Title: A comparative analysis of machine learning algorithms for plant disease detection using leaf images

Authors: Rizwan Ali; Ihtisham Ali; Maqbool Khan; Arshad Iqbal

Addresses: School of Computing Sciences, Pak-Austria Fachhochschule Institute of Applied Sciences and Technology, Haripur, Pakistan ' School of Computing Sciences, Pak-Austria Fachhochschule Institute of Applied Sciences and Technology, Haripur, Pakistan ' School of Computing Sciences, Pak-Austria Fachhochschule Institute of Applied Sciences and Technology, Haripur, Pakistan ' School of Computing Sciences, Pak-Austria Fachhochschule Institute of Applied Sciences and Technology, Haripur, Pakistan

Abstract: This research aimed to evaluate how effectively machine learning algorithms identify plant diseases, with the goal of providing an automated solution for agriculture. The study focused on determining the most accurate model for detecting diseases from plant leaf images, as separating healthy and infected leaves is essential for reliable diagnosis. The algorithms examined included support vector classifier, decision tree (DT), K-nearest neighbour (KNN), random forest classifier (RFC), and naïve Bayes. Accuracy, precision, recall, and F1 score were used as performance indicators. The 'Plant Pathology 2020' dataset, containing leaves with single, multiple, scab, and rust diseases, was used for experimentation. Results showed that the RFC has highest performance across all metrics, while KNN and DT scored the lowest. The study also analysed the detection of single versus multiple diseases, where support vector classifier and decision tree performed well. Automated disease detection offers strong potential for early diagnosis to improve crop productivity.

Keywords: plant disease detection; machine learning; classification; single disease; multiple disease; image processing.

DOI: 10.1504/IJAITG.2026.153483

International Journal of Agriculture Innovation, Technology and Globalisation, 2026 Vol.5 No.2, pp.121 - 135

Accepted: 22 Feb 2025
Published online: 11 May 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article