Scientists at the University of Bristol, UK, have helped develop a powerful new tool called SA-FARI, which can automatically find, identify, and follow individual animals in video footage, according to a study published in the journal arXiv. The breakthrough could save researchers around the world an enormous amount of time and transform how we monitor and protect wildlife.
What SA-FARI does
SA-FARI stands for “Segment Anything in Footage of Animals for Recognition and Identification.” It’s the result of an international collaboration led by ConservationX Labs and Meta, with the Bristol team playing a key role. The system is built on Meta’s latest Segment Anything Model 3 (SAM3), a cutting-edge AI that can understand both text and visual instructions to pick out and follow objects in images or videos.
Here’s how it works in practice: the AI creates what researchers call “masklets”, essentially precise outlines of an animal as it moves frame by frame through a video. This means the animal can be cleanly separated from its background, making it much easier to study its behaviour, distinguish one individual from another, and track how it responds to its environment.
For anyone who has ever scrolled through endless hours of camera trap footage, the appeal is obvious and the potential is huge.
Why it matters for conservation
Knowing where animals are and what they’re doing is the foundation of wildlife protection. Without that information, it’s hard to tell whether conservation efforts are actually working.
Dr Otto Brookes, Lecturer in AI and Animal Biometrics at Bristol, explained: “The ability to locate animals in space and time is incredibly important for wildlife monitoring – it is a prerequisite for many tasks such as recognising behaviour and distinguishing individuals from one another and ultimately measuring how animals respond to conservation interventions.”
In short, before you can protect a species, you need to be able to see it clearly and SA-FARI helps with exactly that.
A massive team effort
To build the system, the team put together a huge dataset: more than 11,000 wildlife videos filmed in natural habitats, all carefully annotated by hand. The AI was then trained to recognise nearly 100 different species and follow them pixel-by-pixel through footage.
Even better, the dataset is being made freely available to biologists, researchers, and conservationists everywhere, so projects around the world can benefit from the technology.
The project pulled together expertise from across the globe, including the Hasso Plattner Institute, the University of Oviedo, Osa Conservation, the Senckenberg Museum of Natural History, the Max Planck Institute for Evolutionary Anthropology, and Climate Corridors.
Tilo Burghardt, Professor of Computer Vision and Animal Biometrics at Bristol, said: “Global problems require global solutions. Based on the group’s pioneering track record of over 20 years, the University of Bristol is regarded as one of the go-to places for using AI for conservation in the UK and beyond, and is an important part of a growing international community working in this area.”
Recognition on the world stage
The SA-FARI paper is being presented on Saturday 6 June at the Conference for Computer Vision and Pattern Recognition (CVPR) in Denver, USA, widely seen as the most prestigious event in visual AI. The paper has also been selected as an Award Candidate, marking the second year in a row that the Bristol team has earned this honour.
What’s next
Professor Burghardt believes SA-FARI is just the beginning. In the future, the system could be expanded to track animals’ body posture, estimate depth, and even generate natural language descriptions of what the animals are doing. For wildlife conservation, that could mean smarter monitoring, faster research, and ultimately a better chance of protecting the species that share our planet.
Wasmuht D, Brookes O, Schall M, Palencia P, Berne C et al. The SA-FARI Dataset: Segment Anything in Footage of Animals for Recognition and Identification, 2026. arXiv:2511.15622