaf images

Abstract

Proper diagnosis of plant disease and nutrient deficiency is crucial in enhancing crop productivity, reducing yield losses and early agricultural interventions. This paper introduces the machine learning approach for automated identification of plant diseases and their severity by analyzing the images of plant leaves. The proposed framework involves image preprocessing, feature extraction, classification and severity estimation, which would enable the accurate identification of various types of disease and estimation of their severity levels while detecting nutrient deficiencies. An extensive image dataset of healthy, diseased and nutrient deficient leaves was used to train and test the models. As illustrated by the experimental results, the proposed framework outperforms the existing machine learning and deep learning methods for plant disease identification and nutrient deficiency detection with the classification accuracy of 98.76% and 98.14% respectively. In addition, the severity assessment module estimates well to enable accurate pesticide and nutrient application, which minimizes chemical use. The proposed framework provides a scalable, efficient, and precise approach for smart crop health monitoring and precision agriculture applications.

Code and data availability

The paper describes a machine learning framework for plant disease and nutrient deficiency detection, but contains no author-deposited dataset, code, model, or supplement. The dataset is only vaguely described as gathered from 'publicly available agricultural image repositories' with no named repository, URL, or access

No evidence-backed public reproduction asset is currently recorded.