North Pacific Anadromous Fish Commission

Technical Report 17

Table of Contents

Development of Age Determination Technique Part 1: Neural-Network Based Prediction of Chum Salmon Age by Scale Images

Authors: 
Ryoma Hoson, Hiroyuki Shioya, Yasuyuki Miyakoshi, Fumi Yamaguchi, and Hirokazu Urabe

Abstract Excerpt:
In recent years, the poor catch of chum salmon has continued in Japan, and the research aimed at monitoring the stock status and elucidating the factors that cause fluctuations in stock has been attracting attention. However, since it takes a lot of time to perform age assessment by hand, it is required to actively introduce information technology such as AI and ICT. Neural networks have been applied to predict or represent fisheries data, and many works in the field of fisheries have been presented (Hyun et al. 2005; Suryanarayana et al. 2008; Zhou 2003). In this study, we constructed an age-assessment system using neural networks to automate salmon age determination using scale images and verified its effectiveness. Salmon age determination was composed as the computational process by using neural networks with respect to the image classification. The age-assessment system will contribute to the analysis of salmon data in Hokkaido.

*This is the first paragraph of an extended abstract. Download the full abstract below.

DOI:
https://doi.org/10.23849/npafctr17/147.149.

Citation

Hoson, R., H. Shioya, Y. Miyakoshi, F. Yamaguchi, and H. Urabe.  2021.  Development of age determination technique part 1: neural-network based prediction of chum salmon age by scale images.  N. Pac. Anadr. Fish Comm. Tech. Rep. 17: 147–149.  https://doi.org/10.23849/npafctr17/147.149.