Google open sources SpeciesNet AI model for identifying animals from camera feeds

Google announced that it is open sourcing its AI SpeciesNet model that helps users identify animal species by analysing photos sourced from camera traps set up to monitor wildlife.

SpeciesNet was trained on a dataset that had over 65 million images, per Google, and there are more than 2,000 labels under which to categorise subjects, such as “mammalia” or “felidae,” or to identify non-animal objects such as vehicles. An object detector and an image classifier work together in order to find the animal in the feed and then identify its species.

However, SpeciesNet is not a new AI model release and has been used by experts for years.

“Since 2019, thousands of wildlife biologists have used SpeciesNet through a Google Cloud-based tool called Wildlife Insights to streamline biodiversity monitoring and inform conservation decision-making,” said Google’s Head of Sustainability Programs and Innovation, Mike Werner, in a blog post on March 3.

“The SpeciesNet AI model release will enable tool developers, academics and biodiversity-related startups to scale monitoring of biodiversity in natural areas,” said Werner.