Seabirds typically inhabit marine environments such as seas and wetlands, where food sources like fish and insects are abundant. However, seabird populations are declining rapidly—faster than many other bird species—due to threats like pollution, commercial fishing, habitat degradation, mineral extraction, and human disturbance in coastal regions. Predation from both land and water species also poses significant risks. Moreover, seabirds have limited natural defenses; they often nest and lay eggs in open habitats like bare ground, sandy shores, or rocky islands, with little to no nesting material, leaving their young vulnerable.

Among the 350+ seabird species, the Royal Tern and Caspian Tern are notable examples. Adult terns range from 45 to 60 cm in length and weigh between 350 to 750 grams, making them relatively small targets for visual detection. As a result, very high-resolution imagery is essential for accurate monitoring.

Recent advances in drone technology and AI have made it possible to collect high-resolution imagery across large geographic areas and detect small objects like seabirds with precision. Drones offer the added advantage of minimal disturbance to bird populations and are well-suited for frequent surveys—even in post-disaster scenarios such as hurricanes or oil spills.

This deep learning model automates the detection of seabirds—specifically Royal and Caspian terns—from high-resolution aerial imagery, enabling efficient monitoring and helping identify priority areas for seabird habitat protection.