How our open-source AI model SpeciesNet is helping to promote wildlife conservation
SpeciesNet is a free, open-source AI model that can identify nearly 2,500 species of mammals, birds, and reptiles in camera trap photos, helping conservationists worldwide process millions of images and protect endangered wildlife.
The Problem: Too Many Photos, Not Enough Time
Motion-triggered cameras give researchers unprecedented views of wildlife behavior — but turning millions of candid snapshots into actionable data is incredibly time-consuming. Biologists and conservationists face backlogs spanning years of images that need manual identification.
How SpeciesNet Works
SpeciesNet is an AI model trained to automatically identify animals in camera trap photographs. Available since 2019 via Wildlife Insights and open-sourced as a free tool, it recognizes nearly 2,500 categories of mammals, birds, and reptiles — processing in days what would take human volunteers months or years.
Real-World Impact
Snapshot Serengeti (Tanzania)
The project has operated camera traps in Serengeti National Park since 2010. Initially relying on online volunteers, the team turned to SpeciesNet to analyze a backlog of 11 million photos, processing decades of data in just days. The analysis provides long-term insights into fauna behavior and abundance in one of Africa's most biodiverse regions.
Humboldt Institute (Colombia)
Colombia's Humboldt Institute uses SpeciesNet as part of a national-scale camera trap network called Red Otus. Tens of thousands of images have revealed behavioral shifts — some mammals are becoming more nocturnal, and birds appear later in the morning in developed areas.
Idaho Department of Fish and Game (USA)
The department uses SpeciesNet to sort millions of camera images, tracking wildlife populations and ensuring species health and stability across the state.
Why Open Source Matters
By making SpeciesNet free and open-source, Google has enabled conservation groups worldwide to leverage AI without budget constraints. The model democratizes access to wildlife monitoring technology.
Key Numbers
| Metric | Value |
|---|---|
| Species recognized | ~2,500 |
| Photos analyzed (Serengeti) | 11 million |
| Regions deployed | Africa, South America, North America |
| Availability | Free, open-source |
Source: Google Blog