SpeciesNet: Google’s AI Breakthrough in Wildlife Identification

Google introduces SpeciesNet, an AI model aimed at identifying animal species through camera trap photos, enhancing wildlife research and data analysis.

Google just dropped SpeciesNet, and it’s a game-changer for wildlife research. This open-source AI model can pick out different animal species just by looking at photos from camera traps—you know, those sneaky little cameras with infrared sensors scientists use to spy on animals. But here’s the kicker: while these cameras are awesome for gathering data, they churn out so many photos that researchers can spend weeks just sorting through them. Talk about a first-world problem in the animal kingdom.

Enter Wildlife Insights, Google’s brainchild from six years ago under its Google Earth Outreach program. It’s like a social network for scientists, letting them share, ID, and analyze wildlife pics online to speed things up. And SpeciesNet? It’s the star player, trained on a whopping 65 million images from big names like the Smithsonian and the Wildlife Conservation Society. That’s a lot of cat pictures, but, you know, the wild kind.

What’s cool is SpeciesNet doesn’t just stop at ‘Hey, that’s a tiger.’ It can sort images into over 2,000 categories—specific species, broader groups, even random objects that photobomb the shots. Google’s hoping developers, academics, and even startups will use it to keep an eye on biodiversity. And the best part? It’s up for grabs on GitHub with an Apache 2.0 license, so pretty much anyone can use it, no strings attached.

But Google isn’t the only tech giant playing wildlife detective. Microsoft’s AI for Good Lab is in the mix too with PyTorch Wildlife, offering its own pre-trained models for spotting and classifying animals. It’s like the tech world’s version of a nature documentary, but with more coding and less David Attenborough.