Google has released SpeciesNet, an AI model that helps researchers identify animals in photos taken by camera traps.
These traps are widely used to track wildlife, but sorting through the huge number of images takes a lot of time.
To speed things up, Google launched Wildlife Insights six years ago, a platform where researchers can upload, identify, and analyse wildlife photos together.
Now, SpeciesNet, trained on over 65 million images from sources like the Smithsonian and the Zoological Society of London, is available as an open-source tool.
Here’s what you should know:
SpeciesNet can recognise over 2,000 labels, including specific species, broader groups like ‘mammals’, and even non-animal objects like vehicles.
It’s open-source on GitHub under an Apache 2.0 licence, meaning developers, researchers, and conservation groups can use it with few restrictions.
Google isn’t the only one in this space, Microsoft’s AI for Good Lab also offers PyTorch Wildlife, a similar AI tool for animal detection.
NatGeo AI
By making SpeciesNet freely available, Google hopes to make it easier to track biodiversity and support conservation efforts worldwide.
Somewhere out there, a squirrel is getting its AI-generated headshot.