Conservation X Labs, Meta Launch Dataset to Transform AI

Conservation X Labs (CXL), in partnership with Meta, has introduced a groundbreaking resource for conservation and AI research: the world’s largest open collection of annotated camera-trap wildlife videos. This newly released dataset features more than 10,000 expertly labeled clips representing nearly 100 species captured across multiple continents. By offering such detailed and diverse footage, the dataset stands as a major milestone in the advancement of ecological monitoring and AI-powered wildlife analysis.

Named SA-FARI, the dataset fills a long-standing void in conservation technology. For years, researchers have lacked high-quality, annotated wildlife video data at scale. While AI models built on static images have contributed value, they cannot fully capture the nuances of animal behavior, movement patterns, subtle health indicators, or interactions between species. Camera-trap video, however, provides a much richer and more dynamic view of ecosystems. Until now, the absence of large-scale annotated video benchmarks has significantly slowed progress in building advanced computer vision models for environmental research.

SA-FARI seeks to change that. It includes detailed annotations applied frame-by-frame, featuring bounding boxes and segmentation labels. This level of precision allows AI models to track animals across complex scenes while generating more accurate insights. Importantly, the dataset is powered by Meta’s Segment Anything Model 3 (SAM 3), an advanced computer vision model capable of identifying, outlining, and tracking objects across every frame of video. Contributions came from an array of global conservation partners, including Osa Conservation in Costa Rica, the Los Amigos Biological Station in Peru, and organizations across Central Africa.

Meta’s Kate Saenko emphasized the importance of the collaboration, stating, "I'm very excited about our partnership with Conservation X Labs and the impact it could have on computer vision research for wildlife monitoring. Challenging benchmarks are very important for driving the AI field forward. We hope our new benchmark will spark new research and lead to better tools for monitoring species in wildlife camera video streams."

True to its mission of promoting open science, the dataset is freely accessible with no limitations. A research paper detailing SA-FARI is now available on arXiv, offering deeper insights into the dataset and its potential applications.

Reflecting on the organization’s decade-long contributions to nature tech, Dr. Alex Dehgan, CEO & Co-Founder of Conservation X Labs, shared, "For a decade, Conservation X Labs has helped pioneer the field of nature tech, pushing forward breakthroughs that expand both the speed and the scale of conservation. Through our partnership with Meta, we can now share two transformative tools a powerful foundation model and an expertly annotated dataset to reimagine how we monitor animals and understand the behaviors that define their adaptation to a changing world. We believe deeply in the promise of technology to inspire hope, spark innovation, and help protect the planet we all depend on."

With SA-FARI, the conservation and AI communities now gain access to an unprecedented digital resource designed to accelerate innovation, strengthen wildlife research, and support global biodiversity protection.

To join our expert panel discussions, reach out to info@intentamplify.com

Contact Center Technology Insights:

• MCE Welcomes Ex-T-Mobile Exec Aleeta Willcuts to Enhance Telco CX 
• ServiceNow Eyes Marketing CRMs After Big Genesys Deal 
• Strolid & Frontline Group Redefine Contact Centers with Virtual CX