Research area
VTOL drones with RGB and multispectral cameras and machine learning for tree counting, crop health, plant physiology and fertiliser zoning.
Overview
Our UAV work has moved from pictures to physiology. Trinity Pro VTOL flights with multispectral cameras, matched to measurements on the ground, become maps of tree counts, crop health, stomatal conductance and where fertiliser is needed.
Our approach
Every project starts in the field: we sense what is happening, model what it means, and turn the result into decisions farmers and operators can act on.
Why it matters
What we do
Center projects
2 projects from the center's research portfolio. Open a slide to see the system, the setup and the results.
Counts papaya trees automatically from RGB orthomosaics captured by a Trinity Pro VTOL, by tiling the mosaic and detecting trees with a convolutional neural network.
Rates papaya as normal, half normal or abnormal from VTOL multispectral data combined with field SPAD and LI-COR readings in a neural network.
Student research
2 undergraduate theses, 1 master's thesis supervised at the center, with their methods, key results and research framework.
Stomatal conductance is a sensitive stress indicator but is measured by hand, leaf by leaf. Five-band hybrid VTOL imagery and 37 vegetation indices, matched to field measurements, map it across the field with tuned models validated by five-fold cross-validation and explained with SHAP.
In practice
Related publications
Other research
Partner with our researchers, pilot smartagri on your fields, or join the team building data-driven agriculture.
