<![CDATA[Bird’s eye chili (Capsicum frutescens L.) is one of Indonesia’s key horticultural commodities, yet it is highly susceptible to diseases such as anthracnose caused by Colletotrichum spp. fungi and fruit rot due to fruit fly (Bactrocera sp.) infestations. Visual similarities between early and advanced symptoms often hinder manual identification, which can lead to misdiagnosis and the excessive use of pesticides. This study aims to develop an image classification system based on machine learning to distinguish between three chili conditions: healthy, infected with anthracnose, and rotten. The proposed model combines the MobileNetV2 architecture as a feature extractor with a Support Vector Machine (SVM) using a linear kernel as the classifier. The dataset consists of 1,020 field-captured images evenly distributed into three classes (340 per class) and split into training (70%), validation (15%), and testing (15%) sets. Preprocessing steps include background removal, resizing, normalization, and image augmentation. Features are extracted using MobileNetV2 without its final classification layer and used as input for the SVM model. Evaluation results show a validation accuracy of 98% and perfect testing accuracy of 100%, with precision, recall, and F1-score also reaching 100% across all classes. These findings indicate that the hybrid approach is highly effective in detecting chili diseases automatically from images. Furthermore, the system has the potential to help farmers avoid incorrect pesticide use, reduce production costs, and improve crop quality and yield.]]>
Copyrights © 2025
