Vessel Encyclopedia for Sea by Software Analysis (VESSA)

VESSA — Vessel Encyclopedia for Sea by Software Analysis VESSA is a structured dataset designed to support the development of computer vision models for the detection, classification, and kinematic estimation of vessels using monocular imagery. It was developed in the context of the research project “A New Approach for Vessel Detection, Classification, and Estimation Using Monocular Vision”, and addresses specific challenges of maritime surveillance with shore-based and ship-mounted cameras. Contents More than 200,000 labeled images of vessels, collected from open maritime sources 26 vessel categories, including: Passenger Ship, Tug, Tanker, Yacht, Sailboat, Fishing Vessel, and others Bounding box annotations in YOLO format, manually revised Auxiliary metadata, including: Vessel physical dimensions and construction characteristics Copyright information for each image, sourced mainly from the ShipSpotting platform Use Cases VESSA is tailored for research involving: Object detection and classification in maritime environments Monocular camera calibration and focal modeling Geospatial estimation of vessel motion and position Real-time monitoring using video streams