Bionote
Olivia Zumsteg is a doctoral student in the Crop Science group at ETH Zurich. She uses computer vision to extract traits form images taken by smartphones or by the FIP (Field Phenotyping Platform) in-field, and makes predictions based on markers to identify resilient genotypes. Modelling strategies such as ordinary differential are able to integrate environmental variables and are used in her work to model growth and to find genotypic differences regarding climate change.
Presentation Abstract
Frequent drought and heatwaves due to climate change pose significant threats to global wheat yields. The spike volume is an important yield component that shows potential for optimization but cannot be measured at a high frequency so far. Volume estimation from RGB images was first tested on the close-up smartphone images. Results indicate that neural networks showed the highest prediction accuracy. We are currently applying the developed models to predict in-field volumes directly or indirectly via 3D point clouds with our new multi-view sensor head containing a rigid setup of 13 RGB cameras that capture top-view images from hundreds of different genotypes. Such insights are crucial to mitigate the impacts of heat and drought on yield under projected climate conditions.
Co-Researcher: Achim Walter, Nico Graf, Aaron Häuser, Lukas Roth, Andreas Hund