Deep Learning based Analysis of Underwater Images of Marine Wildlife

Trail cameras and similar technology are increasingly used to monitor wildlife in a passive, non-invasive way. With the improvement of camera and data storage technology, researchers increasingly face challenges with the time-consuming and inefficient nature of manually sorting photos. Artificial intelligence and image recognition technology have huge potential for analysing large image-based datasets and greatly reducing the labour- and time-costs of initial image processing. In this proposed project, the intern will be assisting Assiniboine Park Zoo by developing tools that will aid in classifying underwater photos of beluga whales. The intern will be developing machine learning algorithms to detect and identify wildlife within underwater photos (beluga and jellyfish). They will also be applying their graduate research to improve how these tools perform when the quality of photos differ across periods of time and environmental conditions (e.g., clear, blue water vs. murky conditions).

Faculty Supervisor:

Ahmed Ashraf

Student:

Partner:

Assiniboine Park Zoo

Discipline:

Engineering

Sector:

Other services (except public administration)

University:

University of Manitoba

Program: