Challenges of Labelling Unknown Seabed Munition Dumpsites from Acoustic and Optical Surveys: A Case Study at Skagerrak

The disposal of unexploded ordnance (UXOs) at sea is a global problem. mapping and remediation historic UXOs can be assisted by autonomous underwater vehicles (AUVs) carrying sensor payloads such as synthetic aperture sonar (SAS) optical cameras. AUVs image large areas the seafloor in high resolution, motivating an automated approach to UXO detection. Modern methods commonly use supervised machine learning which requires labelled examples from learn. This work investigates often-overlooked labelling process resulting dataset using example dumpsite Skagerrak. A counterintuitive finding this that images cannot relied on for ground truth significant number visible SAS are not images, presumed buried. Given lack truth, we ordinal scheme incorporate measure labeller uncertainty. We validate regime quantifying label accuracy compared labels with confidence. Using approach, explore different taxonomies conclude grouping objects into shells, bombs, debris, natural gave best trade-off between discrimination.