bioRxiv preprint doi: https://doi.org/10.1101/2021.02.01.429285; this version posted February 2, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Improved accuracy for automated counting of a fish in baited underwater videos for stock assessment Connolly RM1*, Fairclough DV2, Jinks EL1, Ditria EM1, Jackson G2, Lopez-Marcano S1, Olds AD3, Jinks KI1 1 Australian Rivers Institute Laboratory – Coast and Estuaries, School of Environment and Science, Griffith University, Gold Coast, Queensland 4222, Australia 2 Department of Primary Industries and Regional Development, Fisheries Division, Government of Western Australia, PO Box 20, North Beach, Western Australia 6920, Australia 3 School of Science & Engineering, University of the Sunshine Coast, Maroochydore DC, Queensland 4558, Australia * Correspondence: Corresponding Author:
[email protected] Keywords: automated fish identification, automated marine monitoring, computer vision, deep learning, object detection, stock assessment, relative abundance. bioRxiv preprint doi: https://doi.org/10.1101/2021.02.01.429285; this version posted February 2, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Abstract The ongoing need to sustainably manage fishery resources necessitates fishery-independent monitoring of the status of fish stocks. Camera systems, particularly baited remote underwater video stations (BRUVS), are a widely-used and repeatable method for monitoring relative abundance, required for building stock assessment models.