<![CDATA[Rice is one of Indonesia's main food commodities, whose productivity often declines due to leaf disease. Early detection of rice leaf disease is an important aspect of maintaining sustainable food security. This study aims to optimize the accuracy of early identification of rice leaf disease by fine-tuning the YOLOv11 model. The research stages included dataset collection, annotation, data preprocessing, data augmentation, model training, fine-tuning, and model performance evaluation. The results showed an improvement in model performance after fine-tuning, with the overall recall value increasing from 0.760 to 0.788 and mAP from 0.764 to 0.785. The confusion matrix also shows a more stable prediction distribution in the fine-tuned model compared to the initial model. Thus, fine-tuning YOLOv11 has proven to be effective in improving the accuracy of early identification of rice leaf diseases and has the potential to support the application of artificial intelligence in the agricultural sector to strengthen food security in Indonesia.]]>
Copyrights © 2025
