Researchers and Indigenous rangers in West Arnhem have tested how artificial intelligence can be integrated into Indigenous-led, community-based wildlife monitoring, reporting significant reductions in the time required to process camera-trap imagery while maintaining local oversight of data and decision-making.
The work involved Warddeken Land Management, Charles Darwin University and the Commonwealth Scientific and Industrial Research Organisation, and focused on incorporating AI into an Indigenous-led community-based monitoring initiative, the Mayh (animal) wildlife program.
The Mayh program covers the Warddeken Indigenous Protected Area in West Arnhem and aims to understand how land management practices affect key mayh of cultural and conservation concern in the region. The program uses camera traps for monitoring and provides employment for Traditional Owners and bininj (male) and daluk (female) rangers.
Following consultation on implementation and safeguards, an AI animal-detecting model, MegaDetector, was fed 1,572 hours of data collected by rangers. The organisations said MegaDetector reduced the number of images requiring analysis from 925,088 to 172,290—an 81.4 per cent reduction—shortening processing time from six months to two months.
The collaboration also reported a reduction in average time spent on data processing from 52.6 hours to 41.8 hours per fortnight. With less time spent on image processing, the groups said overall engagement in land management increased, with more attention able to be directed to species identification.
Rangers also used time freed up by automation for language documentation and for producing resources such as poster reports, newsletters and videos, as well as delivering workshops.
Warddeken Land Management Ecological Monitoring Officer and CDU Northern Institute Postdoctoral Researcher Dr Cara Penton said the priority was protecting local voices, knowledge and jobs and enabling land managers to use emerging technologies.
“In its own abstract way, this tackles the crux of what does AI look like in the workplace to actually make people’s jobs more enjoyable, not that we’re just using AI for everything,” Dr Penton said.
“People are still part of the verification. People are still part of the process.
“We have a term called strong data, and what makes data strong is that people have been part of collecting it, part of verifying and processing it, and how they bring it into their own stewardship system with making decisions about Country is up to them.”
Warddeken Land Management Senior Daluk Ranger and Bininj Nawarddeken woman Suzanna Nabulwad said consultation with elders was important to ensure technology is used appropriately on Country.
“It is important to talk to our elders so we use technology the right way on Country, that way we can make sure we look after it and protect the animals,” Ms Nabulwad said.
“I have been a ranger for a long time. Using technology and doing work with cameras and looking at the images, it helps make the job more interesting. It is important to teach new rangers how to do it so that they enjoy their jobs.”
The research paper, Reimagining artificial intelligence applications for Indigenous-led community-based wildlife monitoring: Insights from biodiversity surveys in the Warddeken Indigenous protected area in northern Australia, was published in the journal Conservation Science and Practice.
