• 

  landing page

• 

  globe detection

• 

  ship details

Arctic Overwatch

Team Members: Omid Latifi, Pratik Das, Miran Qarachatani, Adam Wang

🚢 Problem Statement

• 44% of shadow vessels change their key identifying information to evade detection.
  
• Traditional vessel tracking using AIS (Automatic Identification System) fails when ships deliberately turn off or spoof transponders.
  
• This creates a major gap in maritime surveillance, especially in sanction evasion and illegal activity in the Arctic.
  

💡 Solution – Wake Fingerprinting with AI

We propose using Synthetic Aperture Radar (SAR) imagery to fingerprint ships by their wake.

• Every vessel produces a unique wake signature, like a fingerprint.
  
• By training an AI model to detect and classify wake patterns, we can identify ships even without AIS data.
  

Core Components:

1. Custom Detection Model

   • Trained on 458 SAR images of ships.
     
   • Focused on analyzing wake patterns instead of just vessel hulls.
     

2. Wake Fingerprinting

   • Extracts the unique wake pattern of each ship.
     
   • Compares and validates against AIS when available.
     

3. Validation Pipeline

   • Input: Raw SAR GeoTiff images.
     
   • Step 1: Noise reduction.
     
   • Step 2: Custom AI-based wake detection.
     
   • Step 3: Generate unique wake fingerprint.
     
   • Step 4: Validate against AIS records.
     

📊 Inspiration

• Inspired by research on wake detection from the University of Bristol.
  
• Extended into a real-time maritime anomaly detection system for Arctic waters.
  

✅ Key Features

• Detect ships even if AIS is spoofed or disabled.
  
• Scalable for wide-area Arctic monitoring.
  
• Built for clarity, speed, and trust in maritime intelligence.
  

Built With

• next
• python