The dataset includes approximately 4M composite maritime events (e.g., anchored vessels, loitering, ship-to-ship transfer, etc.), recognised by RTEC on semantically annotated AIS position signals, over a period of six months, from approx. 5K vessels sailing around the port of Brest, France. The original dataset is available here.
CAVIAR is a video surveillance dataset. It consists of 28 video sequences of people activities such as meeting, fighting and leaving unattended packages in a public place. The dataset contains coordinates of tracked persons, as well as, precomputed qualitative spatial relations. Moreover, the dataset is annotated by human experts on two levels. The first level contains simple events that concern activities of individual persons, such as when a person is walking or staying inactive. The second level contains complex events, describing the activities between multiple persons and/or objects, i.e., people meeting and moving together, leaving an object and fighting. The original dataset is available here. The first CAVIAR dataset in an Event Calculus syntax is available here.