BACK TO ARCHIVE

MISSION 004

SAR-Based Maritime Surveillance for Dark Vessel Detection and Forecasting

An AI-based maritime surveillance system that uses Sentinel-1 SAR imagery, YOLOv8, and AIS data fusion to detect dark vessels, classify ship types, and forecast vessel trajectories in the Bay of Bengal.

STATUS: RESEARCH / DEVELOPMENT

TECH STACK

  • Python
  • PyTorch
  • YOLOv8
  • TensorFlow
  • OpenCV
  • Rasterio
  • GeoPandas
  • GDAL
  • Shapely
  • Streamlit
  • FastAPI
  • Folium / Leaflet
  • Sentinel-1 SAR
  • AIS Fusion
  • LSTM
  • Kalman Filter

MISSION BRIEF

Maritime surveillance plays a critical role in coastal security, fisheries regulation, illegal vessel detection, and search-and-rescue operations. A major challenge in ocean monitoring is the presence of dark vessels, which intentionally disable their Automatic Identification System (AIS) to avoid detection.

Conventional optical satellite imagery is often ineffective during nighttime or cloudy weather. Sentinel-1 Synthetic Aperture Radar (SAR) imagery provides reliable all-weather, day-and-night ocean monitoring instead.

PIPELINE

  1. 01Sentinel-1 SAR ingestion
  2. 02Preprocessing and tiling
  3. 03YOLOv8 vessel detection
  4. 04Ship classification
  5. 05AIS matching
  6. 06Dark vessel identification
  7. 07Trajectory forecasting
  8. 08Geofence violation checks
  9. 09Dashboard visualisation
  10. 10Alert generation

RESULTS

Region of Interest

Bay of Bengal

Chennai / Tamil Nadu coast

Primary Dataset

xView3 SAR

VV + VH bands

Forecast Horizon

30-60

minutes ahead

VIEW SOURCE