Using artificial intelligence to support marine macrolitter research: A content analysis and an online database. (15th February 2023)

• Record Type: Journal Article
• Authors: Politikos, Dimitris V.; Adamopoulou, Argyro; Petasis, George; Galgani, Francois
• Is Part Of: Ocean & coastal management. Volume 233(2023)
• Publication Date: 2023-02-15
• Subjects: Beached/dune -- Floating -- Seafloor -- Deep learning -- Machine learning; Marine resources -- Management -- Periodicals; Coastal zone management -- Periodicals; Coastal ecology -- Periodicals; Ressources marines -- Périodiques; Littoral -- Aménagement -- Périodiques; Écologie littorale -- Périodiques; Coastal ecology; Coastal zone management; Marine resources -- Management; Periodicals; Electronic journals; 551.46
• DOI: 10.1016/j.ocecoaman.2022.106466
• Languages: English
• ISSNs: 0964-5691

Abstract

Abstract: Marine scientists use a variety of collection and monitoring methods to survey macrolitter in aquatic environments, aiming to assess the level of pollution and design mitigation actions. However, the large volume of collected data often makes the visual recognition and identification of macrolitter items a time-consuming and labor-intensive task, indicating the need for automated and low-cost solutions. In addition, modelling approaches are needed to identify which environmental and anthropogenic factors shape the variability of observed litter concentrations. Artificial intelligence (AI) has emerged over the last years as a promising tool to address these issues. This study provides a literature review of published research that uses AI to process macrolitter datasets derived from imagery and tabular data. The focus is on diverse topics (litter domain, dataset source, sampling system, data type, task to be resolved, region, proposed methodologies, usability) with the aim of identifying the versatile contribution of AI on this theme and providing a reference resource for marine litter scientists. To do so, we release an online database (available here ), in which the user can seek publications based on several categories and tags. Current limitations, challenges and potential future directions are also discussed. Highlights: Use of AI in marine litter research is reviewed. Automatic classification, object detection and segmentation of macrolitter is feasible. Litter assessment with AI can be achieved. Practical implications of AI-marine litter studies are still limited. Engage AI in a more efficient way to support marine litter stakeholders.