Low Cost UAV Based Detection of Coastal Litter: Insights from Lipasmata , Greece

SCOPE Coastal litter pollution is a major threat to coastal ecosystems and nearby communities. Effective mitigation requires monitoring that is rapid, cost-effective and spatially detailed. Unmanned Aerial Vehicles (UAVs) provide high-resolution observations for coastal environmental monitoring. However, many existing litter-detection approaches rely on expensive multispectral sensors or complex analytical workflows. This study investigates a lower-cost alternative based on RGB UAV imagery and artificial intelligence (AI)-enabled image analysis. METHODOLOGY Data were collected during a field campaign in the Lipasmata recreational area, Athens, Greece, in June 2026, within the framework of the ACCELERATE project. The dataset includes RGB UAV imagery and complementary field documentation. State-of-the-art Deep Learning (DL) object-detection methods are applied to identify and map visible litter features and to assess their spatial distribution along the coastal zone. RESULTS AND DISCUSSION Preliminary processing indicates that high-resolution RGB imagery, coupled with DL-based detection, can support the rapid identification and mapping of visible coastal litter, while reducing the time and effort required by conventional ground surveys. The field campaign also revealed practical challenges related to flight planning, weather and illumination conditions, sensor limitations and image interpretation. These factors highlight the need for robust validation and standardized procedures for repeatable mapping. CONCLUSIONS The proposed workflow can provide a scalable and cost-effective solution for coastal litter detection and support pollution-control, as well as coastal-zone-management decisions. It responds to the need for accessible monitoring tools that do not depend on expensive multispectral sensors, while exploiting AI-based image analysis. Future work will compare the performance of the applied DL methods under different environmental and coastal conditions. KEYWORDS: UAVs; Deep Learning; artificial intelligence; coastal litter; coastal pollution; Athens, Greece. ACKNOWLEDGEMENTSThis study is financially supported by the ACCELERATE MSCA Staff Exchanges programme, funded by the European Union’s Horizon Europe research and innovation programme under Grant Agreement No. 101182930. We gratefully acknowledge GeoSense Pco and Christos Gkotsikas for their support in UAV data acquisition. We also thank Adam Wier and Petros Masouridis from WaltR, Giacomo Carli from The Open University, and Miranda Iossifidis from Newcastle University for their valuable assistance during the field experiment, carried out as part of their secondment at Harokopio University of Athens.