Drone Surveillance Integrated with Machine Learning for Enhanced Port Management

We propose deploying a fleet of five autonomous drones for surveillance at ports. These drones stream live video to a central server equipped with three transfer learning based machine learning models. The first model tracks and categorizes vessels, cross-referencing their Hull Identification Numbers (HIN) with the database. Identified vessels are marked in green; unknown ones trigger red alerts and are mapped on a website. Violators are blacklisted upon confirmation. The second model identifies land encroachments by comparing drone-captured images with reference ones, detecting deviations using the Structural Similarity Index Measure (SSIM). The third model detects vessel encroachments by monitoring HINs and alerting authorities if a vessel remains stationary for too long. Alerts for both land and vessel encroachments are sent to port authorities via a dedicated website and a mobile app. A real-time dashboard provides harbor vessel counts, encroachment alerts, recent violations, and weather updates for drone operations, offering comprehensive oversight of port activities. This solution integrates autonomous drones, machine learning, and a user-friendly dashboard to enhance port management.