February 28, 2023 Tuesday
Geometry processing algorithms are crucial tools to reconstruct and understand the 3D environment. However, the 3D data processing and understanding pipeline is subject to all kinds of disturbances, such as irregularity, sparsity and device noise. For 3D reconstruction tasks, we designed optimization algorithms to handle various underlying noises. Moving one step further, we use data-driven approaches to remove the burden of guessing the correct underlying noise distribution. Recently, we propose to use robust geometry processing techniques to detect smoothly moving objects from unlabeled point cloud sequences without using 3D annotations.
Dr. Xiangru HUANG
Post-doc, MIT CSAILPhD degree, University of Texas at Austin
Dr. Huang received his bachelor degree from Shanghai JiaoTong University. Later, he received his PhD degree from the University of Texas at Austin supervised by Prof. Qixing Huang. He then joined MIT CSAIL working as a post-doc working with Prof. Justin Solomon. His research focuses on optimization algorithms, geometry processing and 3D vision.