Parks Victoria Launches Open-Source AI Tool to Accelerate Native Species Conservation
Key Takeaways
- ▸AI species recognition tool processes 20 images per second with 95%+ accuracy, reducing manual review time from weeks to hours
- ▸Available as free, open-source platform enabling land managers, conservation groups, academics, and citizen scientists to access advanced wildlife monitoring
- ▸Immediately deployed for post-bushfire conservation triage in Victorian Otways, directing feral animal control to protect endangered species
Summary
Parks Victoria has deployed a groundbreaking artificial intelligence species recognition tool that can process wildlife camera trap images at unprecedented speed, identifying over 200 native and feral Australian species with 95%+ accuracy. Developed collaboratively by Parks Victoria conservation officers Dr. Erin Nash and Mary Thorpe alongside Dutch data scientist Peter van Lunteren, the AI tool processes 20 images per second—a task that previously took weeks of manual review. The system has been made available on a free, open-source platform for land managers, conservation groups, academics, Traditional Owners, and citizen scientists to use across Australia.
The tool proved its value immediately when Dr. Nash processed 39,000 images from post-bushfire monitoring in the Otways region in just hours, revealing critical threats from invasive species like red deer and pigs. This rapid analysis enabled Parks Victoria to deploy targeted feral animal control programs to protect vulnerable species such as potoroos, bandicoots, and swamp antechinus. The AI system was trained using five million images provided by 18 Victorian research and conservation organizations, including Zoos Victoria, the University of Melbourne, and Deakin University, ensuring it is specifically optimized for Australian fauna and local conditions.
- Trained on 5 million images from 18 Australian research and conservation organizations, optimized for local native and invasive species
