Computer Vision Algorithms for USV mounted Real-Time Marine Vessel Detection Systems

This is a feasibility study for designing an automatic marine vessel detection system that can be used by an Unmanned Surface Vehicle. Following a thorough exploration of the challenges related to the video date acquired on the boat by our partner, we will come up with recommendations for the best camera hardware setup and preprocessing techniques that improve video quality. Next, state-of-art machine learning techniques for target detection will be applied to spot the marine vessels. We will design prototypes for both training our machine learning models on the server side and for testing it on the boat processor. The output of this system, along with the collected data from the other sensors of the boat may be interpreted and integrated by our partner to create a robust, real-time, boat-embedded autonomous marine vessel detection and tracking system.

Faculty Supervisor:

Alexandra Branzan Albu

Student:

Alireza Rezvanifar

Partner:

Open Ocean Robotics

Discipline:

Engineering - computer / electrical

Sector:

Manufacturing

University:

University of Victoria