Cost‐effective seafloor habitat mapping using a portable speedy sea scanner and deep‐learning‐based segmentation: A sea trial at Pujada Bay, Philippines. Issue 2 (30th October 2021)

• Record Type: Journal Article
• Authors: Terayama, Kei; Mizuno, Katsunori; Tabeta, Shigeru; Sakamoto, Shingo; Sugimoto, Yusuke; Sugimoto, Kenichi; Fukami, Hironobu; Sakagami, Masaaki; Jimenez, Lea A.
• Is Part Of: Methods in ecology and evolution. Volume 13:Issue 2(2022)
• Page Start: 339
• Page End: 345
• Publication Date: 2021-10-30
• Subjects: coral -- deep learning -- marine organism -- optical camera array -- seagrass -- structure from motion; Ecology -- Periodicals; Evolution -- Periodicals; 577
• DOI: 10.1111/2041-210X.13744
• Languages: English
• ISSNs: 2041-210X

Abstract

Abstract: Various sampling and monitoring strategies have been developed to assess marine habitats and life‐forms. However, the cost‐effectiveness of such survey methods (e.g. line intercept transects and autonomous underwater vehicles) is still not high. In this paper, a practical seafloor habitat mapping method combining a cost‐effective survey system (P‐SSS: portable speedy sea scanner) and a deep learning‐based quantification method were proposed. P‐SSS is a highly portable transport system and a towed‐type system with five cameras arrayed on its platform. The sea trial was conducted at Pujada Bay, Philippines, on 7 December 2019. The high‐quality orthophotos of the seafloor with a high resolution of ~3.0 mm/pixel were successfully generated. The attained survey efficiency was 12, 900 m 2 /hr. In addition, in this paper, a segmentation method utilizing the U‐Net architecture to estimate the coverage of corals, seagrass and sea urchins using a large‐scale 2D image is proposed. Overall, this highly portable survey system is expected to become a promising tool for marine environmental surveys, especially in the areas where the rich nature of the oceans is susceptible to environmental changes, such as the remote islands that lack sufficient survey facilities.