0
Article ? AI-assigned paper type based on the abstract. Classification may not be perfect — flag errors using the feedback button. Tier 2 ? Original research — experimental, observational, or case-control study. Direct primary evidence. Sign in to save

Detection and Classification of Floating Waste on Water Surfaces Using YOLO-Based Algorithms

Proceedings of Engineering and Technology Innovation 2026
Boonchoo Jitnupong, Jirawan Charoensuk, Supaporn Bundasak, Supakrit Somritjinda, Nonpawit Silabumrungrad, Jaruwan Suraseing

Summary

Researchers trained AI models to spot floating trash—like plastic bottles, bags, and cans—in water using photos, with their best system (YOLOv11) correctly identifying waste about 83% of the time. This kind of technology could help clean-up crews find and remove plastic pollution before it breaks down into microplastics, which can end up in the fish and water we consume and have been linked to potential health concerns.

Floating waste on water surfaces is one of the causes in environmental pollution. Foam, plastics, and glass bottles are not readily biodegradable and adversely affect aquatic life. This research focuses on detecting, classifying, and counting non-biodegradable and hard-to-decompose waste floating on water surfaces using object detection techniques. Five classes of waste, cans, foam, plastic bags, plastic bottles, and miscellaneous items, are evaluated in this study. Detection and classification are performed by using six models-YOLOv5, YOLOv7, YOLOv8, YOLOv9, YOLOv10, and YOLOv11-which can identify overlapping objects. This study utilizes two datasets with varying resolutions, as well as two model sizes and two batch sizes. In the experimental evaluation, the YOLOv11 model outperforms the other models with a precision of 83% and mAP50 of 78%. Plastic bags and plastic bottles are classified more accurately by YOLOv11 than by YOLOv5, with improvements of 22% and 15%, respectively.

Share this paper