PlastOPol: A Collaborative Data-driven Solution for Marine Litter Detection and Monitoring

Abstract

Marine plastic pollution as a generally accepted global challenge has comprehensive impacts on our living environment. However, in practice, it is extremely difficult to monitor the severity of the problem based on the shortage of data and relevant tools. Thus, this has attracted academia and industry not only as an urgent environmental challenge but also as an interesting research problem. Many projects have been developed to collect relevant information with the help of beach clean-up volunteers and semi-professional. Although data collected by beach-cleaners cover large geographical areas and provide invaluable help in the effort of mapping marine litter hot-spots, the lack of uniformity in the classification systems, combined with the irregularity of clean-up activities makes it difficult to obtain the full picture. This issue further impacts on the information processing, analysis, and decision-support potential of the available tools. In this paper, we provide a collaborative data-driven solution to monitor marine litter along the coast. It includes a mobile application to collect images, a back-end server to map and train the classification model, and a general database to store and manage related data. The case study has been conducted in Ålesund city in Norway, preliminary results are further analyzed for system validation. This work aims to contribute to the United Nations' sustainable development goals 14 for life under water and 11 for sustainable cities and communities.