Back to ProjectsData Science / Research · 2024
Maritime Risk Modeling
AIS Trajectory Forecasting & Satellite Ship Segmentation
Complex simulation and deep learning pipeline utilizing historical AIS data and satellite imagery to predict collision risks in congested shipping lanes.
My Role & Contributions
- 01Preprocessed millions of historical AIS transponder records using Pandas and NumPy to reconstruct vessel trajectories.
- 02Trained SARIMA time-series models and U-Net segmentation networks to identify high-density maritime risk zones.
Key Features
- 01SARIMA time-series forecasting for maritime traffic density prediction
- 02U-Net semantic segmentation for ship detection in satellite imagery
- 03Custom spatial risk heatmap visualization dashboard
- 04Automated historical AIS data cleaning and anomaly filtering pipeline
Impact & Highlights
- Improved collision-hotspot identification accuracy in high-traffic strait simulations.