We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Designing Realtime Object Detection Based Plastic Litter Monitoring System
Summary
Researchers built an automated camera system that uses AI to detect and count plastic waste floating in rivers, offering a faster alternative to the current method in Indonesia, which relies on people manually watching and counting debris. This matters because rivers are a major pathway for plastic pollution into the ocean, and better tracking tools can help pinpoint pollution sources and guide cleanup efforts before the plastic breaks down into microplastics that contaminate water, food, and potentially our bodies. While the AI system slightly undercounted debris compared to human observers, it offers the advantage of continuous, real-time monitoring that isn't practical
Abstract Plastic pollution significantly contributes to global marine contamination. Multiple strategies have been implemented to reduce plastic pollution, particularly in waterways that discharge into the ocean. Estimating the volume of plastic waste transported by rivers is a widely adopted method that informs both prevention and mitigation efforts. In Indonesia, manual visual observation is the predominant estimation technique. Although straightforward, this method is labor-intensive and does not enable continuous monitoring. This study presents an alternative approach utilizing Internet of Things (IoT) technology and object detection algorithms for real-time plastic waste monitoring. A camera sensor captures river flows, and an edge-computing device processes these images to detect and classify plastic and non-plastic debris using the YOLOv11 algorithm. The quantified debris data are transmitted to a cloud server and visualized on an online dashboard. System performance was evaluated by comparing automated detection results with manual counts conducted by ten observers on five-minute river video recordings. Human observers consistently reported higher counts than the automated system, likely due to their ability to replay videos for recounting, a capability not available in the system’s real-time operation. System accuracy may be improved by increasing the diversity of training data for the object-detection model. Therefore, the IoT-based real-time plastic waste monitoring system represents a reliable alternative to manual observation of plastic waste in waterways.