<![CDATA[The advancement of artificial intelligence (AI) technology, particularly in the area of computer vision, has encouraged the use of automatic object detection methods for various needs, including the classification of organic and inorganic waste. The problem of waste management in the Solo Technopark area which is still carried out manually causes the waste sorting process to not run optimally. This research focuses on developing and evaluating the performance of several YOLO models for detecting and classifying organic and inorganic waste types in real-time. The research dataset contains 6,758 waste images categorized into 10 object classes, obtained from Roboflow. The preprocessing stages include annotation, auto-orientation, and image resizing to 640×640 pixels. The dataset is then divided into 70% training data, 20% validation, and 10% testing. This study used three YOLO models, namely YOLOv11, YOLOv12, and YOLOv26 with epoch variations of 10, 30, 50, and 100. Model evaluation was carried out using precision, recall, mAP50, mAP50-95, and inference time metrics. The results showed that the best model was obtained on YOLOv26 epoch 100 with a precision value of 0.92, recall of 0.847, mAP50 of 0.892, mAP50-95 of 0.741, and inference time of 3.0 ms. These findings indicate that the YOLOv26 model has good capabilities in detecting and classifying organic and inorganic waste accurately and quickly, so it has the potential to be used as a basis for developing a real-time waste detection system.]]>
Copyrights © 2026
