Suckers Emission Detection and Volume Estimation for the Precision Farming of Hazelnut Orchards

In this work, inspired by the needs of the H2020 European Project PANTHEON(1), we address the hazelnut sucker detection and canopy volume estimation problem on a per-plant basis. Sucker control is an essential but challenging practice in agriculture, given the fact that suckers, i.e., shoots that grow from the tree roots, compete with the tree itself for water and nutrients. This research is motivated by the observation that in current best-practice, sucker control is carried out by applying a non-calibrated amount of chemical inputs to each tree. Indeed, a proper sucker detection and estimation algorithm would represent the enabling technology for an environmentally friendly sucker control approach where the amount of herbicide could be properly calibrated according to the needs of each individual plant. In this work, we propose an end-to-end algorithm for detecting the presence of suckers and for estimating their canopy. First a sparse point cloud-based representation of the suckers is detected, then an approximated canopy estimation is achieved by means of a tailored meshing strategy that performs a leaf-based clustering and an iterative clusters connection. The volume is then estimated by the resulting mesh. Preliminary real-world experiments are provided to corroborate the effectiveness of the proposed canopy estimation strategy.