DONATE

IAHR Document Library


« Back to Library Homepage « Book of Extended Abstracts of the 41st IAHR World Congress, ...

Monitoring Plastic Using Deep Learning Model in a Controlled Laboratory Environment

Author(s): Ana Todorova; Robert Niven; Matthias Kramer

Linked Author(s): Matthias Kramer, Robert K. Niven

Keywords: Plastic transport monitoring Macroplastics Computer vision Surface transport

Abstract: Plastic pollution in aquatic environments has emerged as a critical global concern, with significant threats to ecosystems, wildlife, and human health. Traditional methods for monitoring plastic pollutants, such as manual counting and sampling, are labour-intensive and inefficient for long-term application. Therefore, we propose an automated method that uses computer vision and deep learning approaches to identify, measure, and track floating plastic particles. For our experiment, a GoPro camera was used to capture video footage of 3D-printed plastic particles in a controlled hydraulic environment. Using the YOLOv8 model, we trained a custom object detection algorithm on a dataset of 2,396 images, with data augmentation techniques applied. To enhance analysis, we implemented a line counter for object quantification, instance segmentation to estimate the surface area of pollutants, and trajectory tracking to monitor particle movement and estimate their velocity. These results lay the groundwork for the application of this methodology in real-world scenarios, contributing to more effective plastic waste monitoring strategies.

DOI:

Year: 2025

Copyright © 2025 International Association for Hydro-Environment Engineering and Research. All rights reserved. | Terms and Conditions