Classifying and Identifying Floating Marine Debris
The Department of Engineering cordially invites you to the following presentation:
Classifying and Identifying Floating Marine Debris
Abstract
The amount of plastic debris and generally different kind of litter that enters the oceans is increasing year by year. Due to the sheer scale of plastic marine debris in the world’s oceans— estimated to exceed 150 million tons—and its geographical distribution, the act of manually identifying and mapping sea debris becomes a herculean task. A common method for estimating the amount of marine debris at sea is to tow a net behind a marine vessel and to subsequently count the collected items. In this presentation, we propose a novel way for identifying different types of floating marine debris at the surface of the water using machine learning techniques and computer vision approaches. The performance of Deep Learning techniques using the Convolutional Neural Networks (CNN) architecture and the Bag of Features (BoF) methods were tested in order to figure out which method exhibits the best classification capabilities. For training the classifiers we have used a big image data set that includes images from two marine species and six types of marine debris. These comprised sea turtles, dolphins, plastic bags, plastic bottles, plastic buckets, ghost fishing nets, plastic buoys (fishing gear) and polystyrene. Our investigation compared the performance of Deep Learning and BoF methods as applied to the classification of marine debris as well as the performance of different algorithms of Computer Vision on detecting marine debris. Results reveal that the CNN method attains an accuracy of 99% on classification tasks compared to the BoF which exhibits an accuracy of 83%.
Speaker’s bio:
Ms. Kyriaki Kylili is currently reading for the PhD in Electrical Engineering, at the University of Nicosia. Her doctoral research is concerned with using image processing and sensors to detect, recognize and map plastic debris at sea. The research project Kyriaki works on, namely, “Measuring, mapping and mitigating the effects of plastic debris in the Mediterranean (PLASTICMED)” is funded by the Universitas Foundation and the Leventis Foundation. Kyriaki holds a Master’s Degree in Physics, from the University of Cyprus, during which she completed a graduate thesis related to the optical properties of organic-hybrid materials which can be used in new hybrid devices, such as white LEDs. Kyriaki also possesses a Bachelor’s Degree in Physics from the University of Cyprus.
The talk will be delivered in English and is open to the public. Please save the date:
Date: Thursday, November 30, 2017
Venue/Time: University of Nicosia, RTB-A14, Time 15:00-16:00
Map Link: http://bit.ly/1Q2RyK4
Read also:
• Raising Public Awareness about Plastic Debris at Sea