Fitri Marisa, Muhammad Auzhar Rafli Ramadhani, Anastasia Lidya Maukar, Peti Savitri, Saepudin, Anita Syafariah
Potato leaf diseases such as early blight and late blight pose a significant challenge in cultivation, especially in humid and cool environments. This study proposes a potato leaf disease classification method using the K-Nearest Neighbor (KNN) algorithm, incorporating biological features such as chlorosis and necrosis. In addition to color, texture, and shape features, these features provide physiological indicators that enhance classification accuracy. The dataset comprises 6,000 images, divided into three classes: early blight, late blight, and healthy. The training and testing data are split into ratios of 70:30, 80:20, and 90:10, respectively. Feature extraction was performed using the HSV color space, Grayscale histogram, Hu moments, and biological symptom segmentation. Experimental results showed the highest accuracy of 91% at a 70:30 ratio, with consistent performance across all scenarios. The model demonstrated high reliability in classification, especially for healthy leaves, although there were misclassifications between early and late blight due to visual similarities. KNN proved to be computationally efficient with a prediction time of up to 16 seconds. This study highlights the importance of integrating biological indicators and the potential of lightweight classification models in supporting precision agriculture, particularly in resource-constrained environments. © 2025 IEEE.
Widyagama University of Malang, Department of Informatics Engineering, Malang, Indonesia; President University, Industrial Engineering Study Program, Bekasi, 17550, Indonesia; Vocational Universitas Sangga Buana, Department of Informatics Engineering, Bandung, Indonesia; Universitas Sangga Buana, Bandung, Indonesia; Universitas Sangga Buana, Vocaion, Bandung, Indonesia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock