Towards Robust Shape Correspondence: Learning with Receptive Field Optimization

发布时间:2023年07月06日 作者:刘圣军   阅读次数:[]

报告题目:Towards Robust Shape Correspondence: Learning with Receptive Field Optimization

报告人: Lei Li (Technical University of Munich)

报告时间: 2023年7月9日(星期日)下午14:30-17:30

报告地点:线下:bat365官网登录245智慧教室;线上:Zoom会议(879 0237 3801, password: GGCV2023)

Abstract: Computing accurate correspondence among 3D shapes is a fundamental and challenging task in the fields of Computer Vision and Graphics. In this talk, I will present our recent works on robust geometric deep learning for both 3D rigid and non-rigid shape matching. Our research highlights the importance of receptive field optimization in neural network designs for handling 3D data with significant geometric variations. I will discuss both spatial and spectral methods incorporating learnable receptive fields, which enable end-to-end informative geometric representation learning and ultimately lead to state-of-the-art matching results.

报告人简介:

Lei Li is a postdoctoral researcher working with Prof. Angela Dai at the 3D AI Lab, Technical University of Munich, Germany. Previously, he worked as a postdoctoral researcher with Prof. Maks Ovsjanikov from 2020 to 2022 at LIX, Ecole Polytechnique / Inria, France. Lei earned his Ph.D.

degree in computer science and engineering (2020) from The Hong Kong University of Science and Technology. His Ph.D. thesis advisor is Prof. Chiew-Lan Tai. Lei’s research interests lie in Computer Graphics and Computer Vision, with a focus on geometric deep learning for shape analysis.



打印】【收藏】 【关闭