<![CDATA[Accurate identification of rice maturity is an important factor in determining the optimal harvest period and maintaining grain quality. Conventional field observations are often influenced by subjective judg-ment and may produce inconsistent results across different observers. This study proposes an auto-mated approach for rice maturity segmentation by integrating Unmanned Aerial Vehicle (UAV) imagery with the YOLOv8 deep-learning model. A dataset consisting of 682 aerial images was collected from paddy fields and categorized into three classes: unripe, ripe, and unhealthy rice. The images were an-notated using bounding boxes and divided into training, validation, and testing subsets. Model training was performed using YOLOv8n for 100 epochs with a batch size of 16. Performance evaluation em-ployed accuracy, precision, recall, and F1-score metrics derived from the confusion matrix. Experimental results showed that the proposed framework achieved an accuracy of up to 93%, demonstrating its capability to identify rice maturity conditions effectively. The findings suggest that UAV-based moni-toring combined with deep learning can support precision agriculture by providing a faster, more ob-jective, and scalable alternative to manual field assessment.]]>
Copyrights © 2026
