3D Indoor Scene Understanding from Different Modalities

报告题目：3D Indoor Scene Understanding from Different Modalities

报告摘要：Holistic 3D scene understanding refers to both object semantics understanding and shape reconstruction. With the advance of 3D scanning techniques, extensive works have been proposed to capture and perceive the 3D surroundings of our living environments from different input modalities, e.g., images, point clouds, depth maps, or even human motions. In this talk, I would have a brief overview of the history in 3D indoor scene understanding, with an introduction of our recent works in 3D semantic scene understanding from different input modalities, i.e., single-view images, depth maps, point clouds and human motions.

讲者简介：聂隐愚，现为慕尼黑工业大学视觉计算组（Visual Computing & AI Group, TUM）博士后，博士毕业于英国计算机动画中心，伯恩茅斯大学。主要研究方向为3D视觉，场景理解，3D形状重建与几何学习。研究成果多次发表于CVPR, ICCV, ECCV, NeurIPS, MICCAI，且担任CVPR/ECCV/TPAMI/TVCG/3DV 相关方向审稿人。

时间：2022年11月13日(周日)上午10:00-11:00

腾讯会议：192154620