Artificial Intelligence Based Fish Counts

Biotactic began developing AI-based fish detection and counting systems in 2013 using a proprietary set of training data to generate deep learning models that can accurately detect untagged fish in a variety of environments and conditions. Field testing first began at fish ladders in Ontario in 2021. To date, our deep learning system has been used to accurately count Arctic Char and Lake trout at fish fences in Nunavut, Kokanee, Rainbow Trout, Large-scale Suckers and Northern Pike at fish fences and ladders in British Columbia, Bull Trout and Cutthroat Trout in a hydroelectric dam penstock in Oregon and Steelhead and Chinook Salmon at multiple fish ladders across Ontario. Our system is the Biotactic Research Artificial Intelligence Network (BRAIN). Not only able to count fish the BRAIN tracks fish movement and differentiates between upstream and downstream and true and false passage. Using optical cameras with built in infrared and artificial intelligence coupled with human-based support, the BRAIN provides clients with accurate assessments of fish passage, species-specific fish counts and migration behavior. Videos can either be run through the BRAIN post project collection or camera feeds can be live-streamed through our established BRAVO network allowing the BRAIN to process continuously and for clients to have near-real time fish counts. To learn more about the BRAIN development and implementation click here.



Image

Rainbow Trout
Downstream movement
True passage

Image

Rainbow Trout
Upstream movement
True passage

Image

Rainbow Trout
Upstream movement
True passage

Image

Lake Trout
Downstream movement
True passage