Description of Invention
The accumulation of drifting debris on beaches has become increasingly problematic, with some debris directly deposited by human activities and other debris carried across oceans. In some cases, debris reaches locations that are difficult to access, leading to delayed discovery and substantial accumulation. To address this issue proactively, it is essential to regularly monitor drifting debris. This poster explores the use deep learning combined with drone-based systems, to detect floating debris. We apply models such as YOLO-NAS and YOLOv10 to identify debris in drone-captured images. A case study on a beach in Oshima Island, Fukuoka Prefecture, Japan, demonstrates the effectiveness of these models. Further studies are needed to validate their accuracy across diverse locations.