Underwater Detection: A Brief Survey and a New Multitask Dataset

Article Underwater Detection: A Brief Survey and a New Multitask Dataset Yu Wei 1,2, Yi Wang 1,*, Baofeng Zhu 1, Chi Lin 1, Dan Wu 1, Xinwei Xue 1, and Ruili Wang 3,4 1 School of Software Technology, Dalian University of Technology, Dalian 116620, China 2 Harbin Boiler Co., Ltd, Harbin, 150000, China 3 School of Mathematical and Computational Sciences, Massey University, Auckland 0632, New Zealand 4 School of Computer Science, University of Nottingham Ningbo China, Ningbo 315100, China * Correspondence: [email protected] Received: 27 June 2023 Accepted: 25 April 2024 Published: 25 December 2024 Abstract: Underwater detection poses significant challenges due to the unique characteristics of the underwater environment, such as light attenuation, scattering, water turbidity, and the presence of small or camouflaged objects. To gain a clearer understanding of these challenges, we first review two common detection tasks: object detection (OD) and salient object detection (SOD). Next, we examine the difficulties of adapting existing OD and SOD techniques to underwater settings. Additionally, we introduce a new Underwater Object Multitask (UOMT) dataset, complete with benchmarks. This survey, along with the proposed dataset, aims to provide valuable resources to researchers and practitioners to develop more effective techniques to address the challenges of underwater detection. The UOMT dataset and benchmarks are available at https://github.com/yiwangtz/UOMT.