Video Surveillance and Augmented Reality in Maritime Safety

Recently, augmented reality and machine learning have become integral parts of many developed systems. In the maritime domain, it is particularly interesting to develop a concept that combines augmented reality with the visualization of collision risks, using machine learning for motion prediction as its foundation. Hence, this research aims to propose a system that visualizes the risk in an augmented reality application. The paper presents a distance estimation method that mainly uses a single stationary camera placed at the harbor entrance. The machine learning component involves training the YOLO algorithm on the Split Port Ship Classification Dataset. This distance estimation is an input for the speed estimation algorithm. Speed is a key parameter for the prediction of collision risk. Preliminary experiments were conducted to provide proof of concept for further research, and the description of a case study is included in this paper.