Dark Vessel Detection by SAR data analysis (Defense Tech hackathon winner)

Arctic Overwatch: Tracking Shadow Vessels in the Arctic with AI

Nearly half of "shadow vessels" in the Arctic deliberately disable or spoof their AIS transponders to avoid detection. Traditional tracking fails when ships simply turn off their signals.

Arctic Overwatch solves this by detecting ships through their wakes instead. The project won 2nd place at the DMZ Toronto Defence Tech Hackathon.

How it works: Every ship leaves a unique wake pattern that can't be faked or disabled. Using Synthetic Aperture Radar (SAR) satellite imagery, these wakes remain visible even in harsh Arctic conditions.

We built a CNN trained on 458 SAR images to recognize and classify wake patterns rather than the vessels themselves. The pipeline: clean satellite data, detect wakes with the CNN, generate unique wake fingerprints, and cross-reference with AIS records.

Why it matters: This proves we don't need cooperative tracking systems. By combining satellite imagery with deep learning, we can monitor vessels that deliberately try to disappear which is critical for the Arctic's vast, remote waters where traditional patrols aren't feasible.