Visualization of Building Congestion Analysis and Prediction Using Floor Plans

This study proposes a novel approach for visualizing congestion in architectural spaces using generative AI to enhance design performance. Congestion analysis during early design stages is crucial, impacting spatial quality, functionality, and user experience. Traditional methods are often time-consuming and complex. Our method leverages advanced AI techniques to create more immediate and high-quality congestion maps, aiming to improve design efficiency and stakeholder communication. The research focuses on typical Korean apartments, incorporating BIM-based data preparation, AI model development, and validation against traditional simulations. We developed a domain-specific fine-tuning model to improve visualization accuracy. The findings suggest that generative AI can provide rapid and intuitive congestion analysis, potentially optimizing design decision-making and space utilization. However, further research is necessary to apply these results to diverse building types.