Cad2graph: Automated Extraction of Spatial Graphs from Architectural Drawings

Pratik Maitra, Masahiro Kiji, Talal Riaz, Philip M. Polgreen, Alberto M. Segre, Sriram V. Pemmaraju, Bijaya Adhikari

Abstract: A significant obstacle to spatial epidemiology in healthcare facilities is the absence of computationally amenable maps of the underlying space. Spatial data for built spaces are typically stored in computer aided design (CAD) architectural files which are difficult to parse, query, and combine with other data sources. To alleviate this difficulty, we design a tool, cad2graph, which automatically extracts spatial maps from CAD files. To ensure that the spatial map is easily amenable to computation, we represent it as a graph whose vertices represent spatial units of a uniform size and whose edges represent obstacle-free, walkable paths of uniform length connecting adjacent pairs of spatial units. cad2graph extracts key information such as walls, doors, and room labels from the CAD file and through a series of geometric transformations, extracts a spatial graph.

External IDs:dblp:conf/pkdd/MaitraKRPSPA23