Vancouver Island University

Computing Science Department Student Project Gallery




Machine Learning of Fish Age

Student: Jeremy Kawahara
Course: CSCI 491 Senior Project, 2011
Supervisor: Huizhu Liu
Title: "Development of Computer-Aided Annuli Detection Software"

Abstract: The concentric growth patterns on a salmon scale can be used to determine the age of the salmon. During the winter months, growth is relatively slow and rings become closely grouped together forming a winterzone delineated by an annulus. By correctly identifying which crowded bands of rings formed in the winter months, we can determine how many winters a salmon has lived and thus determine its age.

The goal of this project is to develop Computer Aided Detection software to assist laboratory technicians in accurately and consistently identify the age of chum salmon scales. In this project we are researching various image processing algorithms including the Hough Transform and the Canny Edge Detection algorithm to extract features from the image. Machine Learning methods such as K-nearest Neighbour and Support Vecor Machines are being used to classify and detect noise within the image. This software utilizes the image processing and Machine Learning libraries found within Open CV (Open Source Computer Vision).


Chum salmon scale, unprocessed



Same image with noise classified by the Support Vecor Machine using Hough Trasform, edges detected using Canny Edge detection, and resized.
Images courtesy of Fisheries and Oceans Sclerochronology Laboratory, Supervisor Shayne MacLellan.



Robot

Student: Mike Mills
Course: CSCI 491, 2009, and as graduate work at University of Alberta
Supervisor: Huizhu Liu
Title: "Robot"





Gara Pruesse's Homepage
Computing Science Homepage
Vancouver Island University