The Nature Conservancy (TNC) released its full AI-powered electronic monitoring solution as open source after three years of prototyping and stress-testing in the Eastern Tropical Pacific. Alan Descoins, CEO of Tryolabs, said: "Reviewing EM footage used to take months. With edge AI running onboard the vessel, we brought that down to minutes. The system achieves a 6% catch count error compared to expert reviewers and generates daily risk reports at sea, delivering timely, actionable intelligence to decision-makers."
Ben Gilmer, director of large-scale fisheries at The Nature Conservancy, said: "For electronic monitoring to truly work at a global scale, it has to be accessible, affordable and fast. That's why we're making this technology freely available—to allow governments, scientists and industry to build from the same foundation and accelerate a future where verified on-the-water activities and accurate fisheries data are not just an aspiration, but a global standard critical to science based-management, market access and combating illegal, unreported and unregulated fishing."
Electronic monitoring uses onboard cameras, GPS and sensors to document fishing activity, but adoption has been limited by the high cost and logistics of reviewing massive volumes of video. Hard drives are often shipped to human review centers and analyzed months after a vessel returns to port, after seafood has already entered the supply chain. Without independent monitoring, catch reporting is largely unverifiable, creating conditions for illegal, unreported and unregulated fishing.
The system uses edge computing to analyze electronic monitoring video directly onboard longline fishing vessels, which are equipped with Starlink terminals that transmit daily summary reports over secure Wi-Fi. Computer vision models detect, track and classify catch as it is brought onboard. The system compares its predictions to captains' electronic logbooks each day and flags potential non-compliance before vessels return to port. It is designed to complement, not replace, human expertise: reviewers receive daily summary reports with AI predictions on species-level catch insights, fishing risk profiles and associated footage, with experts kept in the loop to validate AI predictions.
TNC collaborated with partners over the past two years to develop and pilot the system, with Tryolabs setting performance benchmarks and building the full solution stack now available to the public. The prototype was selected as one of 15 global awardees in the Bezos Earth Fund's AI for Climate and Nature Grand Challenge, and the $2 million award will help bring the system to scale in the Western and Central Pacific Ocean, including The Republic of Palau.
Details