Automated Target Classification for Multi-Frequency Echosounders

The oceans cover the majority of our planet’s surface but much of their depths are still a mystery. Improvements in technology have allowed for the development of instruments on underwater platforms and autonomous gliders that are able to survey the world’s oceans. One instrument, called an AZFP (acoustic zooplankton fish profiler), emits high-frequency sonar pulses and listens for backscatter (reflections) to observe fish, zooplankton, suspended sediments, and other quantities in the water column. Backscatter data are complex and time consuming to process and interpret. This study seeks to use recent improvements in Machine Learning to automate the processing and interpretation of backscatter data to reduce the time and manual effort required. Some studies using Machine Learning have already been carried out, but these focused specifically on certain species of fish and plankton and ignore everything else. However, animals in the ocean are also affected by their environment. TO BE CONT’D

Faculty Supervisor:

Stan Dosso

Student:

Alexander Slonimer

Partner:

ASL Environmental Sciences Inc

Discipline:

Geography / Geology / Earth science

Sector:

University:

University of Victoria

Program: