Google's open-source model "SpeciesNet" simplifies species annotation tasks, enhancing conservation efficiency

Google announces the open-source release of the SpeciesNet model to the public, greatly improving scientists’ efficiency in studying species. Through collaboration with Wildlife Insights, the model can accurately identify over 2,500 mammal species worldwide. This artificial intelligence technology not only frees scientists from tedious manual screening but also enables precise image analysis, opening a new chapter in protecting Earth’s life.

What is Google SpeciesNet?

SpeciesNet is an open-source AI model developed by Google that is trained to automatically recognize nearly 2,500 species of mammals, birds, and reptiles. Increasingly, organizations and academic institutions are using SpeciesNet for conservation research. Since 2019, the model has been in use, and a year ago, Google released it as a free open-source tool. Now, research teams are using this model to analyze and organize image data. SpeciesNet can identify species from multiple angles and under different lighting conditions, even when only part of the animal is visible.

How does SpeciesNet work?

SpeciesNet runs in the Google Cloud environment. It helps Wildlife Insights users annotate images. Any verified annotation can, in turn, provide training data for SpeciesNet. Wildlife Insights is a community platform hosting 200 million annotated images.

SpeciesNet addresses the biggest bottleneck in traditional conservation work: data processing speed.

Features include:

• Massive recognition: Can identify nearly 2,500 species of mammals, birds, and reptiles.
• High accuracy: Recognition accuracy reaches up to 99.4%.
• Rapid processing: Even with a standard laptop, it can process 30,000 photos per day; with a GPU, over 250,000 photos.

How does SpeciesNet assist research?

Cameras can capture animal activity 24/7, but converting millions of images into usable data is time-consuming for wildlife managers, biologists, and conservationists. The Humboldt Institute in Colombia uses SpeciesNet to monitor species living in the Amazon rainforest, analyzing tens of thousands of collected images. They discovered changes in bird migration timing and daily activity patterns of wildlife, with results showing that mammals have become more nocturnal to avoid threats and predators.

The Idaho Department of Fish and Game (IDFG) deployed hundreds of cameras in northern forested areas. Using SpeciesNet to classify images by species significantly sped up the review process of millions of images collected annually.

Australia hosts many unique species not found elsewhere. WildObs in Australia uses SpeciesNet to identify locally important rare species, which are key to monitoring and conservation efforts. Trained with AI, SpeciesNet helps organizations monitor threatened and endangered species, maintaining the ecological balance of wild populations.

This article about Google’s open-source model “SpeciesNet” simplifying species annotation and enhancing conservation efficiency first appeared on Chain News ABMedia.