Virtual try-on and stylization for full-body human images

Cui, Aiyu

Permalink

https://hdl.handle.net/2142/127194 Copy

Description

Title

Virtual try-on and stylization for full-body human images

Author(s)

Cui, Aiyu

Issue Date

2024-11-18

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

Lazebnik, Svetlana

Doctoral Committee Chair(s)

Lazebnik, Svetlana

Committee Member(s)

• Forsyth, David
• Hoiem, Derek
• Liu, Yanxi

Department of Study

Siebel School Comp & Data Sci

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Computer Vision
• Virtual Try-On
• Image Generation

Abstract

Human image generation has seen significant advancements in recent years, enabling the creation of highly realistic human images. This thesis delves into various aspects of human image generation, focusing on virtual try-on and related tasks. By leveraging cutting-edge techniques in image generation, we explore innovative virtual try-on approaches to manipulate and synthesize human images, opening up new possibilities for creative expression, virtual fashion, and personalized experiences. In this dissertation, we first introduce a multi-task human generation pipeline that supports not only virtual try-on but also pose transfer and fashion editing operations. Second, we investigate one-shot human stylization to apply artistic styles to virtual try-on images, creating unique and visually appealing results. Third, we investigate robust warping techniques for virtual try-on that can address the challenging garment warping problems in the real-world virtual try-on. Finally, we extend virtual try-on to in-the-wild scenarios, enabling users to visualize clothing on themselves using real-world photos. Throughout this thesis, two key technical themes emerge: (1) learning human generation tasks without direct supervision and (2) expanding the capabilities and use cases of virtual try-on. These advancements can potentially revolutionize various industries, including fashion and entertainment.

Graduation Semester

2024-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/127194

Copyright and License Information

Copyright 2024 Aiyu Cui

Owning Collections