UrbanGenAI: Reconstructing Urban Landscapes using Panoptic Segmentation and Diffusion Models
Project Overview
The document explores the use of UrbanGenAI, a generative AI tool that integrates computer vision with artificial intelligence to facilitate urban landscape reconstruction in educational settings, focusing on its application in the Architectural Design Studio at the University of Derby. This innovative tool empowers students and community members to collaboratively visualize and modify urban designs, enhancing their learning experiences through active participation and creativity. By allowing users to input their ideas, UrbanGenAI fosters an interactive environment that promotes collaboration and deepens understanding of urban dynamics. Preliminary findings suggest that the tool significantly supports participatory planning efforts, indicating its potential to transform education in architecture and urban design by engaging stakeholders in the design process and making urban planning more accessible and inclusive. Overall, the document underscores the transformative impact of generative AI in education, particularly in fostering a collaborative approach to urban design and enhancing students' and community members' educational experiences.
Key Applications
UrbanGenAI
Context: Applied in educational settings such as the Architectural Design Studio module at the University of Derby for urban landscape design education and in community engagement initiatives like the Living Streets research study to involve local residents in co-designing their neighborhoods.
Implementation: Users (students or residents) upload urban landscape images, which are processed using the OneFormer model for segmentation and the SDXL model for image generation based on their textual inputs. This approach allows for visualizing proposed changes and enhances user interaction with the design process.
Outcomes: Enhances understanding of urban design dynamics, increases creativity and engagement among students and community members, empowers residents to actively participate in urban design, facilitates collaborative discussions, and improves visual communication of design ideas.
Challenges: Requires user interaction and understanding of the tools; potential for technical issues in image processing; ensuring all community members have access to the technology and know how to use it effectively.
Implementation Barriers
Technical Barrier
The complexity of image processing and the need for technical knowledge to operate the AI tool.
Proposed Solutions: Simplifying user interfaces and providing training workshops for users.
Project Team
Timo Kapsalis
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Timo Kapsalis
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai