Extracting and Matching Keypoints using Deep Learning to Estimate Salmon Biomass

In land-based aquaculture being able to estimate the mass of a fish is critically important in monitoring their health as they move through their lifecycle as well as knowing when to harvest. Current methods are invasive to the fish in its environment and result in mortality during sampling. We aim to use two cameras placed closely together underwater to take simultaneous pictures of fish and enable the estimation of their biomass. This is a complicated process involving the use of AI and ML to automate the identification of a fish in the cameras’ field of view and then make specific measurements of the fish to determine its biomass. Working with land-based fish farms we will collect data that can be validated and use these examples to train our models.

Faculty Supervisor:

Michael Bauer

Student:

Partner:

ReelData Inc

Discipline:

Computer science

Sector:

Information and cultural industries

University:

The University of Western Ontario

Program: