Deep Learning-Based Marine Buoy Detection System

In recent years, the number of fishermen aged 65 or older has been decreasing in Japan. Despite the mechanization of operations on fishing boats, many tasks performed on unstable seas place a heavy burden on elderly workers. In particular, the installation and collection of marine buoys used to collect information and serve as landmarks and floats on the sea have not yet been fully automated and require manual operation in many situations. The problems associated with automation can be solved using artificial intelligence technology by detecting appropriate locations for installing marine buoys and using robotics to realize installation and collection technology. In this study, we propose a method to obtain and identify the basic location of marine buoys from camera images and investigate a marine buoy detection system using a deep learning algorithm. We compared the accuracy of our model to those of models trained using different algorithms and examined the learning model.