A Faster, Smarter, and More Accurate Tool for Combating Illegal Fishing
The fight against illegal, unreported, and unregulated (IUU) fishing has significantly progressed with Skylight's latest real-time fishing event detection breakthrough. Powered by the latest state-of-the-art machine learning model built in-house by our team at Ai2, Skylight's AI can now identify fishing activity within minutes. This is a significant improvement from the previous 6-hour average delay. This real-time capability is a game-changer, enabling conservationists and enforcement agencies to respond to potentially illegal fishing happening at the moment, protecting our ocean more effectively than ever.
Breaking Barriers with Better, Faster, and More Comprehensive Detections
For enforcement agencies and protected area managers, time and accuracy are crucial for making real-time decisions to combat IUU fishing effectively. Skylight’s new Fishing Events showcase the potential of AI when applied to extraordinarily rich and complex datasets like AIS. By processing approximately five billion AIS messages daily—at a rate of 50,000 messages per second—and delivering actionable insights through the platform within minutes, Skylight empowers those on the frontlines with the information they need, exactly when they need it.
The Skylight team has continued striving toward world-class excellence with its AI-driven analysis needed to inform critical, time-sensitive, and resource-intensive action. As of October 2024, in the latest fishing model, the precision of Fishing Event detection has improved significantly, achieving nearly 80% accuracy due to dedicated efforts in minimizing false positives. For perspective, even maritime experts only agree on what fishing behavior is 85% of the time due to the inherent complexity and subjectivity of analyzing AIS data. This advancement ensures users can act on Skylight information with greater confidence, whether enforcing regulations or managing marine protected areas.
The revamp of how users view fishing detections in Skylight has also introduced several noteworthy features. Maritime analysts can now see multiple Fishing Events from a single vessel within a day. Combined with higher recall rates for unknown vessels and enhanced buoy identification, this helps paint a far more detailed picture of fishing activity in any given area.
“This breakthrough marks a turning point for Skylight in the fight against illegal fishing,” said Namrata Kolla, leader of the Skylight program. “The ability to detect fishing activity in real-time allows our partners to push the boundaries of what’s possible in ocean conservation, particularly the ability to respond and take action. For Skylight, it also paves the way for exciting new developments we’re set to explore in 2025, like nowcasting, forecasting, and building AI models that truly make a difference for the oceans."
Preparing for the Future of Ocean Protection
While these advancements mark a major milestone, Skylight is already planning for more. Future enhancements will integrate vessel information from Global Fishing Watch, allowing users to conduct deeper analyses by correlating Skylight’s real-time data with detailed vessel registries. This step will elevate investigative capabilities and further strengthen efforts against IUU fishing. Additionally, Ai2’s commitment to open-sourcing the Atlas model and dataset will also pave the way for broader community engagement. These efforts demonstrate Skylight’s ongoing goal to deliver the critical information and tools needed to help those on the frontlines stay ahead in the ongoing battle to protect our oceans.