<![CDATA[Banana is one of the tropical fruit commodities with high economic value, but it is vulnerable to diseases such as anthracnose (spot), crown rot, and fruit rot, which negatively affect yield and fruit quality. The manual detection method, which is still commonly used, has limitations in terms of accuracy and efficiency. Therefore, this study aims to develop a banana disease detection system based on digital images using the You Only Look Once version 5 (YOLOv5) method. This research applies a quantitative experimental approach with a dataset consisting of 1,005 images that were labeled using the Roboflow platform. The training process was carried out in Google Colaboratory with four epoch configurations, namely 20, 50, 80, and 100. Model performance was evaluated using accuracy, precision, recall, F1-score, and mean Average Precision (mAP), as well as confusion matrix visualization. The best training results at 50 epochs achieved an average mAP-50 of 0.817%. The final results of this study demonstrate that YOLOv5 is effective in automatically and accurately detecting banana diseases. The web-based implementation provides added value in terms of accessibility and ease of use. The study recommends further development with a larger dataset and the utilization of mobile applications to support field implementation in real time.]]>
Copyrights © 2026
