- Partnering with First Nations, a new interdisciplinary study proposes harnessing artificial intelligence and computer-based detection to count and produce real-time data about salmon numbers.
- Monitoring their population when they return to the rivers and creeks is crucial to keep tabs on the health of the population and sustainably manage the stock, but the current manual process is laborious, time-consuming and often error-prone.
- Fisheries experts say the use of real-time population data can help them make timely informed decisions about salmon management, prevent overfishing of stocks, and give a chance for the dwindling salmon to bounce back to healthy levels.
- First Nations say the automated monitoring tool also helps them assert their land rights and steward fisheries resources in their territories.
Between spring and fall each year in coastal British Columbia, when salmon migrate upstream, the region’s First Nations manually count the number of fish passing through to get a sense of how healthy the population is. But it’s work that takes place in remote and hard-to-access streams of the province, making it laborious, time-consuming, and often error-prone.
So for a recent study, marine scientists, computer scientists and conservation practitioners partnered with Indigenous-led fisheries organizations to build and deploy an automated system to monitor and count salmon.
This first-of-its-kind tool harnesses the power of artificial intelligence to “learn” how to differentiate objects using computer vision algorithms. It can recognize and count 12 species of fish found in the Pacific Northwest, including the five species of wild Pacific salmon, by merely scanning video clips. The study was published in the journal Frontiers in Marine Science.
“This is the first time that anyone has automated counting of salmon from a video,” said Will Atlas, a salmon watershed scientist at the Oregon-based Wild Salmon Center. “We’ve come sort of the closest to having a tool that’s ready to be rolled out into actual management applications.”
In coastal British Columbia, Pacific salmon holds a unique place as a culturally revered fish for Indigenous peoples, and is a prized delicacy for seafood aficionados. The many coastal First Nations had sustainably managed salmon numbers for thousands of years, until logging and overfishing destroyed the delicate balance in the last century. As a result, the number of salmon returning to the many creeks and rivers where they spawn has fluctuated dramatically, casting doubts about their future.
“A central part of managing and conserving salmon is monitoring the number of adult salmon that return to the river to spawn,” Atlas told Mongabay.
Doing so manually, however, just isn’t feasible. “It’s challenging work because we’re at the whim of Mother Nature and the environment,” said fisheries biologist Mark Cleveland from the Indigenous-led Gitanyow Fisheries Authority in Kitwanga, British Columbia.
To train their AI-based tool, researchers used more than half a million video clips recorded by the Gitanyow Fisheries Authority and the Skeena Fisheries Commission. In recent years, these two Indigenous-led fisheries management organizations have begun using high-definition underwater cameras to monitor salmon migration in the Kitwanga and Bear rivers. But they still depend on humans to review the video and count the salmon.
In its initial stages of development, the AI-based tool needed humans to “teach” it to identify salmon — a task field technicians helped with by annotating salmon in the video clips. Over time, the tool learned to recognize the fish so well that it got it right seven times out of 10. Its accuracy surpassed 90% for sockeye (Oncorhynchus nerka) and coho salmon (Oncorhynchus kisutch), two of the important species in the North Pacific.
However, its accuracy was low in identifying pink (Oncorhynchus gorbuscha) and chinook salmon (Oncorhynchus tshawytscha) because the individuals of these species differ in their looks. During the spawning season, the male pink salmon develops a large hump and hooked jaw, and the chinook salmon change colors.
The researchers say they hope that training the algorithm with more data sets, collected from different rivers with more salmon species, can improve its accuracy.
While Indigenous traditional knowledge was not used in developing the AI tool itself, it formed the basis for building Indigenous weirs — fence-like structures built across rivers with a small passage for the fish to pass through — where the cameras were placed. Without the weirs, counting multiple fish captured in a video frame would have been far more challenging. Traditional knowledge also played a role in determining the monitoring season for different salmon, like correcting partners to monitor returning sockeye as early as April, and placing cameras in streams where elders knew salmon migrated.
“It’s a really good proof-of-concept feasibility study,” said AI scientist Justin Kay, co-founder of Ai.Fish, who was not involved in the study. “What they have done well is bringing together all of the stakeholders in developing and deploying [this] technology. I think it’s really impressive.”
Real-time data for better decisions
The newly developed tool automates the existing salmon count carried out manually by First Nations, which can be a bottleneck for making management decisions regarding salmon, such as when to close commercial fisheries or limit the catch in rivers where salmon numbers are low.
