Developing cutting-edge tools for 3D reconstruction of plant architecture
Supervisors: Matt Adcock (CSIRO), Mingze Xi (CSIRO) and Tim Brown (ANU/APPF).
In digital agriculture we now have the ability to scan plants in the lab and field at millimetre resolution. However, converting these measurements into meshed 3D plant models with real plant architecture is still a largely unsolved problem. Techniques such as NeRF, deformable models, GANs and pose estimation linked with AI and ML are being used very effectively to solve similar problems in other domains, but these techniques are not yet widely implemented in plant phenotyping. This project will extend and adapt these techniques to support development of cutting-edge 3D reconstruction and plant phenotyping tools using lab and/or field acquired data. A wide range of potential approaches are possible depending on the interests and expertise of the applicant.