Google Open Sources SpeciesNet AI to Automate Wildlife Identification
Google releases SpeciesNet, an open source AI model trained on millions of images to automatically identify animal species in camera trap photos. The tool aims to help researchers process massive amounts of wildlife data more efficiently.
Google releases SpeciesNet, an open source artificial intelligence model designed to identify animal species in photos from camera traps. Researchers use these motion-triggered cameras to study wildlife populations, but the devices generate massive volumes of data that take weeks to manually sift through. SpeciesNet aims to solve this problem by automating the identification process.
The tech giant trains SpeciesNet on over 65 million publicly available images contributed by major organizations like the Smithsonian Conservation Biology Institute and the Wildlife Conservation Society. The model classifies pictures into more than 2,000 labels that cover specific animal species, broader taxa like "mammalian," and non-animal objects such as vehicles. This technology currently powers the analysis tools on Wildlife Insights, a Google Earth Outreach platform where researchers collaborate on camera trap data.
Google makes SpeciesNet available on GitHub under an Apache 2.0 license, allowing developers, academics, and startups to use the model for commercial purposes with very few restrictions. This release enters a growing field of open source tools for wildlife monitoring, competing with similar frameworks like Microsoft's PyTorch Wildlife. Ultimately, Google intends for this technology to scale global efforts to monitor and protect biodiversity in natural environments.