Abstract
With an ever-increasing number of vessels at sea, the modelling, analysis and visualisation of maritime traffic are of paramount importance to support the monitoring tasks of maritime stakeholders. Sensors have been developed in this respect to track vessels and capture the maritime traffic at the global scale. The Automatic Identification System (AIS) is transmitting maritime positional and nominative information at highest frequency rate, making it a valuable source for maritime traffic modelling. From an original AIS dataset covering the area of Brest, France, we extracted a set of 17 maritime routes, connecting ports in this area. Two different representations for the routes are provided: (1) clusters of AIS contacts, and (2) route prototypes, representing the nominal trajectory of the vessels following the route. Additionally, a set of tracklets (built by five consecutive AIS contacts from the same vessel trajectory) has been extracted from the set of routes and the original dataset, and labelled either with the route name to which they belong or as off-route tracklets. This dataset provides thus some ground truth on the routes followed by vessels and is aimed at testing and validating vessel-to-route or track-to-route association algorithms.