Graph Neural Networks for Point Cloud Processing

3D Point clouds are a rich source of information that enjoy growing popularity in the vision community. However processing unordered and sparse point clouds using neural networks has been a challenge but in recent years there are models proposed to learn point clouds using deep learning. Graph neural networks have also shown great capacity to capture geometrical features from point clouds in tasks such as classification and segmentation. In this presentation we discuss how graphs can be utilized to describe point cloud patches, detect salient points and use them in downstream tasks such as 3D registration.

Talk is based on the speaker's paper:

Graphite: GRAPH-Induced feaTure Extraction for Point Cloud Registration
https://arxiv.org/abs/2010.09079

Presenter BIO:
Mahdi Saleh studied Bachelor of Electrical Engineering at IUST. He then moved to Germany to study for his Master's at the Technical University of Munich Computer Science department. Meanwhile, he was working as a researcher in 3D computer vision and AI in Framos GmbH, IBM Watson Munich, and AR Experts. Before starting his Ph.D. at TUM, he worked on industrial computer vision and applied research for two years. He is now a Ph.D. student at the CV group of the CAMP chair at TUM focused on Point cloud processing and 3D pose estimation. At the moment, he is also a research Scientist intern at Facebook Reality Lab, Menlo Park.