Important points
- Nallawilli Bunjil is an Indigenous-led technology company that uses drones and data modeling to support environmental land management.
- Through CSIRO’s Kickstart program, the company worked with researchers to build a machine learning model that can detect and classify vegetation from drone imagery data.
- In an ongoing collaboration with CSIRO, Nallawilli Bunjil is developing hyperspectral algorithms to detect early signs of plant stress and land health changes.
The Alpine Jaismatan Nation is under increasing pressure.
Climate change is reshaping the region’s ecosystems, making extreme weather events more frequent and making it increasingly difficult to monitor vast, remote areas with traditional tools.
In the legend of Jaismatan, Bunjir, the long-tailed eagle and great creator, watches over the country from the sky. The responsibility based on this tradition continues today. It is about taking care of our totems and passing on a healthy country to future generations. Totems are relatives, whether animals, plants, or landforms. To protect the totem is to protect the spiritual and ecological health of the entire population.
It is from this story that Nallawilli Bunjil takes its name and purpose. Established to protect the sacred totems and biodiversity of Jaismatane Bhimbul in the Victorian Highlands, the company trades as BunjilView.a modern-day “eye in the sky.”
Narawiri Bunjir is currently expanding its collaboration with traditional owners and land managers across Australia.
Nallawilli Bunjil uses drones, machine learning, and indigenous knowledge to monitor, protect, and improve the nation’s health while creating new tools, jobs, and economic futures for indigenous communities.
This research is supported by a partnership with the Pauline E. McLeod Foundation.ensuring that our community, country and future generations remain at the heart of every project.
Unleash the power of data
Nallawilli Bunjil had already acquired high-quality aerial data and created visually impressive three-dimensional models of the land.
The problem, as Chief Commercial Officer Andrew Basford has articulated, was that these images were essentially impressionistic images, rich in detail but limited in actionable information.
“What the team needed was a way to extract deeper meaning from the data, to start to understand what the land was telling us beyond what it looked like,” Basford said.
“This meant combining multiple sensors that capture color, heat, and structure and developing a machine learning workflow that can turn raw data into land management insights.”
The challenge was that while Nallawilli Bunjil had the vision and the data, it didn’t yet have the in-house ability to define a starting point. They turned to CSIRO’s Kickstart program for help.
From ideas to actionable analysis
Kick-Start provides dollar-worth of funding to start-ups and small businesses to undertake research and development projects with CSIRO researchers, from product prototyping and testing to technical problem solving. For Nallawilli Bunjil, it offered something equally valuable: clarity.
The program connected the company with CSIRO researcher Dr Nariman Havili and worked with the Narawili Bunjir team to define the problem, identify a practical starting point and translate the concept into a working vegetation detection model.
“From a technical perspective, this collaboration was about turning complex sensor data into something that can generate real insights. We developed a computer vision pipeline that can detect and classify vegetation from drone data and turn visually rich images into actionable output,” said Dr. Havili.
“What characterized this project was a thorough and iterative process of translating the initial vision into a practical, deployable model. In doing so, we established a strong foundation from which future work can extend to hyperspectral analysis (reading hundreds of optical wavelengths).”
The project was originally scheduled for six months, but was extended to 18 months as the team tackled the complex task of converting raw drone imagery into a working model.
The result is a computer vision model that can identify and classify vegetation types from drone images, turning beautiful photos into structured data that land managers can act on.
“Kick-Start helped us clarify our goals, identify where to start, and translate concepts into practical solutions. CSIRO provided technical guidance, mentoring, and logical, grounded advice to help us move from idea to action and integrate cultural knowledge and technology,” Basford said.
canary in the country mine
The trained model is currently being refined and transferred to Nallawilli Bunjil’s in-house platform, and the team is already planning its next evolution.
A second collaboration with CSIRO is underway and will focus on improving the technology to detect subtle changes in plant health. In addition to identifying vegetation in images, it will also be able to see signs of stress before they are visible to the human eye.
“We will improve accuracy and act as the ‘canary in the mine’ for land management and biodiversity, enabling early insights into plant health, stress and early changes due to climate change,” Basford explained.
Our long-term vision extends beyond the Victorian Highlands. Having proven the model in Jaithmathang Country, Nallawilli Bunjil has already expanded its approach to other regions across Australia, applying the same blend of geospatial precision and Indigenous knowledge to protect sensitive landscapes and the communities that depend on them.
Advice for companies considering Kick-Start
This collaboration also opened the door to the wider CSIRO ecosystem beyond the Kick-Start program, allowing Nallawilli Bunjil’s work to impact the tool far beyond the original project. The collaboration included a partnership with Terria, a spatial data platform originally developed by CSIRO and spun out in 2024. Integration with Spark, an AI toolkit for fire prediction and analysis developed by CSIRO.
“What started as a single technical collaboration quickly grew into something much more. Our collaboration with CSIRO has not only helped us solve one problem, but also connected Indigenous-led innovation to a broader ecosystem, allowing our knowledge and tools to impact national platforms and support real-world outcomes far beyond a single project.”
For Nallawilli Bunjil, the value of the program goes far beyond its technical accomplishments, and Basford’s advice to others considering Kick-Start is clear:
“Our advice to anyone considering a Kickstart program is to leverage CSIRO’s guidance and deep expertise to solve problems. Our journey has prepared us to turn ideas into reality and plan the next evolution of detection algorithms,” he concluded.
