Advanced Representations for 3D/4D Scene Reconstruction and Understanding

PhD Qualifying Examination


Title: "Advanced Representations for 3D/4D Scene Reconstruction and 
Understanding"

by

Mr. Xinhang LIU


Abstract:

NeRF has achieved unprecedented results by utilizing MLPs to representing 
continuous fields, especially in the task of novel view synthesis. Three years 
after its emergence, representations based on 3D Gaussians have attained 
improvements in both quality and efficiency in comparison to NeRF.

This presentation will introduce a series of works based on NeRF and 3DGS from 
recent years. In addition to the original design intentions of NeRF/3DGS and 
the novel view synthesis task, it will also cover topics such as the modeling 
of dynamic (4D) scenes, modeling under adverse conditions, and scene 
understanding.


Date:                   Friday, 10 May 2024

Time:                   11:00am - 1:00pm

Venue:                  Room 5508
                        Lifts 25/26

Committee Members:      Prof. Chi-Keung Tang (Supervisor)
                        Prof. Chiew-Lan Tai (Chairperson)
                        Prof. Pedro Sander
                        Dr. Dan Xu
                        Prof. Dit-Yan Yeung