APPS: Vessel detection and classification in harbors
As part of the international ITEA project Advancing Plug & Play Surveillance (APPS), ViNotion has developed recognition technology for maritime applications.
The APPS project aims at extending existing radar-based Vessel Traffic Management Systems (VTS) with additional sensors, to be used in harbors and along shorelines to support maritime traffic management.
ViNotion has investigated advanced vessel detection and classification algorithms. From all moving objects, vessels have the largest variation in visual appearance, which makes the detection and classification a complex and challenging problem. Using modern machine learning techniques, ViNotion is able to detect vessels with high accuracy. Moreover, each vessel is classified into one of the 25 different vessel types, ranging from small pleasure yachts, up to large container ships.
Features
- Vessel detection and tracking in video sequences
- Detects vessels invisible to radar (small speedboats)
- Real-world (GPS) positioning for integration into a Vessel Traffic management System (VTS)
- Extension to existing radar systems
- Distinguishes between 25 different vessel types
Applications
- Security for oil- and or windmill platforms
- Validate and improve vessel trajectories by radar and AIS, through integration into a Vessel Traffic management System (VTS)
- Virtual fences in the water to count the number of ships in an area, for example in the harbor.
- Shoreline protection using virtual fences
- Automatically measures the speed and dimensions of a vessel for enforcement
- Estimated Time of Arrival for bridges and locks
The result
In February 2017 the detection and classification system has been demonstrated in the port of Rotterdam. This demonstration was part of a joint research project with Dutch partners Microflown, Siqura, Thales, de Technische Universiteit Eindhoven, Havenbedrijf Rotterdam, and international partners from Turkey (Aselsan, SRDC, Nanobiz), Spain (Nunsys, Prodevelop, ITI) en Southern Korea (GMT).