SpeciesNet: The Open-Source AI Boosting Conservation

Since SpeciesNet was made open-source a year ago, its use has spread among researchers and wildlife enthusiasts, facilitating the identification of animals and the study of their habitats. This artificial intelligence (AI) model aims to promote wildlife monitoring and conservation on a global scale.

Motion-triggered cameras provide a unique perspective on animal behavior in the absence of humans, capturing scenes ranging from a puma in the forests of Colombia to a cassowary in Australia. However, for wildlife managers, biologists, and conservationists, the challenge lies in processing millions of these images to extract useful data. This is where SpeciesNet comes into play.

SpeciesNet is an AI model specifically designed to automatically identify nearly 2,500 categories of mammals, birds, and reptiles. Since 2019, it has been integrated into the Wildlife Insights platform and, for the past year, has been available as a free and open-source tool. This allows research groups to analyze data from camera traps more efficiently.

African Partner: Snapshot Serengeti

The Snapshot Serengeti project operates in the Serengeti National Park in Tanzania, in collaboration with the Tanzanian Wildlife Research Institute since 2010. Initially, the project relied on online volunteers to analyze the images, but the volume of data was too large. Todd Michael Anderson, the project lead at Wake Forest University in North Carolina, used SpeciesNet to process a backlog of 11 million photos, transforming decades of data in just a few days. This analysis provides an overview of wildlife behavior and abundance in one of Africa's most biodiverse regions.

South American Partner: Humboldt Institute of Colombia

In Colombia, the Humboldt Institute uses SpeciesNet via the Wildlife Insights platform to monitor species living in the Colombian Amazon rainforest, a region of exceptional biodiversity that is rapidly changing. The Red Otus project was recently launched to capture images on public and private lands across the country. This national network has allowed for the analysis of tens of thousands of images, revealing changes in bird migrations and daily wildlife behaviors. Analyses indicate that some mammals are becoming more nocturnal to avoid threats, while birds are appearing later in the morning in developed areas, likely to escape predators.

North American Partner: Idaho Department of Fish and Game

In the United States, the Idaho Department of Fish and Game (IDFG) also uses SpeciesNet to identify animals captured by their camera traps. While aerial surveys are regularly conducted in southern Idaho, the agency deploys hundreds of camera traps, particularly in the northern forests of the state. The images are first sorted by SpeciesNet, significantly speeding up the review process by human experts, thus facilitating the analysis of millions of images collected each year.

Australian Partner: Wildlife Observatory of Australia

In Australia, the Wildlife Observatory of Australia (WildObs) has adapted SpeciesNet to identify local species not included in the initial model. Australia, with its many endemic species, requires particular monitoring. A version of SpeciesNet trained on local wildlife allows for the tracking of iconic, threatened, or endangered species specific to the region. This helps protect these unique species and ensure their conservation.