Problem Statement

Recent advances in vision-language models (VLMs) and generative AI have opened new possibilities for human-centered image editing. Traditional editing tools often require expert knowledge or laborious manual adjustments, leaving a gap for more intuitive and accessible solutions. This thesis investigates how intelligent interfaces can enhance image editing by automatically analyzing photo content and suggesting meaningful edits. These suggestions will be presented as dynamic, interactive UI widgets that adapt to the context of the image and the user’s editing intent. By combining VLM-driven scene understanding with adaptive interface generation, this project aims to support a human-in-the-loop workflow that balances AI guidance with user control. The outcome will contribute design and technical insights to the growing field of AI-assisted creativity.

Tasks

• Perform literature review of related work on AI-driven image editing, vision-language understanding, and adaptive user interfaces to establish a conceptual and technical foundation.
• Design and prototype an interactive system that combines image analysis with dynamic user interface generation for guided photo editing.
• Explore and implement strategies for presenting AI-suggested edits through intuitive and context-aware UI components.
• Evaluate the prototype in terms of usability, user control, and perceived assistance, using both qualitative and quantitative methods.
• Document findings and insights in a Master’s thesis and present them in the Disputationsseminar.
• (Optional) Contribute to a research publication.

Skills Required

• Strong programming skills in Python for integrating AI pipelines.
• Experience with web development using React (frontend) and Flask or FastAPI (backend).
• Familiarity with machine learning libraries (e.g. PyTorch, HuggingFace Transformers); prior exposure to vision-language models is a plus.
• Interest in UX design, interactive prototyping, and human-AI interaction.
• Experience working with REST APIs and deploying AI models in web contexts.

References

• Shen et al. (2024) - Empowering Visual Creativity: A Vision-Language Assistant to Image Editing Recommendations
• Liu et al. (2024) - MagicQuill: An Intelligent Interactive Image Editing System