Shipborne Sea Ice Classification Using Neural Networks

The purpose of this project is to take existing state-of-art machine learning techniques and implement them for ice classification in polar seas. Ice classification plays a critical role in any icebreaker voyage. An ice specialist onboard the icebreaker is required to classify all ice environments encountered. This process is tedious and time consuming. This project aims to automate this process. In using cutting edge neural networks, images taken from aboard icebreakers can be used to classify each pixel in an image giving overall context and information about the environment. These ice classifications would serve to generate important documentation for icebreakers as well as contribute to the formation of ice maps for the Canadian Ice Service.

Faculty Supervisor:

Oscar de Silva;Weimin Huang

Student:

Benjamin Dowden

Partner:

Springboard Atlantic

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

Memorial University of Newfoundland