AIを活用した航行船舶の船種識別に関する研究
Abstract
In a traffic census, there are several problems to solve. It is time taking to identify the type of small ships which are not legally obligated to install the Automatic Identification System (AIS). It requires time to check the observed record. The observation accuracy becomes less as it tends to depend on the subjective judgement of the observation personnel, which is outstanding particularly in nighttime. Also, the experienced personnel in observation is short in number, which makes difficult to keep successors of the observation techniques.
In order to solve the problems, the authors review work saving and efficient ship type identification judgement by the ship images and the radar reflection intensity mainly with the observation of small ships. In our previous report, we reviewed the ship type identification using the radar reflection intensity distribution (B scope data) and Artificial Intelligence (AI) to solve the problems. In this study, while keeping the ship type identification by the ship images, we reviewed the ship detection and ship type identification using the object detection technique for the first time, and herein we report the outcome.