The Marine Debris Forward-Looking Sonar Datasets
This dataset addresses a data scarcity problem for researchers in underwater robotics and AI, though it is incremental as it primarily offers new data rather than novel methods.
The paper tackles the lack of public sonar datasets for underwater robotics by presenting the Marine Debris Forward-Looking Sonar datasets, which include three settings and support multiple computer vision tasks, with initial results provided.
Sonar sensing is fundamental for underwater robotics, but limited by capabilities of AI systems, which need large training datasets. Public data in sonar modalities is lacking. This paper presents the Marine Debris Forward-Looking Sonar datasets, with three different settings (watertank, turntable, flooded quarry) increasing dataset diversity and multiple computer vision tasks: object classification, object detection, semantic segmentation, patch matching, and unsupervised learning. We provide full dataset description, basic analysis and initial results for some tasks. We expect the research community will benefit from this dataset, which is publicly available at https://doi.org/10.5281/zenodo.15101686