Summary
International Conference on Machine Vision Applications
2023
Session Number:P1
Session:
Number:P1-02
LOTS: Litter On The Sand for litter segmentation
Barra Paola, Citarella Alessia-Auriemma, Orefice Giosué, Santana Modesto-Fernando-Castrillón, Ciaramella Angelo,
pp.-
Publication Date:2023/07/23
Online ISSN:2188-5079
PDF download
Summary:
The marine ecosystem is threatened by human waste released into the sea. One of the most challenging marine litter to identify and remove are the small particles settled on the sand which may be ingested by local fauna or cause damage to the marine ecosystem. Those particles are not easy to identify because they get confused with maritime/natural material, natural elements such as shells, stones or others, which can not be classified as ``litter''. This work aims to create a system that analyzes sand images and recognizes likely small-sized litter on the sand. We also present a dataset of Litter On The Sand (LOTS), with images of clean, dirty and wavy sand from 3 different beaches. The results represent a benchmark for this dataset with known state-of-the-art techniques.