Enriching BIM models with fire safety equipment using keypoint-based symbol detection in escape plans

Abstract

In the context of fire safety inspections, Building Information Modeling (BIM) models enriched with Fire Safety Equipment (FSE) components can be used to complete compliance checks and other analyses. However, BIM models often lack the required FSE information. To address this issue, escape plans are a convenient source of data, as they show the position and type of FSE on floor plans. Therefore, this study proposes an automated method to analyze escape plans and extract FSE component information to enrich existing BIM models. The method employs the deep learning model Keypoint R-CNN for symbol detection. Symbol locations are then translated into physical positions within the BIM model. Through a real-building case study, the method demonstrates promising results. Future research may focus on improving the symbol detection performance and the registration between the BIM models and fire escape plans, as well as utilizing the extracted information for actual fire safety analyses.