A Real Time Monitoring System for Accurate Plant Leaves Disease Detection using Deep Learning

Abstract

Agriculture plays a vital role in ensuring food security, yet plant leaf diseases continue to cause significant crop losses due to delayed and inaccurate detection. Traditional manual inspection methods are time-consuming, subjective, and not suitable for large-scale farming. To address these challenges, this paper presents a real-time plant leaf disease monitoring system that combines Deep Learning and Internet of Things (IoT) technologies for fast and accurate disease detection. The proposed system continuously captures leaf images using IoT-enabled camera modules and processes them through an advanced image preprocessing pipeline to remove noise and enhance relevant features. A Convolutional Neural Network (CNN), optimized using transfer learning techniques, is employed to automatically extract features and classify plant leaves into healthy or diseased categories, including multiple disease types. The model is further optimized for deployment on edge devices such as Raspberry Pi, enabling low-latency and real-time predictions directly in the field without heavy computational requirements. The system provides instant alerts, confidence scores, and disease insights through a user-friendly dashboard, helping farmers take timely preventive actions. Experimental analysis demonstrates high accuracy, robustness under varying environmental conditions, and scalability across different crops. By enabling early detection and reducing unnecessary pesticide usage, the proposed solution contributes to sustainable agriculture, improved crop productivity, and cost-effective farm management.

Code and data availability

The paper trains a CNN on the public New Plant Diseases Dataset (Kaggle) but provides no authors' code, trained model, or data deposit; no availability statement or public repository URL for the paper's own assets appears in any block, and no allowed URL hosts a paper-specific asset.

No evidence-backed public reproduction asset is currently recorded.