New method of mussel survey by using high-resolution acoustic video camera-ARIS and deep learning

Zhao, Fan; Mizuno, Kazuya; Tabeta, Shigeru; Asayama, Takato; Hayami, Hiroki; Fujimoto, Yasufumi; Shimada, Takehiko

doi:10.1109/oceanschennai45887.2022.9775335

Cited by 4 publications

(1 citation statement)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Figure 8 presents acoustic mosaic maps of the survey routes for the 2019 lake survey together with boxes predicted to contain mussels at multiple scales (Zhao et al, 2022). The results of this survey indicate a density of $0.4 individuals/m 2 .…”

Section: Resultsmentioning

confidence: 99%

Survey of freshwater mussels using high‐resolution acoustic imaging sonar and deep learning‐based object detection in Lake Izunuma, Japan

Zhao,

Mizuno,

Tabeta

et al. 2023

Aquatic Conservation

Self Cite

View full text Add to dashboard Cite

The use of traditional in situ methods for underwater surveys to map freshwater mussel habitats is limited by challenges such as water transparency, depth and high labour demands. In this study, adaptive resolution imaging sonar (ARIS) was applied to monitor mussel distribution and abundance. In contrast to conventional quadrat surveys, this acoustic survey is non‐invasive and enables direct observation of mussels to allow their survival status to be determined in turbid water. ARIS produces high‐quality acoustic images that facilitate the creation of a distribution map for broader‐scale monitoring, especially if paired with deep learning methods such as the YOLOv4 algorithm for automatic mussel detection and classification. ARIS was successfully applied to surveying over 2000 m2 of Lake Izunuma in Miyagi, Japan. In one site, a high mussel abundance of ~0.6 individuals/m2 was detected, while other sites had low densities. The detection model achieved a mean average precision of 0.97. The survey results were used to generate a distribution map of living mussels. This study illustrates the feasibility of using an acoustic video camera with an open‐source deep learning algorithm to monitor mussels and other benthos in turbid water.

show abstract

Section: Resultsmentioning

confidence: 99%