Deteksi Sampah Organik dan Anorganik Menggunakan Model YOLO Berbasis Dataset Roboflow

-Waste separation practices in Indonesia remain far from optimal due to limited supporting infrastructure and low public participation, indicating the need for more automated approaches that reduce reliance on manual sorting. This study investigates the capability of the YOLOv8 object detection algorithm to classify organic and inorganic waste from images. A dataset of 3,169 annotated images sourced from the Roboflow platform was used, and the model was trained through a transfer learning scheme using the YOLOv8n pretrained weights at a resolution of 640×640 pixels for 100 epochs. Model performance was evaluated using Precision, Recall, and mean Average Precision (mAP). The resulting scores — 0.968 for precision, 0.970 for recall, 0.986 for [email protected], and 0.922 for [email protected]:0.95 — demonstrate that YOLOv8 can reliably distinguish waste categories with high accuracy. These findings highlight the potential of YOLOv8-based object detection as a foundational component in future computer vision – driven waste management systems.