Ontology-based approach for vessel activity recognition
Maximilian Zocholl, Clément Iphar, Anne-Laure Jousselme, Cyril Ray · OCEANS 2021: San Diego – Porto · 2021
Recognising vessel activities like illegal fishing or human trafficking helps focusing resources and allows targeted inspections on suspicious vessels. This paper proposes an ontology-based approach for vessel activity recognition, aligned with prior works in the area of ontology-based activity recognition as well as with existing ontology design pattern including a discussion of the Common Core ontology. Existing definitions of vessel behaviours and activities are reviewed and summarised in a shared conceptualisation. The semantic description is technology-independent, supports portability and allows making a first step towards a combined data- and knowledge-driven vessel activity recognition, for an increased interpretability of information sources. A generic design pattern is proposed to cover more specific existing and potential future vessel activity definitions in a common way. We argue that the modelling and exploitation of uncertain information is a key enabler for vessel activity recognition, given the large variability of vessel behaviours and inherent uncertainty related to their intent.