Neural network generation of 3D models for replenishment of BIM catalogs of medical equipment
Abstract
Neural network generation of 3D models for replenishment of BIM catalogs of medical equipment
Incoming article date: 13.05.2025The design of medical facilities requires careful coordination between different disciplines, especially when developing the section ""Technological Solutions"" (TS), which describes functional processes, equipment layout and compliance with sanitary standards. A common problem in this process is the lack of ready-made 3D families for specialized medical equipment in BIM systems such as Autodesk Revit. The purpose of this study is to develop an algorithm for creating equipment families based on a single image, which will eliminate time and resource constraints during manual modeling. To achieve this goal, the study examined modern design methods for TS and MEP sections, analyzed existing tools for creating families, and tested five neural network-based web applications capable of generating 3D models from 2D images. The proposed algorithm includes creating a 3D model based on images using artificial intelligence web tools, preprocessing the model in Autodesk 3ds Max, and importing the final geometry into a Revit family template. The algorithm was tested on a real example, an anaesthetic gas supply system, demonstrating the feasibility and effectiveness of the approach. Hyper 3D Rodin has been rated as the most effective tool for generation. The study concludes that integrating artificial intelligence into BIM workflows can significantly streamline the development of unique equipment families, reducing manual labor and improving project coordination.
Keywords: Revit family of equipment, medical equipment, neural network generation, 3D geometry of equipment, BIM, technological solutions