Mobile Potato Leaf Disease Detection Using Ensemble Learning | InformativeBD

Karen W. Cantilang, and Laarni M. Ladiao, from the institute of Philippines . wrote a Research article about, Mobile Potato Leaf Disease Detection Using Ensemble Learning. Entitled, Mobile-based potato leaf disease identifier using ensemble modeling. This research paper published by the International Journal of Biosciences | IJB. an open access scholarly research journal Biosciences. under the affiliation of the International Network For Natural Sciences| INNSpub. an open access multidisciplinary research journal publisher.

Abstract

Potato leaf diseases pose a significant threat to crop productivity, necessitating accurate, accessible, and real-time diagnostic solutions. This study proposes a mobile-based potato leaf disease identification system using ensemble modeling to improve classification accuracy and support early disease detection in agricultural environments. The system classifies seven categories, including six disease types—bacteria, fungi, nematode, pest, Phytophthora, and virus—and one healthy (normal) class. A dataset of 3,000 potato leaf images was utilized following the Knowledge Discovery in Databases (KDD) framework, including data selection, preprocessing, transformation, data mining, and evaluation. Deep feature extraction was performed using the Inception v3 convolutional neural network to generate high-dimensional image embeddings. These features were classified using Support Vector Machines (SVM) and further enhanced through a stacking-based ensemble approach to improve predictive performance. Experimental results show that the proposed model achieved an overall classification accuracy of 88% and a macro-averaged Area Under the Curve (AUC) of 0.92, demonstrating strong discriminative capability across all classes. The ensemble model outperformed individual classifiers, particularly in distinguishing visually similar disease categories. The system is designed for mobile deployment with both online and offline functionality, making it suitable for real-world agricultural applications, especially in resource-limited settings. This study highlights the effectiveness of integrating deep learning-based feature extraction with ensemble learning techniques for robust plant disease detection and scalable precision agriculture solutions.

Read more : Curry Leaf (Murrayakoenigii): A Rich Source of Beneficial Fatty Acids | InformativeBD 

Introduction

Potatoes are one of the most important food crops globally, providing sustenance and economic benefits to millions of people (Dolničar, 2021). However, their productivity is often threatened by various leaf diseases, such as early blight, late blight, and bacterial wilt, which can significantly reduce yield quality and quantity. These diseases spread rapidly if not detected and treated promptly, resulting in substantial losses for farmers (Muzammil Khan et al., 2024). Traditional methods of disease identification rely heavily on visual inspection by farmers or agricultural experts, which can be time-consuming, prone to human error, and inaccessible to many smallholder farmers, particularly in remote areas. Recent research highlights the potential of deep learning and computer vision techniques for automated potato leaf disease detection, addressing limitations of traditional visual inspection.

Recent advancements in artificial intelligence (AI) and machine learning (ML) have revolutionized plant disease detection by enabling automated image-based diagnosis. A mobile-based potato leaf diseases identifier using ensemble modeling offers a promising solution by combining the strengths of multiple machine learning algorithms to analyze leaf images captured through a smartphone camera. Ensemble modeling, which integrates models such as convolutional neural networks (CNNs), random forests, and support vector machines (SVMs), improves prediction accuracy by aggregating predictions from different models. This approach minimizes errors, enhances reliability, and ensures consistent disease identification across various environmental conditions, empowering farmers to take timely and appropriate action (Ahmed and Reddy, 2021).

Despite the potential of AI-based disease detection systems, most existing models are limited by their reliance on a single algorithm, which often results in lower accuracy and susceptibility to variations in environmental factors, such as lighting and leaf orientation. While CNNs excel at feature extraction, their performance may be compromised when dealing with noisy data, making it essential to incorporate additional models to improve overall classification performance. Ensemble modeling addresses this limitation by combining the strengths of diverse algorithms, ensuring a more robust and accurate disease identification system. However, the application of ensemble modeling in mobile-based disease identification for potato plants remains underexplored, highlighting a critical research gap.

Moreover, most current mobile-based disease detection applications focus primarily on highvalue crops, such as tomatoes and apples, with limited attention given to potato plants, despite their economic significance. Existing studies often neglect the unique characteristics and challenges associated with identifying potato leaf diseases, including overlapping symptoms and variations in disease progression.

Furthermore, many available solutions are designed for laboratory or controlled environments, limiting their practicality for realworld deployment in agricultural settings (Jafar et al., 2024). Addressing this gap by developing a mobile-based potato disease identifier that utilizes ensemble modeling can significantly enhance disease management practices for potato farmers, particularly in developing regions.

This study aims to fill this research gap by developing a mobile-based potato leaf diseases identifier that leverages ensemble modeling to improve classification accuracy and provide real-time, actionable insights to farmers. By combining the predictive power of CNNs, random forests, and SVMs, the system will offer a more reliable, cost-effective, and user-friendly solution for detecting and managing potato leaf diseases. The successful implementation of this technology has the potential to improve potato yield, reduce crop losses, and contribute to the overall sustainability of agricultural practices.

The main objective of this study is to develop a mobilebased potato leaf disease identification system using ensemble modeling to enhance classification accuracy and support early disease detection. To achieve this goal, the study is guided by the following specific objectives:

1. To collect and pre-process a dataset of potato leaf images, including six disease categories—bacteria, fungi, nematode, pest, Phytophthora, and virus— along with a healthy (normal) class.

 2. To develop a classification model using deep feature extraction (Inception v3) combined with Support Vector Machines (SVM) and a stacking-based ensemble learning approach.

 3. To evaluate the performance of the proposed model using standard metrics, including accuracy, precision, recall, F1-score, and Area Under the Curve (AUC), based on a labeled image dataset.

The scope of this study to develop a mobile-based potato leaf disease identifier using ensemble modeling to enhance disease detection accuracy. The system will allow farmers and agricultural experts to capture images of potato leaves and identify six pre-defined disease categories: bacteria, fungi, nematode, pest, Phytophthora, and virus. The model will integrate multiple machine learning algorithms for improved classification accuracy and will be evaluated based on metrics such as accuracy and precision. The mobile application will function in both online and offline modes to ensure accessibility in areas with limited internet connectivity. The study is delimited to identifying potato disease visible on the leaves. The model’s accuracy depends on the quality and diversity of the dataset, and underrepresented diseases may be harder to detect. Image quality, environmental factors, and overlapping disease symptoms could also affect classification performance. Despite these limitations, the research aims to provide an accessible and effective tool for early disease detection in potato farming.

Article source : Mobile-based potato leaf disease identifier using ensemble modeling