About MaCVi
The Maritime Computer Vision Initiative
An open maritime vision community
MaCVi connects researchers, practitioners, challenge teams, and organizations working on computer vision and robotics in maritime, marine, and underwater environments. The initiative brings together work on autonomous vessels, search and rescue, environmental monitoring, underwater perception, remote sensing, and related fields.
We create shared points of contact across research areas through workshops, benchmark challenges, datasets, publications, and community discussion.
What the community does
Workshops
Conference workshops connect technical papers, invited talks, demonstrations, and discussion.
Challenges and datasets
Community-led benchmarks provide shared tasks, evaluation protocols, and public results.
Exchange and collaboration
The community shares research, events, resources, opportunities, and proposals for new activities.
How MaCVi is organized
MaCVi is a community initiative rather than a fixed organizing team. Responsibilities are attached to concrete activities and may change between workshop editions.
- Workshop organizers
- Plan and deliver a specific workshop edition. Each edition lists its organizers on its own people page.
- Challenge teams
- Own individual tasks, datasets, evaluation procedures, participant support, and reporting.
- Community coordination
- Maintains shared infrastructure such as the website and Discord and supports continuity between activities.
- Contributors
- Can propose workshops, challenges, datasets, events, or other resources relevant to maritime computer vision.
History
- 2023 The first MaCVi workshop took place at WACV and established a shared venue for maritime computer vision challenges and research.
- 2024 MaCVi 2024 expanded the challenge programme and introduced stronger emphasis on deployable and embedded perception.
- 2025 MaCVi 2025 combined benchmark challenges with a broader paper, keynote, and community programme.
- 2026 MaCVi at CVPR 2026 broadened the initiative across maritime, marine, multimodal, and deployable computer vision.
Origins and founding contributors
MaCVi grew from the first Workshop on Maritime Computer Vision at WACV 2023. Benjamin Kiefer, Lojze Žust, Matej Kristan, and Janez Perš are acknowledged as founding contributors to the initiative. Current organizers and contributors vary by activity and workshop edition; the historical organizer list is preserved on the MaCVi 2023 people page.
Researchers from the University of Tübingen and the University of Ljubljana initially connected their datasets, benchmarks, and workshop activities to make maritime research easier to discover and to encourage exchange between previously separated application domains.
Acknowledgements and platform history
SeaDronesSee origins
The SeaDronesSee benchmark originated in the Avalon project, which explored unmanned aerial assistance for humanitarian search and rescue at sea. The lack of suitable public data motivated a benchmark for developing and comparing robust maritime vision systems.
The original SeaDronesSee team included Benjamin Kiefer, Martin Messmer, and Leon Varga.
Original evaluation platform credits
The original SeaDronesSee evaluation website was developed during the COVID-19 pandemic as a semester-long Scrum team project at the University of Tübingen, supervised by Benjamin Kiefer and Martin Messmer.
The student development team was:
- Jannik Beran
- David Carbon
- Fabian Dehn
- Noah Dörre
- Jasper Gerken
- Johannes Koch
- Anna Lefarova
- Nils Model
- Timo Moosmann