Using aerial imagery and computer learning to detect lignum on the floodplain

Monitoring lignum is challenging. Lignum shrublands can be difficult to access, particularly when conditions are wet, and they are difficult places to move through. Current condition assessment approaches rely on visual assessments which are subjective and methods vary across jurisdictions. With the emergence of drone-based techniques for vegetation monitoring (e.g.(Higgisson et al. 2021, McCann et al. 2022)), there is an opportunity to develop approaches for monitoring the condition of lignum using aerial imagery. The first step in that process is to investigate if aerial imagery and machine learning approaches can identify lignum of varying condition on the floodplain and in the wetlands of the Murray Darling Basin.

Project Objectives: Determine if aerial imagery can be used to identify lignum on the floodplain to determine the feasibility of a broad cost-effective assessment of lignum health.

This objective can be divided into three aims:
1. To determine if lignum of varying condition can be identified from aerial images from floodplain and wetland environments.
2. To determine if machine learning can be used to accurately identify lignum of varying condition from aerial images from floodplain and wetland environments
3. To evaluate the use of aerial imagery and machine learning for landscape scale assessments of condition and extent.