<![CDATA[Potatoes are one of the important sources of carbohydrates whose quality greatly affects the food industry. The potato quality inspection process that is still carried out manually is often time-consuming and prone to human error. This research developed a quality detection system for fresh and rotten potatoes using the YOLOv8n version of the You Only Look Once (YOLO) algorithm. The study began with the collection of 1000 potato photos that were split into 85% for training, 10% for validation, and 5% for testing. The dataset was then labeled using the Roboflow platform and was aggregated to bring the total to 2304 photos. The training results showed that the YOLOv8n model achieved 99.9% accuracy, 100% recall, 99.5% mAP50, and 97.9% mAP50-90. The model is implemented in a Flask-based website to enable real-time detection. Although the model produces good performance, there are some errors in recognizing object classes. Overall, this system is capable of effectively detecting the quality of potatoes, reducing waste, and maintaining product quality.]]>
Copyrights © 2024
