Orders are now being taken with in Japan for a new automated fish counting system developed by Yanmar Marine Systems Co., Ltd. (YMS). Designed for tuna farming, this system utilises image recognition technology to count the number of fish.
Keeping track of the number of fish inside each net in order to control production volume and the amount of feed used is essential for fish farming. In addition, a drive in recent years towards more sustainable resource management, fishing vessels are required to accurately report the number of juvenile Pacific bluefin tuna caught in the wild, which end up being used for breeding at fish farms.
Current methods are laborious and time-intensive requiring manual counting of the total number of fish caught, and visually counting underwater images when moving the fish to aquaculture nets.
To tackle this problem and support the aquaculture industry, YMS has developed the Automated Fish Counting System which significantly reduces the time needed to count fish.
This is being developed in collaboration with the aquaculture division of Maruha Nichiro Co., Ltd. In preparation for mass production, Maruha Nichiro arranged several sites for Yanmar to conduct trials under various environmental conditions to evaluate the product’s performance.
Yanmar’s Research & Development Center developed image recognition and processing technology and an integrated system including hardware, such as a specially designed underwater camera and image processing computer, to make possible real-time automated counting. Under optimal conditions, the system has succeeded in automatically counting tuna with an accuracy of more than 98%. In the future, YMS aims to contribute to greater efficiency and development of the fishing and aquaculture industry by reducing feed costs, cutting work time, and facilitating the accurate reporting of catches.
The counting system enables users to analyse and check results during or immediately after video capture and the angle of view of underwater cameras can be adjusted by remote control after installation, reducing the time required to install and adjust the camera.
The angle of view and brightness can be adjusted to reduce image disturbances and reduce impact on automated analysis. Screen display and correction functionality to enable easy manual counting and correction after automated analysis even for images affected by environmental disturbances.