Plant structure analysis: some statistical approaches based on graphical hidden Markov models

This presentation will focus on diverse contributions related to statistical analysis of growth, branching and flowering patterns in plants. In a first part, I will present the statistical challenges raised in this context and illustrate how the notions of probabilistic graphical and hidden Markov models and multivariate counts can be relevant. In a second part, I will show how nonparametric Bayesian statistics can be used to tackle problems where the number of variables is a priori unknown and unbounded, as well as their potential benefits in some projects related to LiDAR data analysis. Eventually, I will mention several ongoing or forcasted projects to be conducted with colleagues in AMAP / AGAP / Cirad.