Publication
On the Development of a Candidate Selection System for Automated Plastic Waste Detection Using Airborne Based Remote Sensing
Christoph Tholen; Mattis Wolf
In: Max Bramer; Frederic Stahl (Hrsg.). Artificial Intelligence XL. SGAI International Conference on Artificial Intelligence (AI-2023), Cham, Pages 506-512, ISBN 978-3-031-47994-6, Springer Nature Switzerland, 2023.
Abstract
This paper presents the initial steps of the development of a candidate selection system (CSS) for the PlasticObsthinspace+thinspaceproject. The aim of the PlasticObsthinspace+thinspacesystem is the monitoring of plastic waste in waterways and onshore utilizing airborne based remote sensing and Artificial Intelligence. The intended use of the CSS is to use the output of an upstream AI-system used for waste assessment utilizing low resolution imagery to choose promising regions for further investigation by a downstream AI system depending on images with a higher resolution. The CSS is modelled as a cost constrained travelling salesman (CCTSP). For solving the CCTSP a greedy algorithm based on weighted distances is utilized. In this research two different methods of calculating the weighted distances are used and their impact strategies on the performance of the greedy algorithm is investigated. It is shown by the experiments conducted within this research that the ratio combination of the value of the next node and the distance outperforms the linear combination of value and distance within all the experiments conducted.