Improving situational awareness of autonomous vessels using computer vision

Abstract

The aim of the thesis is to develop a computer vision system suitable for automated vessel maneuvering tests. This involves selecting appropriate hardware and computer vision model, developing software for the computer vision system, conducting system tests, and analysing the results. The analysis compares the accuracy and speed of different computer vision models. Additionally, research is conducted to determine ways to optimize the software in order to efficiently utilize the hardware and thereby make the system run smoother and faster.

Keywords: computer vision, machine vision, embedded system, optimisation, containerisation, virtual environment, neural network, machine learning, hardware acceleration.