Underwater object detection and datasets: a survey

• DOI: 10.1007/s44295-024-00023-6
• Publication Date: 2024-03-04T08:27:08Z
• Authors: Muwei Jian; Nan Yang; Chen Tao; Huixiang Zhi; Hanjiang Luo
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Abstract The rapidly growing exploitation and utilization of marine resources by humans has sparked considerable interest in underwater object detection tasks. Targets captured environments differ significantly from those general images owing to various factors, such as water turbidity, complex background conditions, lighting variations. These adverse factors pose a host challenges, high intensity noise, texture distortion, uneven illumination, low contrast, limited visibility images. To address the specific difficulties encountered environments, numerous methods have been developed recent years response these challenges. Furthermore, there significant effort constructing diverse comprehensive datasets facilitate development evaluation methods. This paper outlines 14 traditional used based on three aspects that rely handmade features. Thirty-four more advanced technologies deep learning were presented eight aspects. Moreover, this conducts study seven representative missions. Subsequently, challenges current tasks analyzed five directions. Based findings, potential research directions are expected promote further progress field beyond.