Underwater Source Detection and Classification for Signal-based Surveillance: Audio Dataset Curation and Cross-Domain Evaluation
Updated
Updated · arxiv.org · Jun 29
Underwater Source Detection and Classification for Signal-based Surveillance: Audio Dataset Curation and Cross-Domain Evaluation
1 articles · Updated · arxiv.org · Jun 29
Summary
Researchers have curated a new underwater audio dataset to improve machine learning for maritime surveillance and source classification.
The dataset features over 1,000 labeled segments across eight acoustic classes and introduces a lightweight CNN benchmark with margin-enhanced loss and feature alignment.
This work aims to address data scarcity and domain shift in underwater acoustics, supporting robust model development and cross-domain evaluation for future research.