Vous êtes invités à venir assister à la présentation de Yandi Yang, ce vendredi 25 octobre de 11h30 à 12h30 à la salle CSL-2306.

Yandi termine un stage Mitacs Globalink sous la supervision de Éric Guilbert. Il nous présentera les résultats de ses travaux.

Airborne LiDAR Semantic Segmentation based on Superpoint Graph

In these recent years, several deep learning approaches for point cloud classification and segmentation have been presented. These methods all rely on a convolutional neural network and were developed for dealing with 3D objects and indoor scenes but also for point clouds obtained from terrestrial laser scanning. However, little consideration was given to airborne lidar scanning. Hence, in this presentation, we assess superpoint graph, a popular segmentation approach, on an airborne point cloud. After introducing superpoint graph, we will present results obtained on a point cloud in an urban environment and on a point cloud in a forest environment. We will also address technical issues we encountered during the project.