The purpose of this study was to save manpower and reduce costs on water quality measurement in cage culture. An unmanned aerial vehicle system was applied to locate the target net cage and detect the water quality and temperature in the desired cage automatically. This paper presents the use of image recognition and deep learning to find a predefined target location of cage aquaculture. The whole drone control and image recognition process was based on an onboard computer and was successfully realized in an actual environment. When the drone approached the net cage, image recognition was utilized to fix the position of the unmanned aerial vehicle on the net cage and drop a sensor to check the water quality. The proposed system could improve conventional manned measurement methods and reduce the costs of cage culture.
Keywords: Drone; deep learning neural network; net cage aquaculture; object identification.