Classification of Herring, Salmon, and Bubbles in Multifrequency Echograms Using U-Net Neural Networks

被引:5
作者
Slonimer, Alex L. [1 ]
Dosso, Stan E. [1 ]
Albu, Alexandra Branzan [2 ]
Cote, Melissa [2 ]
Marques, Tunai Porto [2 ]
Rezvanifar, Alireza [2 ]
Ersahin, Kaan [3 ]
Mudge, Todd
Gauthier, Stephane [4 ]
机构
[1] Univ Victoria, Sch Earth & Ocean Sci, Victoria, BC V8P 5C2, Canada
[2] Univ Victoria, Elect & Comp Engn, Victoria, BC V8P 5C2, Canada
[3] ASL Environm Sci, Victoria, BC V8M 1Z5, Canada
[4] Fisheries & Oceans Canada, Sydney, NS B1P 6J9, Canada
关键词
Aquaculture; deep learning; oceans; semantic segmentation; underwater acoustics; ACOUSTIC CLASSIFICATION; FISH-SCHOOLS; BACKSCATTERING; IDENTIFICATION; SYSTEM; SOUND;
D O I
10.1109/JOE.2023.3272393
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Echosounders are used by fisheries and ocean observatories, but significant manual effort is required to classify species of interest within multifrequency echograms. This article investigates the use of modified U-Net convolutional neural networks for the pixel-level classification of biological and physical data in echogram imageswith accurate classification of herring and salmon schools, bubbles, and the sea surface. Data were collected on the coast of British Columbia, Canada, over two years using an Acoustic Zooplankton and Fish Profiler at four frequencies (67, 125, 200, 455 kHz). In addition, simulated data (water depth and solar elevation angle) provide spatial and temporal context to improve the quality of predictions. Redundancy is built into the model by using a tiling strategy during training and classification. During training, using a limited set of annotated data, translational augmentation encodes theU-Netswith robust features that enable applications for alternate deployment configurations (lower sampling rates or alternate water depths). To ensure broad applicability, these networks were trained to classify echograms with noise left intact. The best-performing model classifies herring, salmon, and bubble classes with F1 scores of 93.0%, 87.3%, and 86.5%, respectively. The results are accurate even when multiple classes are in close proximity, thus, retaining biological data that would otherwise be discarded due to surface bubble noise.
引用
收藏
页码:1236 / 1254
页数:19
相关论文
共 54 条
[1]   Classifying multi-frequency fisheries acoustic data using a robust probabilistic classification technique [J].
Anderson, C. I. H. ;
Horne, J. K. ;
Boyle, J. .
JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA, 2007, 121 (06) :EL230-EL237
[2]  
Bateriwala M., 2019, UNET SEGMENTATION PY
[3]   COMPARISONS OF HERRING BEHAVIOR IN THE LIGHT AND DARK - CHANGES IN ACTIVITY AND RESPONSES TO SOUND [J].
BLAXTER, JHS ;
BATTY, RS .
JOURNAL OF THE MARINE BIOLOGICAL ASSOCIATION OF THE UNITED KINGDOM, 1987, 67 (04) :849-859
[4]   Support vector machine under uncertainty: An application for hydroacoustic classification of fish-schools in Chile [J].
Bosch, Paul ;
Lopez, Julio ;
Ramirez, Hector ;
Robotham, Hugo .
EXPERT SYSTEMS WITH APPLICATIONS, 2013, 40 (10) :4029-4034
[5]   Acoustic classification in multifrequency echosounder data using deep convolutional neural networks [J].
Brautaset, Olav ;
Waldeland, Anders Ueland ;
Johnsen, Espen ;
Malde, Ketil ;
Eikvil, Line ;
Salberg, Arnt-Borre ;
Handegard, Nils Olav .
ICES JOURNAL OF MARINE SCIENCE, 2020, 77 (04) :1391-1400
[6]   Use of a shoal analysis and patch estimation system (SHAPES) to characterise sardine schools [J].
Coetzee, J .
AQUATIC LIVING RESOURCES, 2000, 13 (01) :1-10
[7]  
Cornwall C., 2010, ESRL GLOBAL MONITORI
[8]   A post-processing technique to estimate the signal-to-noise ratio and remove echosounder background noise [J].
De Robertis, Alex ;
Higginbottom, Ian .
ICES JOURNAL OF MARINE SCIENCE, 2007, 64 (06) :1282-1291
[9]   Development and application of an empirical multifrequency method for backscatter classification [J].
De Robertis, Alex ;
McKelvey, Denise R. ;
Ressler, Patrick H. .
CANADIAN JOURNAL OF FISHERIES AND AQUATIC SCIENCES, 2010, 67 (09) :1459-1474
[10]   Classifying fish schools and estimating their species proportions in fishery-acoustic surveys [J].
Fablet, Ronan ;
Lefort, Riwal ;
Karoui, Imen ;
Berger, Laurent ;
Masse, Jacques ;
Scalabrin, Carla ;
Boucher, Jean-Marc .
ICES JOURNAL OF MARINE SCIENCE, 2009, 66 (06) :1136-1142