An Intelligent Information System for Plant Disease Detection using Machine Learning and Image Processing

Abstract

Objectives: This article focuses on designing and developing an AI-based plant disease identification system using machine learning and image processing techniques to identify crop diseases and health from images of the leaf. The main objectives include developing a real-time automated investigation tool reachable to farmers, creating datasets of both healthy and disease plant images, designing an easy-to-use interface of the application comparing existing solutions. Methods: This study observed mixed-method approach as research methodology. The method merged with qualitative feedback and quantitative surveys. The data set contains images of healthy and infected plant leaves gathered from multiple crops to train the machine learning models for classification. The experiment used image preprocessing and feature extraction for better accuracy, and performance parameters like response time and usability were assessed. The model’s performance was compared with present gold-standard Pashu Poshan, apps—Plantix and Leaf Doctor—aiming on localization, prediction capability and response time. The usability tests were applied to 80 stakeholders, consisting of farmers of both small and medium scale. Findings: Proposed system got an accuracy rate approximately 90% in plant disease detection, average response time of less than one minute, surpassing other available systems that required time of 20–60 seconds for some elementary or basic recognition. The user assessment provides a usability score of 4.5 out of 5, where nearly 87% of participants valued the software as easy to use. As most rural areas face limited internet connection, the system’s offline feature offered significant advantages. Novelty: Unlike other systems with general disease identification and poor interfaces, the proposed system, integrated with voice commands, disseminates real-time localized treatment guidance projecting prediction based on weather circumstances. It also joins NGOs and local agricultural teams giving extended support such as funding and crop insurance. This wide-ranging integration of intelligent automation, approachability and user adaptation makes Plant Guard a valuable and novel solution for technology enabled agriculture. Keywords: Machine Learning, Plant disease detection, Image processing, Smart agriculture, Crop disease identification Introduction

Plant phenotyping relevance

葉画像から植物病害を検出する画像処理・機械学習システムの開発、データセット作成、性能比較が研究の中心であり、植物の病害状態を直接推定するため対象範囲に含める。

Code and data availability

The article describes a plant disease detection app (Plant Guard) and survey-based usability testing, but contains no public phenotype/trait dataset, image dataset deposit, author analysis code, or trained model with availability language or URL. The dataset is only described as internally gathered leaf images; no data

No evidence-backed public reproduction asset is currently recorded.