The Integration of Artificial Intelligence in Architectural Surveying and Cultural Heritage: Developments, Methodologies, and Perspectives of the Digital Twin
TL;DR
This paper comprehensively analyzes the latest innovations in architectural surveying, semantic segmentation of point clouds, and Heritage Building Information Modeling (HBIM) and outlines prospects for a holistic, transparent, and technologically advanced approach to protecting built heritage.
Abstract
The digital evolution in construction engineering, architecture, and heritage conservation has accelerated at an unprecedented pace, thanks to the integration of Artificial Intelligence (AI), advanced acquisition systems, and the Digital Twin paradigm. This paper comprehensively analyzes the latest innovations in architectural surveying, semantic segmentation of point clouds, and Heritage Building Information Modeling (HBIM) [1,2]. The analysis unfolds through the classification of Machine Learning and Deep Learning algorithms for the automatic recognition of architectural elements, exploring the effectiveness of neural architectures such as PointNet++ and DGCNN, and their ability to generalize on heterogeneous datasets. The transition from descriptive to predictive and prescriptive models is also examined, with an in-depth focus on case studies of global relevance, including the In-Heritage platform applied to the Palazzi dei Rolli in Genoa for IoT microclimate monitoring, and the Aïoli platform used for the post-disaster reconstruction and virtual anastylosis of Notre-Dame de Paris. Through a critical evaluation of the ethical implications, the interpretability of algorithms (overcoming the "black-box" paradigm), and the integration of immersive reality (VR/AR) for spatial use, the research synthesizes the state of the art. It outlines prospects for a holistic, transparent, and technologically advanced approach to protecting built heritage [3].