Dante Wasmuht clicks play on a video of a Great Tinamou, a shy, ground-dwelling bird of Central American forests, and within seconds, a shimmering outline traces the animal’s every movement—frame by frame, with pixel-perfect precision. This is SA-FARI in action, a breakthrough AI system that can detect, name, and track nearly 100 wildlife species in video footage, transforming how conservationists study animals in the wild. Developed by an international consortium led by ConservationX Labs and META, with key contributions from the University of Bristol, SA-FARI builds on META’s advanced Segment Anything Model 3 to deliver unprecedented accuracy in wildlife monitoring. At its core, the technology uses 'masklets'—dynamic outlines that follow individual animals across time—enabling researchers to isolate animals from their backgrounds and unlock detailed behavioral and population insights.
This isn’t just a technical achievement—it’s a practical revolution for conservation. Camera traps generate millions of hours of footage, most of which is reviewed manually, a process that can take researchers months or even years. SA-FARI slashes that time, automating what once required painstaking human effort. The system was trained on over 11,000 wildlife videos from natural habitats, making it one of the most comprehensive datasets of its kind—and it’s freely available to scientists and conservationists worldwide. The work has earned global recognition, with the SA-FARI paper selected as an Award Candidate at the Conference for Computer Vision and Pattern Recognition (CVPR) in Denver, marking the University of Bristol’s second consecutive year receiving this honor in AI for conservation.
Dr. Otto Brookes of the Bristol team emphasizes the real-world impact: “The ability to locate animals in space and time is incredibly important for wildlife monitoring—it is a prerequisite for many tasks, such as recognizing behavior and distinguishing individuals from one another.” The team, led by Professor Tilo Burghardt, has spent over two decades pioneering AI applications in animal biometrics, positioning Bristol as a global leader in this emerging field. Collaborators span institutions from the Max Planck Institute to Osa Conservation, reflecting the project’s truly interdisciplinary reach. Looking ahead, the platform could be expanded to track animal body pose, depth, and even generate natural language descriptions—opening new doors for ecological research.
As climate change and habitat loss accelerate, tools like SA-FARI offer more than efficiency—they offer hope. By turning raw footage into actionable data at scale, this technology empowers conservationists to respond faster, smarter, and with greater precision. In the quiet tracking of a single tinamou, there’s a vision of a future where AI doesn’t replace human care, but amplifies it.