Why it matters
Myrtle rust (Austropuccinia psidii) was detected in Australia on the NSW central coast, 2010. Since then, the pathogen has spread rapidly in the mainland and recently, New Zealand. It has been established along the east coast of Australia with recorded impacts across a range of ecosystems. As of April 2016, about 350 native species have proved susceptible to myrtle rust, with estimated yield loses up to 70%. Their primary effects on local hosts include deformation of leaves, heavy defoliation of branches, reduction of fertility, stunted growth, and plant death. Federal, state and local government bodies have collaborated with industry organisations and owners of affected properties to suppress the outbreak and eradicate the disease, with regular results.
By integrating the best outcomes of site-based tests and UASs, this project aims the desing of a novel framework to detect alterations in plantation and vegetation forests caused by exotic pathogens. The designed pipeline process and algorithms are able to correlate successfully the labelling of healthy and infected plants in the study areas. Also, it is capable of process data from multiple georeferenced image sensors, filter and work with the most relevant image features and optimise the computation effort to predict vegetation conditions.
Aims and Objectives
This project aims to provide end-users with the value of these technologies in guiding decisions and adopting systems based on capabilities to detect exotic pathogens on host plantation and natural forests. The project focuses on the application of UAVs, hyperspectral cameras, machine learning and on-ground testing data to offer an integrated system that informs stakeholders the actual conditions and locations of affected plants. Specific and measurable benefits include substantial reductions in data acquisition times, assessment tools for Myrtaceae vegetation conditions in a broader scale (e.g. integrating insights from ground-based and remote sensing techniques), practical use of resources (personnel, equipment and budget), and potential minimisations in economic losses and preservation of forest ecosystems.
Milestones & Achievements
- Evaluate the utility of multi- and hyperspectral, thermal, LiDAR and high-resolution RGB sensors and cutting-edge UAVs in identifying exotic pathogens such as myrtle rust on the Myrtaceae family plants from early to mature stages of the disease.
- Evaluate the scope of current machine learning algorithms on the processing of large datasets of various image sensors for the detection and mapping of deteriorated plants accurately and efficiently.
- Synthesise the provided outcomes of site-based and UAS-based approaches into a single system to demonstrate a practical application for surveillance of deteriorated Myrtaceae forests by myrtle rust in local and national monitoring campaigns.
- Development of a pipeline to process data from multiple sensors, and detect and map Myrtaceae plants affected with myrtle rust through a classification algorithm using machine learning, hyperspectral imagery and a multi-rotor UAS.
- Integration of multiple sensors on a multi-rotor UAS such as high-resolution RGB, thermal, LiDAR, multispectral and hyperspectral cameras.
- Design of a customised gimbal to maximise the quality of acquired data of hyperspectral sensors for orthorectification and stitching techniques, noise reduction and distortion attenuation.
- Sandino, J.; Pegg, G.; Gonzalez, F.; Smith, G. A Novel Approach to Detect and Map Natural and Plantation Forests Affected by Pathogens using UAVs, Hyperspectral Sensors and Artificial Intelligence. Sensors 2018. (Submitted: under peer-review)
- Vanegas F.; Pegg G.; Kok J.; Sandino J.; Gonzalez F. UAS Remote Sensing Efforts for Myrtle Rust Management. In Science Protecting Plant Health 2017, Brisbane, Australia, 2017. Abstract link: (http://apps-2017.p.yrd.currinda.com/days/2017-09-26/abstract/3798#)
For further information regarding this project please contact the project contacts below:
Funding / Grants
- Plant Biosecurity CRC 2135
- Juan Sandino
- Dr Fernando Vanegas
Other Team Members
- Dr Geoff Pegg (DAF Queensland)
- Dr Grant Smith (Plant and Food Research, New Zealand)