Utilizing Artificial Intelligent (AI) Tools to Discern Plastic Debris at the Marine Environment


Plastic debris in the marine environment is one of the most pressing environmental concerns of our time. Detailed knowledge of the abundance and geographical distribution of plastics across the world’s oceans can help better monitor, control and address the environmental hazards. Utilizing Artificial Intelligence (AI) tools for the detection of plastic objects in the marine environment can help reveal the true scale which is key to suggesting remedial measures.
In this talk, we will present smart ways for identifying plastic debris in the seas and the coastlines. Starting from simple machine learning tools and concluding with sophisticated AI algorithms, we propose a number of intelligent plastic debris classifiers. Such classifiers are able not only to identify plastic items but also to localize and segment the litter items in images and video footage. The AI-based image classifiers can recognize nine categories of objects: (1) plastic bottles, (2) buckets, (3) bags, (4) fishing nets, (5) plastic straws, (6) food wrappings, (7) a fish species, (8) aluminium cans and (9) cigarette butts. When validated on still images displaying plastic debris, YOLOv5 scored an average precision of 92.4%. Applying the tools on video footage, the YOLOv5 and YOLACT++ tools attained very similar average precision at almost 75%. Thanks to the ability of YOLACT++ to apply a mask over each detected plastic item, it was used to develop the intelligent method which is proficient in deducing the density and the dimensions of plastic litter found at shorelines. Utilising images depicting plastic litter from 6 coasts in Cyprus, the intelligent method identified four categories of plastic debris, namely, bottles, bags, straws and food wrappings. The abovementioned result translated into a plastic litter density of 0.035 items/m2 and extrapolated to the entire shorelines of the island, the method estimated about 66,000 plastic articles. Concluding, the intelligent method, with the aid of the OpenCV Contours image processing tool, predicted the length of the plastic debris which spanned between 10 and 30 cm.

Speaker’s bio:

Ms. Kyriaki Kylili is currently a PhD candidate in Electrical Engineering, at the University of Nicosia. Her field of investigation deals with the creation of smart ways, based on deep learning tools, for identifying and localising plastic debris in the marine environment. Kyriaki in the past has examined the presence and impact of micro-plastics in water and has modelled the fate of plastics in the sea and the coast. As a researcher, Kyriaki participated in the PLASTICMED project, funded by the Universitas Foundation, which dealt with the detection and identification of macro- and micro-plastics in the Mediterranean Sea. Her doctoral research is jointly funded by the University of Nicosia and the A. G. Leventis Foundation. Kyriaki holds a Master’s Degree in Physics, from the University of Cyprus, during which she conducted research on the optical properties of organic-hybrid materials which can be used in LEDs. Kyriaki also holds a Bachelor degree in Physics, from the University of Cyprus.

The talk to be delivered in English will held face-to-face and will be lived streamed via WebEx:
www.webex.com; Link: https://bit.ly/3nePcQx; Meeting number: 27307177287; Meeting Pass: 3DBpMERS7t6. For more info please visit the Marine & Carbon Lab: www.carbonlab.eu

Venue/date: Conf. Room 102, RTB, U. of Nicosia; Wed., Nov. 17th, 2021.
Time: 10:00-11:00am.

You may also find interesting:

Leave a Reply

Your email address will not be published. Required fields are marked *