Deep learning models and best practices for measuring the effects of seismic oil and gas exploration on commercial fish

Principal Investigator
Dr. Christopher Whidden, Dalhousie University, Faculty of Computer Science

Email: cwhidden@dal.ca
Website: https://www.dal.ca/faculty/computerscience/faculty-staff/chris-whidden.html?cq_ck=1602763734288

NRC Collaborator
Philippe Lamontagne, Data Science and Artificial Intelligence Team Lead, Ocean, Coastal and River Engineering Research Centre

Project description

The objective of this project is to develop new machine learning models, analysis software, and best practices to detect and classify fish by species using a combination of video and acoustic data, collected as part of collaborative research to measure effects of seismic exploration on commercial fishing.  The science tools to conduct this research combines video data (from 100 cameras), acoustic telemetry tracking, and acoustic sonar (ship-board and bottom mounted transducers), as part of before/after/controlled/impact experiments in the Newfoundland offshore.

This project will be completed in partnership with the Department of Fisheries and Oceans (DFO), NRC, and academic and commercial partners. As collaborator on this multi-year project, the NRC will process video data and develop the necessary tools required to analyse those large and novel datasets.  This work is important to support ocean industries while ensuring ocean health, resource sustainability, and economic prosperity.

Requirements

The PhD student must meet the standard Dalhousie University Faculty of Computer Science admission requirements including:

  • A strong Grade Point Average (GPA) in a master’s degree in Computer Science or a related technical discipline
  • Evidence of research experience, such as publications and thesis summaries
  • A PhD, Post Bachelor is reserved for outstanding graduates of a four-year program with demonstrated research experience and/or excellent research potential

In addition, the ideal student will have:

  • Evidence of machine learning skills and experience, such as publications, course projects, or internships
  • Experience with the Python programming language and a machine learning framework such as Tensorflow or Pytorch
  • Basic knowledge of data science skills including reading data, cleaning data, data analysis, and graphing the results of data analysis
  • Excellent writing and presentation skills
  • Knowledge or experience with classifying fish, acoustic analysis, or other related experience is ideal but not required

Position is still open.