Maritime monitoring system against illegal fishing

Illegal fishing threatens marine biodiversity, disrupts ecosystems, and creates unfair competition for legal fisheries. To address this, this project proposes a maritime monitoring system that combines satellite imagery with Automatic Identification System (AIS) data, using deep learning to detect and evaluate potentially unauthorized vessels. A YOLOv11-Oriented Bounding Box (OBB) model was trained for both optical and SAR imagery, achieving F1-scores of 94.8% and 99.1% respectively. Detected ships are geolocated using GeoTIFF metadata and matched to AIS broadcasts based on position and time. Matches are evaluated using parameters such as position, time and speed. An interactive web-based interface allows users to explore detections and AIS matches. While the full integrated system was only tested on optical imagery, results demonstrated effective vessel detection and matching. Some errors occurred due to reliance on limited parameters. A user study confirmed the tool’s clarity and usefulness, though no domain experts were involved. Despite current limitations, the system provides a strong foundation for improving maritime surveillance and combating illegal fishing.