Automated River Floating Debris Detection Using Advanced YOLO Models

River pollution from floating debris, particularly plastics, poses a severe threat to aquatic ecosystems and human health, demanding efficient automated detection for timely intervention. This project develops an advanced object detection system tailored for identifying floating debris in river environments using state-of-the-art YOLO variants, specifically YOLOv11 and YOLOv12. The model is trained to accurately detect four primary classes: 'container', 'plastic bottle', 'plastic cup', and 'trash_plastic', which represent the most common plastic pollutants in waterways. Comprehensive evaluation was conducted using key performance metrics including precision, recall, and mean Average Precision (mAP) to determine the optimal model balancing accuracy and inference speed in challenging real-world river conditions. A user-friendly web application was built using Flask as the backend framework, complemented by modern CSS and JavaScript for enhanced interactivity and aesthetics. The platform features secure user registration and login, enabling authenticated users to upload river images for debris detection and visualization. This integrated solution bridges cutting-edge deep learning with accessible web technology, supporting scalable environmental monitoring and sustainable waste management efforts.

Keywords

object detection, river floating debris, plastic pollution, YOLOv11, YOLOv12, model evaluation, deep learning, Flask web application, user authentication, image classification, computer vision, environmental monitoring, waste management.