The 4th Workshop on Maritime Computer Vision (MaCVi) brings together computer vision, robotics, and marine science to advance perception for autonomous surface vessels (USVs) and related domains. MaCVi 2026 features technical papers, keynotes, and five benchmark challenges. Emphasis is on accuracy and embedded real-time feasibility. By uniting datasets, tasks, and communities, MaCVi accelerates reliable, deployable maritime perception.
Open waters cover over 70% of the planet and facilitate about 80% of global trade. The growing use of cameras and robotic platforms in this domain generates vast visual datasets, demanding robust methods for detection, classification, and scene understanding.
As most goods move by sea, autonomous vessels offer major economic and environmental potential. Ships and ferries can reduce emissions, while smaller USVs already support environmental monitoring. Reliable perception remains essential: smaller craft often lack automatic identification systems (AIS), radar and sonar are noisy, and thus computer vision—via cameras or sensor fusion—is vital for situational awareness.
Unlike UAVs, USVs also rely on vision for navigation in dynamic coastal or inland waters—still a challenging problem. Underwater vision adds difficulties such as low visibility, scattering, and light absorption, which degrade algorithm performance.
These diverse applications motivate a wide range of computer vision challenges, including but not limited to:
