SPEAKER 0
Our first finalist is Talon Rimmer from the Department of Biology. Talon's presentation is titled Farms Fish and Robots.

SPEAKER 1
Imagine you've achieved the dream. You've quit your job or thesis and started a new life as a full time commercial gardener and your friends have told you they will support you and buy whatever you grow provided you can prove that your garden also serves as habitat for animals that live in the area. The question is, how could you prove that your farm does act as habitat for creatures in the long term? Well, there are companies faced with the same question in British Columbia right now, but rather than growing things on land, they grow their crop in the ocean. Seaweed farming is the practice of growing seaweed, a plant like organism similar to how you might grow a garden on land. But instead rows of these plants are attached to ropes and lowered as plots into the ocean to grow, which can then be pulled up to be harvested later on to answer this question of habitat. We put down underwater video cameras in seaweed farms here in BC and recorded over 9000 hours of video over 16 months to find out, do we see more fish and other sea life in seaweed farms than we do in surrounding areas without seaweed. But as you may expect, looking through 9000 hours of video would be extremely challenging. So this is where my thesis comes in. My work is to reduce the effort involved in processing these videos by training artificial intelligence to essentially become a robot biologist. What we do is take a small portion of all videos collected. We identify every animal in each of those videos and then we use that information to train a custom built computer program to also become really good at identifying these same species. And once we're confident in its ability, we can set it loose on the entire video data set to answer our question. And using this method, we've already had pretty good results. This fish school you see up on the screen was correctly plucked from raw video data and identified to a species pacific herring entirely by a computer. And we've trained the same same program to identify over 20 different types of fish, jellyfish and other marine life found right here in the s of sea. This is really exciting because not only can this help us answer our question of seaweed farm impacts on marine life, but it can create a tool that can support a future with more accessible, robust and accurate underwater biodiversity data, which can help us tackle big questions in our rapidly changing ocean and it can empower industry to better monitor the impacts of their own operations, impacts like and questions like the one we're tackling with this project. What is going on on the farm? Thank you.