<![CDATA[Diseases on tomato leaves are a major agricultural issue in Indonesia that can significantly reduce crop yields. This research aims to develop a tomato leaf disease detection system using YOLOv8, a fast and accurate object detection model. The dataset is classified into several categories, such as Healthy, Bacterial Spot, Early Blight, and Black Spot, with data augmentation and normalization processes applied to enhance performance. Evaluation is conducted using F1 curve, accuracy, and confusion matrix metrics. Results show that the YOLOv8 model achieves an average F1 score of 0.72 at an optimal confidence level of 0.506, with the best performance observed in the Healthy class. However, classes like Bacterial Spot and Black Spot exhibit high error rates due to overlapping features among classes. Confusion matrix analysis indicates the need for improved data distribution and representative features to enhance accuracy. This study highlights the potential of YOLOv8 for real-time tomato leaf disease detection. Optimization through dataset balancing, more diverse augmentation, and model fine-tuning is necessary to improve system sensitivity. The system holds significant potential for implementation in IoT-based devices, supporting the advancement of agriculture in Indonesia.]]>
Copyrights © 2026
