Salmon Vision BC — Indigenous-led wild salmon AI, Gitanyow Fisheries Authority, peer-reviewed in Frontiers in Marine Science 2023
Canada (BC, Central Coast)
Salmon Vision BC
Indigenous-led wild salmon AI deployment. Gitanyow Fisheries Authority + Skeena Fisheries Commission + SFU. Peer-reviewed in Frontiers in Marine Science 2023. DOI 10.3389/fmars.2023.1200408.
Lead actors + named partners
- Lead author: Dr. William I. Atlas, Wild Salmon Center (Portland, OR). Native to Heiltsuk Nation territory.
- Simon Fraser University School of Computing Science — Sami Ma, Ryan Shea, Jiangchuan Liu.
- Simon Fraser University Biological Sciences — Daniel Scurfield, Brandon Nam, Jonathan W. Moore.
- Pacific Salmon Foundation — Katrina Connors.
- Douglas College Computing Science Department — Xiaoqiang Ma.
- Gitanyow Fisheries Authority — Mark Cleveland.
- Skeena Fisheries Commission — Janvier Doire.
- Heiltsuk Nation — joined October 2023 (per CBC 28 Oct 2023) on the Koeye River.
What this deployment does
Salmon Vision combines AI with age-old fishing weir technology — First Nations communities have used fish weirs for salmon monitoring for centuries. The AI integration automates species identification, count, and behavioural monitoring.
Substantive performance:
- Mean average precision (mAP) 67.6% overall.
- >90% coho accuracy (per Atlas et al. 2023).
- >80% sockeye accuracy.
- 7 species of anadromous salmonids (Frontiers in Marine Science 2023).
- 500,000+ individual video frames annotated (per AZoRobotics 2024).
- Multi-object tracking + species detection deep learning.
- Real-time monitoring prototype deployed and tested at fish weirs.
Operational locations:
- Bear River fish counting weir, BC Central Coast.
- Kitwanga River fish counting weir (Gitanyow territory).
- Koeye River (Heiltsuk Nation) — added October 2023.
Peer-reviewed status
Atlas WI, Ma S, Chou YC, Connors K, Scurfield D, Nam B, Ma X, Cleveland M, Doire J, Moore JW, Shea R, Liu J (2023). “Wild salmon enumeration and monitoring using deep learning empowered detection and tracking,” Frontiers in Marine Science 10:1200408. Published 20 September 2023. DOI: 10.3389/fmars.2023.1200408.
Public coverage: CBC News (28 Oct 2023); EurekaAlert (28 Sep 2023); AZoRobotics (2024); Innovate BC; multiple primary sources.
Why this deployment matters
Salmon Vision operationalises the integration of Indigenous-led methodology with cutting-edge AI. The fish weir is a centuries-old First Nations technology; the AI integration is the contemporary layer. The deployment is Indigenous-led and academically-anchored — substantively distinct from vendor-captured AI.
William Housty (Heiltsuk Integrated Resource Management Department Associate Director) on the integration (UBC CTLT 19 Nov 2025): “To us it is… we’re utilizing our own Traditional Knowledge to inform AI. AI is turning around and helping us to gather information, and we’re making decisions based on that information… That in itself is ethical, in that we’re not relying on the technology to make a decision for us.”
Funding source
BCSFA (BC Salmon Farmers Association) + Ocean AID + DFO (Department of Fisheries and Oceans). Mixed federal + industry + Indigenous funding. No single federal programme funds Salmon Vision. This is the framework-deployment funding-instrument gap pattern (G-313).
Operational status
Operational at Bear River + Kitwanga River (Gitanyow); expanded to Koeye River (Heiltsuk) October 2023. Pilot 2023 → broader deployment 2024+ ongoing. Goal stated as “real-time count data by 2024” (Atlas per EurekAlert 28 Sep 2023). Forward-looking: stated expansion to a half-dozen new watersheds on B.C.’s North and Central Coast (per Atlas, EurekAlert). G-336.
Framework citation
Salmon Vision does not formally cite OCAP®/CARE/NISR/IEEE 2890-2025 in its published documentation. The deployment integrates Traditional Knowledge with AI but does not formally name a framework as governing document. Same framework-deployment operationalisation pattern as PolArctic — alignment implicit, not explicit.
Why this matters for talks
For archetype 05 (critical-lens Indigenous sovereignty): Salmon Vision is the substantive operational case for Indigenous-led AI in BC. The talk-ready claim: “Salmon Vision is what data sovereignty by design looks like — Indigenous-led methodology (fish weir + Gitanyow knowledge) integrated with cutting-edge AI (computer vision, 500,000+ frames annotated). The deployment is peer-reviewed, operational, and expanding to new rivers.”
For archetype 02 (data-sovereignty Canadian): Salmon Vision is data-sovereignty-by-design — community-stewarded, peer-reviewed, anchored in centuries-old Indigenous methodology. The talk-ready claim: “Salmon Vision’s 90% coho detection accuracy and Frontiers in Marine Science 2023 peer review put Indigenous-led wild salmon AI on par with vendor-deployed aquaculture AI — and the Indigenous-led deployment outperforms on community accountability.”
For archetype 06 (regional-cluster-comparison): Salmon Vision anchors the BC Central Coast Indigenous-led fisheries AI cluster, distinct from the EU cooperative cluster model and the US land-grant-extension cluster model.
Substantive contested claims surfaced
- C-291: “Indigenous-led AI agrifood deployment in Canada does not exist.” Counter: Salmon Vision is peer-reviewed in Frontiers in Marine Science 2023.
Substantive gaps surfaced
- G-336: Salmon Vision’s stated expansion to Koeye and KwaKwa rivers — operational status unclear.
Freshness and source provenance
- last-verified: 2026-07
- sources:
- frontiersin.org/journals/marine-science/articles/10.3389/fmars.2023.1200408/full (Atlas et al. 2023 Frontiers in Marine Science 10:1200408)
- cbc.ca/news/canada/british-columbia/artificial-intelligence-salmon-monitoring-1.7007573 (CBC 28 Oct 2023)
- eurekalert.org/news-releases/1004954 (EurekAlert 28 Sep 2023)
- azorobotics.com/News.aspx?newsID=14403 (500,000+ frames)
- facebook.com/innovatebc/posts/1628409355952460 (Gitanyow + Ocean AID + Salmon Vision)
- thenarwhal.ca/heiltsuk-salmon-ai/ (The Narwhal coverage)
- indigenousinitiatives.ctlt.ubc.ca/2025/11/19/ai-reflections-indigenous-data-sovereignty-and-artificial-intelligence/ (UBC CTLT 19 Nov 2025; William Housty quote)