Exploiting the Deep Learning Potential for Sea Plastic Detection

Sergio Vitale, Giampaolo Ferraioli, Vito Pascazio

Abstract

Plastic debris are one of the most harmful product for the health of the marine ecosystem. Usually, they enter the oceans as macroplastics through river deltas and tend to aggregate with other materials, floating on the sea surface. With the time passing, the macroplastics tend to degrade in microplastic and enter in the marine life because of ingestion. A fast and precise detection of floating plastics is necessary for monitoring and saving the sea ecosystem. Recent studies have demonstrated how remote sensing (and in particular satellites) can be helpful in such detection. Moreover, in the recent years, deep learning (DL) methods have shown great performance particularly in classification and detection. DL methods can help to overcome some pre-processing step that are time consuming and speed-up the detection. The aim of this paper is to exploit the possibility of constructing a large database of satellite images and correspondent mask of detected plastic. Such database will be freely available in order to promote the research on this topic and on the use of DL.

Download

IMEKO-TC19-MetroSea-2020-07.pdf

DOI

-

IMEKO TC

TC19 - Environmental Measurements

Back to the proceedings