Graph-based Learning for Tree Morphologies

Graph-based Learning for Tree Morphologies

Given 3D shapes of individual trees, the goal of the project is to use state-of-the-art graph neural networks to learn a low-level representation of the morphologies of different tree species with the help of self-supervised learning. The learned representation will then be used in a subsequent clustering step to classify the species and potentially other attributes of the trees.


  • Good mathematical understanding (in particular statistics and linear algebra)
  • Python programming (including Pytorch)
  • Experience in machine learning


Marissa Weis

Neural Data Science Group
Institute of Computer Science
University of Goettingen