18 Septembre – Thesis defense - Jessica Sodjo

10 h30 Amphi Jean-Paul Dom - Laboratory IMS (Building A31 / Talence)

Bayesian nonparametric model for joint segmentation of a set of images with shared classes.

We propose a Bayesian nonparametric approach for the joint segmentation of a set of images. Our model combines the hierarchical Dirichlet process that allows the number of classes to not be fixed in advance and the Potts model which favors the homogeneity in the image.

Event localization