Maritime Anomaly Detection using SAR & Edge AI

Monitoring maritime activity across vast ocean regions is a critical yet challenging task. Traditional surveillance methods struggle with scale, weather conditions, and data transmission limitations. This project addresses the challenge by developing an onboard and edge-compatible anomaly detection framework using Synthetic Aperture Radar (SAR) imagery combined with AIS-style vessel motion modeling. I designed a complete synthetic data generation pipeline that simulates multi-ship trajectories along real-world sea routes, integrates environmental context such as wind, waves, and currents, and injects realistic maritime anomalies including off-route deviations, sudden speed changes, sharp turns, and formation breaks. Using these datasets, I built multi-frame synthetic SAR image sequences and developed a layered anomaly detection pipeline with statistical and machine learning-based evaluation. The project delivers a reproducible research-grade framework for maritime anomaly detection, complete with ROC and Precision–Recall evaluation, visualization tools, and modular scripts. It forms the core of my Honors Thesis research in electrical and computer engineering.