On the remote island of Stora Karlsö in the Baltic Sea, researchers are pioneering a new model for ecological research that fuses advanced video surveillance, artificial intelligence, and sensor technology to study seabirds and their rapidly changing environment.
Led by the Swedish University of Agricultural Sciences (SLU), this ambitious project relies on Milestone Systems’ XProtect open platform video management software (VMS) to automate the monitoring of one of Europe’s most important seabird colonies, while enabling data-driven insights into the effects of climate change and ecosystem shifts.
For years, researchers on Stora Karlsö studied seabird behavior manually by observing colonies through binoculars and jotting notes in the field. But these methods often missed critical, short-lived events such as predator attacks or heatwave-driven chick fatalities. With mounting pressure to understand ecological impacts in real time, the team needed a non-intrusive system to collect continuous, high-resolution video while integrating additional data from thermal cameras, audio sensors, and weighing scales.
The island’s harsh coastal conditions, limited electricity, and lack of wired internet further complicated the challenge. Traditional systems could not scale or support the required level of automation and scientific analysis. The research team required a flexible platform that could operate in a remote environment and integrate seamlessly with their growing ecosystem of sensors and AI tools.
To address these challenges, the SLU team partnered with Swedish systems integrator Prudencia Security to deploy a video monitoring solution based on Milestone XProtect VMS. The open platform architecture of XProtect made it possible to integrate a mix of IP cameras from Axis Communications, ranging from cliff-mounted colony monitors to forest-edge nesting views.
Thermal imaging devices and high-quality microphones were also used, and plans for automated weighing stations will be deployed soon. Prudencia designed and pre-tested a custom server rack with UPS protection to manage the island’s unstable power grid and used satellite-based Starlink internet, with future support for 5G remote access.
To ensure long-term data accessibility, the integrator also developed a custom export utility that archives one-hour video files for offline AI model training. This flexible, resilient setup enables researchers—most of whom are not engineers—to configure views, retrieve past recordings, and build automation workflows directly atop the video stream. The result is a self-sustaining, scalable monitoring platform ready to support new scientific discovery.
The system is already transforming research operations on the island. Scientists can now go back in time to identify causes of nest disturbances or chick mortality, detect predator activity like eagle intrusions, and correlate heat-related stress behaviors with microclimate data. Milestone’s interface makes it easy to scan footage, compare seasons, and extract high-fidelity data for AI processing.
Advanced models can identify specific birds, measure chick growth over time, and even recognize behavioral signals such as wing spreading in response to temperature spikes. With synchronized audio and video, researchers are beginning to triangulate vocalizations with precise bird actions. What was once anecdotal and delayed is now empirical and immediate, offering new avenues for ecological research, biodiversity protection, and scientific storytelling.
The deployment on Stora Karlsö underscores how Milestone’s platform can be repurposed as a scientific tool. The ability to manage video, metadata, and third-party sensors within a unified interface has created a dynamic, research-grade monitoring solution that supports both real-time observation and long-term archival analysis.
Read the full case story and see the video for details about the challenging remote location, and how custom integrations enable new frontiers. It’s a model solution for the future of environmental monitoring, using multiple sensors and AI learning.
