IOW Logo

OTC Stone:
OTC Rostock: Automatic localization and measurement of boulders in acoustic datasets based on neural networks

Duration:
01.10.2021 - 30.09.2024
Project manager:
Dr. Svenja Papenmeier
Funding:
BMBF - Bundesministerium für Bildung und Forschung
Researchfocus:
Focus 4: Coastal seas and society
Partners:
Subsea Europe Services GmbH

The goal of OTC-Stone is to make the mapping of boulders effective and objective in order to provide a reliable and reproducible data basis for diverse economic and ecological questions. In OTC-Stone, an operational software will be developed that automatically locates and measures boulders in hydroacoustic data sets by integrated processing of bathymetric data and acoustic backscatter intensities with means of neural networks. An extensive input dataset is essential for training the algorithms. The more training data sets are available, the more accurate and reliable the result of the automatic analyses. Consequently, in addition to software development, another goal is the spatial expansion of existing local data sets, since only after a comprehensive training phase can an independent problem solution be guaranteed. The utilization of reliable and reproducible statements about boulder occurrences is manifold: Larger stones can pose a hazard to shipping; international standards (IHO S-44 Order 1a and 1b) require reliable detection of obstacles along all main shipping routes. Under EU directives, European states are required to map geogenic reefs, place them under nature conservation and monitor their ecological condition. For the first time, the automated mapping of individual boulders, including size information, makes it possible to calculate the colonization areas of hard substrates and to derive ecological parameters such as biomass from them. Previous values were based purely on estimation. Accurate detection of stones is also required in planning approval procedures for offshore infrastructure.

Publikationen

  • Hinz, M., P. Westfeld, P. Feldens, A. Feldens, S. Themann and S. Papenmeier (2024). AI-based boulder detection in sonar data - bridging the gap from experimentation to application. Int. Hyd. Rev. 30: 78-98, doi: 10.58440/ihr-30-1-a08