Marine Debris Detection from Sentinel-2

Abstract

This paper is devoted to using Machine Learning methods to detect Marine Debris from other objects on the surface of the ocean. Data from the MARIDA dataset gathered over the years by the Sentinel-2 (S2) satellite was used. The aim is to recognize debris from other surface features, such as waves, ships, foam, etc. The paper compares two methods for marine debris detection. The decision tree and random forest methodologies were used.