Towards democratizing generation of 3D experiences

Ren, Zhongzheng

Permalink

https://hdl.handle.net/2142/122204 Copy

Description

Title

Towards democratizing generation of 3D experiences

Author(s)

Ren, Zhongzheng

Issue Date

2023-11-27

Director of Research (if dissertation) or Advisor (if thesis)

Schwing, Alexander Gerhard

Doctoral Committee Chair(s)

Schwing, Alexander Gerhard

Committee Member(s)

• Forsyth, David
• Lazebnik, Svetlana
• Hoiem, Derek
• Agarwala, Aseem

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Computer vision
• 3D

Abstract

3D experiences offer a significantly more immersive platform for learning and entertainment when compared to 2D media, enabling a deeper understanding of complex real-world processes and fostering the creation of lasting memories. However, the conventional approach to generating such experiences is both labor-intensive and cost-prohibitive. Additionally, state-of-the-art deep learning solutions often encounter challenges in scaling due to the limited availability of large-scale 3D supervision. This limitation arises from the fact that while real-world 3D data is abundant, it is often sparsely annotated. In light of these challenges, this thesis aims to democratize the process of generating 3D experiences by exploring the following key research objectives: 1) Developing an innovative generation pipeline starting from recognizing 3D objects efficiently without relying on 3D labels. 2) Creating a class-agnostic approach for reconstructing dynamic objects from unstructured video data. 3) Investigating methods for manipulating and selecting 3D assets based solely on 2D user scribbles. Through these research efforts, this thesis seeks to pave the way for more accessible and cost-effective 3D content creation, with broader applications in fields such as science, entertainment, and beyond.

Graduation Semester

2023-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/122204

Copyright and License Information

Copyright 2023 Zhongzheng Ren

Owning Collections