Automated ship image analysis using Artificial Intelligence

To improve the situational awareness in the maritime domain, this project will enhance anomaly detection and support the authorities in decision-making. For this, we propose a 4-year project to prototype an automated imagery analysis platform that performs behaviour extraction and activity detection of the hybrid/military vessels. To secure that the ‘Innocent passage’ through territorial waters of a military ship is conducted without incidents, it is essential to enhance the accuracy in surveillance. For Belgium, this responsibility lies with the MIK (Maritiem Informatie Kruispunt) in close collaboration with the Navy (see recent article on shadowing of Russian vessels by the Castor). To date, the identification and analysis of potential threats is typically achieved manually by the operators monitoring the imagery of security camera’s (in the visible domain). This way of working applies a number of shortcomings: 1) no persistent surveillance, 2) processing of a limited amount of data and 3) no easy follow-up of events.

The project Ship AI is financed under the "Innovation4Def" call.