Queensland University of Technology’s Dr Fernando Vanegas Alvarez has been awarded an Advance Queensland Industry Research Fellowship $240,000 grant for a project to develop a proven framework for classifying and mapping invasive plant species using drone-collected imagery.
Dr Vanegas Alvarez, from QUT’s School of Electrical Engineering & Robotics and QUT Centre for Robotics, said his research project would identify the most appropriate protocols for aerial and ground data collection and the best AI models for each species and site.
“Invasive plants have negative effects on agriculture and farming, damage existing infrastructure, and in some cases cause irreversible damage to unique ecosystems,” Dr Vanegas Alvarez said.
“My industry partner, the Sunshine Coast Council, is committed to using innovative technologies to better understand the extent of invasive plants, how they spread and their effects on ecosystems to inform on-ground decisions.”
Dr Vanegas Alvarez is working with drone operator Aspect UAV Imaging to develop a standardised methodology to survey and collect data, detect different species and then map vegetation using aerial imagery.
“Fine detail and high-resolution mapping of vegetation and weed species are required to understand and assess the environmental impact and support decision-making,” Dr Vanegas Alvarez said.
“Species classification in natural environments is challenging owing to high floristic and spectral diversity, as well as a mix of light in pixels reflected by multiple species.
“However, drones equipped with ultra-high resolution, visible-light (RGB), multispectral (MS) cameras, and high-precision global positioning systems can capture imagery over 4ha of land in one flight.
“Their high resolution enables us to distinguish individual leaves and patterns of plants and multispectral cameras capture light reflected by plants in wavelengths outside the visible spectrum for humans.”
Dr Vanegas Alvarez said camphor laurel, cat’s claw creeper and Madeira vine were among the Sunshine Coast’s targeted invasive species.
“Our proposed georeferenced, high-resolution mapping could be combined with other geographical features such as creeks, transport corridors to understand how these species spread and the potential vectors that allow their proliferation.”
