Revolutionizing Maritime Monitoring: The Atlantes AI Model Tackles Illicit Fishing

In a significant stride towards combating illegal fishing, researchers have introduced a groundbreaking open-source AI model capable of accurately identifying the activities of nearly all of the world’s seafaring vessels. This innovative technology not only tracks vessel movements but also assesses whether boats may be engaging in illicit fishing practices.

Developed by the Seattle-based Allen Institute for AI (Ai2), the new model, named Atlantes, is designed to analyze an astonishing five billion GPS signals daily from close to 600,000 ocean-going vessels. This artificial intelligence marvel can predict the activities of vessels with approximately 80% accuracy in real-time, making it an invaluable tool for maritime authorities.

How Atlantes Works

Integrated into Ai2’s broader maritime monitoring platform known as Skylight, the Atlantes model functions by processing Automatic Identification System (AIS) messages—data that nearly every ship is mandated to broadcast. The model was trained using data collected from January 2022 to June 2024, utilizing NVIDIA H100 Tensor Core GPUs and PyTorch to ensure optimal performance.

With the help of 20 maritime experts who meticulously annotated over 15 million location-related signals, Atlantes can effectively identify fishing behaviors and other vessel activities. When suspicious behavior is detected, the system can issue alerts to coast guards or patrolling vessels within approximately 15 minutes, enabling swift action against potential illicit activities.

Real-World Impact: Success Stories

One of the model’s first significant successes occurred earlier this year when Argentina’s Coast Guard intercepted a vessel engaged in illegal fishing within its exclusive economic zone, thanks to an alert generated by Skylight. This incident underscores the potential of AI-driven solutions to enhance maritime law enforcement and protect valuable marine resources.

Data Processing Capabilities of Skylight

The Skylight platform is a data powerhouse, capable of processing nearly two terabytes of information each day. This includes a rich mix of satellite information, radio frequency data, and vital AIS location data. By leveraging public data from sources such as NASA and the European Space Agency’s Sentinel-2, Skylight provides a comprehensive monitoring solution that is both precise and efficient.

Accessible via web and API, Skylight offers its services free of charge to governments and conservation organizations, emphasizing the importance of collaboration in tackling illegal, unreported, and unregulated (IUU) fishing.

The Economic and Ecological Stakes

The ramifications of IUU fishing are profound, particularly for economically disadvantaged countries. According to the Financial Transparency Coalition (FTC), IUU fishing results in annual losses of up to $23 billion, representing around 20% of the global fisheries catch. Alarmingly, nearly half of all vessels involved in IUU fishing operate in African waters, where local communities heavily depend on fishing as their primary source of income and nutrition.

Namrata Kolla, Ai2’s program lead for Skylight, articulated the project’s mission: “The purpose of what we’re doing is to help under-resourced nations protect their waters, their natural resources, and their livelihoods.” The integration of AI technologies empowers enforcement agencies to identify critical behaviors within vast datasets, enhancing their capacity to combat illegal fishing effectively.

Future Aspirations for Atlantes

Ai2’s ambitions for the Atlantes model extend beyond maritime applications. Plans are underway to adapt the model for integration into another conservation platform, EarthRanger, which aggregates diverse datasets from various sources, including satellites and sensor networks. This platform is particularly focused on wildlife conservation, providing authorities with a comprehensive view of relevant data overlaid on geographical maps.

One exciting direction involves training the Atlantes model to predict elephant behavior, addressing the pressing issue of human-elephant conflict. Jes Lefcourt, EarthRanger’s director, highlighted that poaching and conflicts with farmers are significant threats to elephants, making this initiative both timely and crucial for conservation efforts.

Conclusion: A New Era for Conservation Technology

Through its decade-long commitment to global conservation, Ai2 has built the world’s largest dataset of elephant movements, continuously tracking tens of thousands of collared elephants. By analyzing this data, Ai2 aims to identify patterns leading up to conflicts between elephants and humans, ultimately enabling proactive interventions.

As Ai2 harnesses advanced AI models like Atlantes, the intersection of technology and conservation becomes increasingly evident. By providing actionable insights and real-time alerts, these innovations pave the way for more effective protection of marine and terrestrial ecosystems alike.