SalmonVision

The SFU NetMedia team visits the Skeena River site to install our edge-computing systems.

Left: An aerial view of a SalmonVision field-monitoring site on the Skeena River.
Right: A field-deployable edge-computing and communications system developed to support local data processing, remote operation, and resilient data delivery at salmon-monitoring sites.

Intelligent systems for wild salmon monitoring

SalmonVision is a collaborative research initiative developing practical technologies for monitoring wild Pacific salmon. As a research partner, the NetMedia Lab at Simon Fraser University contributes expertise in artificial intelligence, multimedia analysis, distributed sensing, edge computing, and remote communications.

Our research focuses on transforming underwater video, imaging sonar, and other field observations into timely and interpretable ecological information. Working with SalmonVision collaborators, we investigate computational methods that support fish detection, classification, counting, tracking, and information retrieval across diverse monitoring environments.

Research designed for field deployment

Wild salmon monitoring frequently takes place in remote and operationally demanding environments. Monitoring systems must operate under constraints such as limited power, intermittent network connectivity, variable water conditions, reduced visibility, acoustic interference, and large volumes of sensor data.

NetMedia Lab studies how machine-learning models, edge-processing systems, and communication technologies can operate effectively under these conditions. Our research includes:

• computer vision and multimodal learning for fish detection, classification, counting, and tracking;
• efficient processing and analysis of underwater video and imaging-sonar data;
• edge intelligence for resource-constrained monitoring installations;
• resilient data transmission over satellite and terrestrial networks; and
• multimedia retrieval and data-management methods for distributed ecological observations.

This research emphasizes accuracy, computational efficiency, system robustness, and compatibility with established field-monitoring practices. Human expertise remains central to the process, providing validation, ecological interpretation, and quality assurance.

Collaboration across western North America

Active SalmonVision collaboration and monitoring locations across Alaska, British Columbia, Washington, and Oregon. The network includes video-weir, imaging-sonar, and fish-ladder installations.

Research and project resources

Through our collaboration with SalmonVision, NetMedia Lab develops and evaluates methods at the intersection of artificial intelligence, multimedia systems, networking, and ecological monitoring. This work contributes to the broader goal of producing reliable and accessible information for salmon research, assessment, and stewardship.

Visit the SalmonVision project website to learn more about the broader initiative, its collaborators, and its monitoring activities. Additional information about NetMedia Lab’s research contributions, publications, field activities, and team members is available on the SFU SalmonVision research website.