Without automation, it takes months to manually review the videos and compile the counts, and by the time the numbers arrive, they’re too outdated to have much practical use.
“At the moment, we get the results postseason — after we recover the hard drive from the site,” said Janvier Doire, fisheries biologist at the Skeena Fisheries Commission.
“Depending on how many data files and different projects are going on, sometimes it could be weeks or months before we have the information we need to make those important decisions,” said Cleveland from the Gitanyow Fisheries Authority.
Both said they hope AI can hasten the process and provide real-time data needed to make decisions on the fly. “Once we can get the number of fish that are coming back to rivers quicker using AI, on a day-to-day basis, First Nations that are harvesting salmon will be able to manage their harvest accordingly,” Doire said.
Computer scientist Robert Moorhead from Mississippi State University, who was not involved in the study, deployed a similar AI-based system to monitor snapper and mackerel in the southeastern United States. “This type of technology is going to be very useful for real-time fish monitoring,” he said. “I think they’re doing something very useful and in the right direction.”
Can AI strengthen Indigenous stewardship?
The Heiltsuk Nation, on the central coast of BC, about 500 kilometers (310 miles) northwest of Vancouver, is larger than the state of Connecticut. Only a handful of staff monitor the number of returning adults and out-migrating juvenile salmon each year in the streams and rivers of the nation’s vast territory.
“Monitoring is important because the salmon are such a huge part of the ecosystem and our own lives — not only for food sustenance but culturally,” said William Housty from the Heiltsuk Integrated Resource Management Department, the nation’s stewardship arm. It’s a way the “salmon people,” as he calls his community, to pay homage to salmon.
To ease this laborious task, the Heiltsuk Nation partnered with Atlas’s team to run a pilot of the AI-based monitoring tool in their weir across the Koeye River. The pilot focused primarily on monitoring sockeye salmon in Heiltsuk territory — a species about which the nation had no information and struggled to make management decisions. After the staff learned how to use the AI-based tool, their productivity transformed: Instead of spending all their time between April and October watching the salmon go by, they could focus 60% of their time on other priority projects.
“It was amazing to be able to sit here in our office and watch the live views of the salmon swimming through the weir and getting notifications on your phone that nine fish just passed through,” Housty said. “It’s unreal to think that that’s actually happening.”
For the Heiltsuk, integrating groundbreaking technology such as AI with their Indigenous knowledge of when and where salmon migrate in the territory is a way to strengthen stewardship over their territories. “It really is an extension of exerting our title and rights over management of salmon populations in our territory,” Housty said. “It’s never before we have seen data produced like this that’s so accurate and so quick and in a form that’s usable for making decisions on the spot.”
Impressed by its performance, the nation plans to use the system in the future too.
Atlas said the AI-based tool could play a vital role in managing salmon populations under a recently signed historic agreement, where coastal First Nations like Heiltsuk work closely with Fisheries and Oceans Canada (DFO), the federal agency responsible for fisheries management, to jointly manage fishery resources in coastal British Columbia. “This information that we’re generating can be foundational to managing salmon under those co-governance agreements,” he said.
By 2025, the researchers plan to roll out the tool with 10 partnered First Nations for monitoring salmon in real time throughout the season. They also plan to build a web application where anyone can upload video clips to automatically count fish numbers.
Atlas said this collective effort of developing a real-time monitoring system is a way to ensure the salmon populations remain healthy and can be fished for centuries to come. “This isn’t just about developing a scientific tool, but it’s about putting food on people’s plates.”
Banner image: Sockeye salmon (Oncorhynchus nerka). Image by Dr. Jonathan Moore.
Spamming streams with hatchery salmon can disrupt ecosystems, study finds
Related listening from Mongabay’s podcast: Indigenous peoples’ long relationship with, and stewardship of, marine environments through two stories of aquaculture practice in New Zealand and Canada. Listen here:
Citation:
Atlas, W. I., Ma, S., Chou, Y. C., Connors, K., Scurfield, D., Nam, B., … Liu, J. (2023). Wild salmon enumeration and monitoring using deep learning empowered detection and tracking. Frontiers in Marine Science, 10. doi:10.3389/fmars.2023.1200408
FEEDBACK: Use this form to send a message to the author of this post. If you want to post a public comment, you can do that at the bottom of the page.
