A Submodular Approach to Find Interpretable Directions in Text-to-Image Models

Text-to-image models have significantly improved the field of image editing. However, finding attributes that the model can actually edit is still a remaining challenge. This thesis proposes a solution to this problem by leveraging a multimodal vision-language model (MMVLM) to find a list of potential attributes that can be used to edit an image, using Flux and ControlNet to generate edits using those keywords, and then applying a submodular ranking method to find which edits actually work. The experiments in this paper demonstrate the robustness of this approach and its ability to produce high-quality edits across various domains, such as dresses and living rooms.

Name thesis:degree_name

Master of Science

Level thesis:degree_level

masters

Discipline thesis:degree_discipline

Computer Science & Applications

Department dc:contributor.department

Computer Science and#38; Applications

Grantor dc:publisher

Virginia Tech

Year dc:date.issued

2025

Author dc:creator

• Allada, Ritika

Chair dc:contributor.committeechair

• Yanardag Delul, Pinar

Committee members dc:contributor.committeemember

• Eldardiry, Hoda Mohamed
• Thomas, Christopher Lee
• North, Christopher L.

• Diffusion Models
• Interpretability
• Image Editing
• Explainable AI
• Recommendation Systems

Statement dc:rights

• Creative Commons Attribution 4.0 International

Licence dc:rights.uri

• http://creativecommons.org/licenses/by/4.0/

Language dc:language.iso

en

Dc Identifier Other

vt_gsexam:44224

OAI identifier oai:identifier

oai:vtechworks.lib.vt.edu:10919/135475