British Columbia’s extensive coastline, numerous salmon-bearing rivers, and remote areas create significant challenges for monitoring wild salmon populations. By combining artificial intelligence (AI) tools with traditional fishery weir infrastructure, the Pacific Salmon Foundation and the Wild Salmon Center are supporting First Nations across BC in automating counting of returning adult salmon to facilitate population assessment.   

Advanced technologies like AI, machine learning, and computer vision, are revolutionizing the way our society processes and analyses large quantities of data. These tools, which have been pioneered over the last 40 years by computer science and technology industry leaders now touch almost every part of our everyday lives. But far too often the benefits of these cutting-edge computing tools have not reached rural and remote communities in Canada. Salmon population monitoring is an area where machine learning (where computers learn from data without explicit programming) and computer vision (where computers interpret visual data) has the potential to catalyze transformations that benefit both wild salmon and the people that depend on them. Through a grant provided by the British Columbia Salmon Restoration and Innovation Fund, PSF is working with Simon Fraser University (SFU), the Heiltsuk, Kitasoo Xai’xais, Haida, Gitanyow, Nuxalk, Wuikinuxw, Gitga’at, and Taku River Tlingit First Nations, Skeena Fisheries Commission, Fisheries and Oceans Canada, and the Wild Salmon Center to develop AI-based computer vision models that automate salmon identification and counting from video and sonar monitoring projects. This technology saves vital staff time enabling broader application of video and sonar monitoring at lower cost. In addition, rapid video and sonar processing will provide vital in-season information on salmon returns, enabling precautionary management that reduces fishing impacts when runs are low, and mobilizes fishing opportunities when fish are abundant. The Heiltsuk Nation were able to harvest sockeye for the community for the first time in years due to the estimate of returning sockeye that was generated from the AI video systems on the Tankeeah and Koeye weirs. This victory was featured in a Narwhal article.

In 2024, PSF and our partners made exciting progress in our effort to bring AI tools to remote communities. Through our efforts to train and refine computer-vision models to automate salmon counting, these models now routinely achieve greater than 90 per cent accuracy in species counts. However, for these tools to make the impact we envision, we needed to apply them in field with partners.

Using machine learning systems designed by our partners in the SFU NetMedia Lab, we ran computer vision models at five First Nation-led salmon counting weirs including the Kitkiata weir (Gitga’at), Kwakwa weir (Kitasoo Xai’xais), and the Tankeeah and Koeye weirs (Heiltsuk). These efforts generated thousands of motion detected video clips, and technicians with partner Nations reviewed a portion of these videos to inform final estimates of salmon returns. Video data collected at the weirs were delivered to partners weekly throughout the fishing season, contributing to decision making around Food, Social, and Ceremonial fisheries. In addition, AI tools were used at the Haida Nation’s Yakoun River sonar program to produce an annual estimate of Chinook returns, a task that would otherwise be impossible due to high numbers of pink salmon obscuring the ability of human reviewers to detect larger Chinook.

Over the past year, we also made major investments in developing an easy-to-use web application called Salmon Vision to enable access to AI tools to evaluate wild salmon returns. The web app provides our partners with a platform for workflow management, including the ability to automatically upload video data from field projects, utilize AI-powered analysis of these data, and review and annotate data and make corrections to update initial AI counts. The sonar program will be integrated into the app in 2025.

Finally, we formed the Salmon Vision Collaborative, an organization comprised of PSF, Wild Salmon Center, SFU NetMedia Lab, and Lumax AI. This partnership will serve as a foundation for ongoing research, development, and implementation of AI tools in salmon conservation over the coming years. By leveraging our organization’s distinct skill sets and geographic focus areas, we aim to scale the application of AI around the Pacific Rim in the next decade, delivering on the promise of AI for conservation of wild Pacific salmon for future generations.