Anzeige
Erschienen in:
2018 | OriginalPaper | Buchkapitel
The “Watch” Support System for Ship Navigation
verfasst von : Masaki Kondo, Ruri Shoji, Koichi Miyake, Ting Zhang, Tadasuke Furuya, Kohta Ohshima, Masaaki Inaishi, Masaki Nakagawa
Erschienen in: Human Interface and the Management of Information. Information in Applications and Services
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Floating and non-floating objects such as other ships, buoys and so on must be alarmed before becoming obstacles for ship navigations. In this research, we have aimed to predict obstacles around a ship from maritime navigation images using an image recognition method and display them effectively to its operator. Faster R-CNN was used as detection method. We prepared a dataset composed of three categories for training and testing machine learning. We enumerated parameter values to obtain the best detection rate of obstacles by CNN. Then, we employed the best set of parameters for further experiments. The results are summarized as follows: (1) the detection rate of buoys is about 55 [%]; (2) large ships are sometimes mistaken for small boats. It remains to improve the detection rate and to decrease misclassifications; (3) the detection rate of small boats with distance of about 3 nautical mile(nm) from the ship is 86 [%], the detection rate of buoys with distance of about 2 [nm] from the ship is 100 [%].