Angus Galloway

Machine Learning Research Scientist @Engineering Technologies Canada Ltd

Toronto, ON, CA

Independently led two ML-R&D projects for fisheries management: MusselFinder and Echofilter 2.0. MusselFinder is a vision transformer method for invasive mussel density estimation from underwater drone photography. The method is four orders of magnitude faster than status quo, learns from user-friendly point annotations without loss of accuracy, and is adopted by USGS. Using Echofilter, Cornell University reduced the processing time of annual acoustic echosounder surveys from 6-8 months to 2-3 weeks.