Empowering Farmers: AI Solutions for Early Pomegranate Disease Detection

Abstract

Pomegranate farming often suffers due to diseases like Bacterial Blight, Alternaria, and Anthracnose, which cause heavy losses for farmers. Identifying these diseases through manual inspection is not only slow and labor-intensive but also prone to errors. This study introduces an automated system that uses machine learning and image processing to detect and classify diseases in pomegranate fruits and leaves at an early stage. A carefully prepared dataset of both healthy and infected pomegranate images was used. The images were cleaned and segmented using Gaussian filtering and k-means clustering. Texture and color features were extracted using GLCM and HSV methods. We trained and tested three models: Convolutional Neural Network (CNN), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). Among them, CNN delivered the best performance, correctly identifying Bacterial Blight with 98.2% accuracy, while SVM and KNN reached 95.1% and 92.3%, respectively. The model's reliability was confirmed using 5-fold cross-validation. This approach shows how machine learning can effectively support farmers by providing faster and more accurate disease detection, helping reduce crop damage. Future work aims to turn this model into a mobile app for real-time use in the field.