A new scientific paper developed within the AI4COPSEC project has been published in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, following its presentation at the XXV ISPRS Congress 2026 “From Imagery to Understanding”, held from 4–11 July 2026 in Toronto, Canada.
The paper, “Multi-Source Remote Sensing for Maritime Security: A Performance Evaluation of SAR and RGB Imagery for Small-Scale Fishing Vessel Detection,” was authored by Stefano Chiappini, Alessandro Galdelli, Andrea Fiorani, Marsia Sanità, Carmen Ferrà, Anna Nora Tassetti and Roberto Pierdicca. The research brings together researchers from AI4COPSEC partner Università Politecnica delle Marche (UNIVPM), CNR-IRBIM and the National Biodiversity Future Center.
The study explores how AI and multi-source satellite imagery can improve the detection of small vessels in coastal waters, where vessels are often not systematically tracked and can be difficult to identify in satellite imagery.
Focusing on the Adriatic coasts of Marche and Puglia, Italy, researchers used Sentinel-2 and high-resolution PlanetScope RGB imagery to build a dedicated dataset and fine-tune deep-learning models. Several YOLOv26 configurations were evaluated, with YOLOv26-M achieving the best balance between detection performance and computational efficiency.
Click here to download the paper in Open Access.
The research contributes to AI4COPSEC’s work on combining Artificial Intelligence, Earth Observation and multi-source data to strengthen maritime surveillance and enhance Copernicus Security Services.
