CKSRI Seminar Series 2024 “Large Reconstruction Model for Efficient 3D Reconstruction and Generation"
ABSTRACT:
This seminar has invited Dr. Yinghao Xu, postdoctoral fellow at the Stanford Computational Imaging Lab, Stanford University to discuss advancements in large-scale 3D reconstruction through Gaussian Splatting (GRM).
Over the past year, the large language model has achieved significant milestones, approaching human-like intelligence in various domains. However, the literature lacks substantial exploration into large-scale 3D reconstruction. This talk will focus on our recent breakthroughs in this area, specifically highlighting our advancements in large-scale 3D reconstruction through Gaussian Splatting (GRM).
GRM represents a robust and highly adaptable 3D reconstruction system, leveraging extensive, high-quality 3D datasets. Functioning as a large-scale reconstructor, GRM can reconstruct a 3D asset from sparse-view images in approximately 0.1 seconds. Furthermore, GRM demonstrates promising capabilities in generative tasks, such as text-to-3D and image-to-3D, through integration with existing multi-view diffusion models.
SPEAKER(S): Dr. Yinghao Xu is a postdoc at the Stanford Computational Imaging Lab, Stanford University, working with Prof. Gordon Wetzstein. He was a Ph.D. student at the Multimedia Lab (MMLab) at The Chinese University of Hong Kong. He is deeply interested in problems lying at the intersection of Computer Graphics and Computer Vision. Currently, his research focuses on generative models and neural rendering, particularly in the area of 3D generative models.