In recent years, the automatic segmentation and classification of digital survey data has been experimented with in built heritage studies. Despite the encouraging progress in the use of Machine and Deep Learning techniques, the semantic segmentation of point clouds is more complex, especially for the historic environment for which, due to the heterogeneity of shapes, it is more difficult to recognize homogeneous regions with similar properties. Given the need to process a large volume of already annotated data for the training and recognition of new scenes, the type and quality of the initial data play a fundamental role in the classification process, as they influence the subdivision into predefined categories that are not always consistent with a decomposition into architectural elements and sub-elements shared by the scientific community. This is an interpretative problem that already emerges from traditional manual labelling, which, being highly subjective, reduces the reproducibility of the results. The paper focuses on understanding to what extent the recognition of homogeneous regions is influenced by factors such as: manual labelling carried out by annotators with different specializations; density value of the point clouds; and type of data acquired depending on the acquisition sensor. These evaluations were conducted by employing the Random Forest algorithm on specific pre-processed point cloud datasets, with reference to the typology of the Franciscan cloister, in order to make the recognition flows more controlled and less ambiguous, aiming at an advancement towards more efficient modelling and management of the existing architectural heritage.
Evaluation of Annotation Ambiguity in Common Supervised Machine Learning Classification Approaches for Cultural Heritage
The paper investigates the impact of annotation ambiguity on supervised machine learning classification processes applied to cultural heritage data. Focusing on semantic segmentation of 3D point clouds, the study highlights how manual annotations—used as training data—are inherently subjective and influenced by the annotator’s background, expertise, and interpretation of architectural elements. Through an experimental framework involving multiple annotators, the research quantifies variability in labeling using metrics such as Intersection over Union (IoU), revealing significant discrepancies in the definition of semantic classes (e.g., columns, arches, moldings). The work proposes integrating uncertainty-aware approaches, such as semantic maps, to encode annotation ambiguity and improve the robustness, interpretability, and interoperability of AI-based classification systems for heritage documentation.
Monitoring Systems Design with Real Time Interactive 3D and Artificial Intelligence
This paper presents a methodological framework for the design and simulation of monitoring systems for cultural heritage, integrating real-time interactive 3D environments and artificial intelligence techniques. The research focuses on preventive conservation, proposing a workflow that combines multiscalar digitization, semantic annotation, and AI-driven analysis to detect and predict the evolution of degradation phenomena.
The methodology is structured into three main phases: (i) detailed digital acquisition of the architectural artifact through integrated survey techniques (TLS, photogrammetry, and image-based methods); (ii) construction of an information-rich 3D model, semantically annotated and enriched with diagnostic data (thermal, material, and colorimetric); and (iii) simulation of monitoring systems within a real-time interactive environment using Unreal Engine 5. As illustrated in the workflow diagram (Fig. 1, p. 722), the system enables the placement of virtual sensors and the generation of synthetic datasets through simulated damage patterns. These data streams are used to train machine learning models for anomaly detection, which are integrated with natural language–based AI systems to support user interaction and decision-making. The results demonstrate the potential of combining digital twins, simulation environments, and AI to design and test monitoring infrastructures prior to on-site implementation, while highlighting the challenges related to data integration, model accuracy, and real-world validation.
Virtual Canova: a Digital Exhibition Across MANN and Hermitage Museums
The paper presents the results of a scientific collaboration between the Interdepartmental Research Center Urban/Eco of the University of Naples Federico II and the MANN (Museo Archeologico Nazionale di Napoli, National Archaeological Museum of Naples).The research activity was aimed to the digitisation, design, and development of an AR/VR-powered narrative experience regarding Antonio Canova’s statuary that is currently exhibited at the MANN, loaned by the Hermitage in St. Petersburg: Cupid, Hebe, Dancer, Cupid and Psyche, the Genius of Death and The Three Graces.The project is motivated by the will to realize an active example of a digital museum, where cultural and formative experiences related to the fruition of architectural and artistic artifacts can be relived over time, even when manufacts are not physically and/or temporally located in the space where the experience takes place.
CHROME Project: Representation and Survey for AI Development
The paper shows the results of the PRIN CHROME Cultural Heritage Orienting Multimodal Experi-ences project, about the three charterhouses of Campania, with a specific focus on research activities related to the connections between representation, survey, AI and VR. The project has formalized a methodology of collection, analysis and modeling of multimodal data, useful for designing virtual agents in 3D environments, which can be applicable in museum environments. The achievement of the goal is pursued through: (i) an integrated range–based acquisition and morphometric data modeling process coherent with VR management, (ii) the use of semantic maps linked with thesauri published as LOD to solve both the theme of ambiguity and annotation uncertainty and the inter-pretability of information by an AI; (iii) the modeling of a virtual agent with the development of a mathematical model for computational control of gestures and prosody.
Semantically Annotated 3D Material Supporting the Design of Natural User Interfaces for Architectural Heritage
With the advent of artificial intelligence and natural user interfaces, the need for multimedia material that can be semantically interpreted in real time becomes critical. In the field of 3D architectural survey, a significant amount of research has been conducted to allow domain experts represent semantic data while keeping spatial references. Such data becomes valuable for natural user interfaces designed to let non-expert users obtain information about architectural heritage. In this paper, we present the architectural data collection and annotation procedure adopted in the Cultural Heritage Orienting Multimodal Experiences (CHROME) project. This procedure aims at providing conversational agents with fast access to fine-detailed semantic data linked to the available 3D models. We will discuss how this will make it possible to support multimodal user interaction and generate cultural heritage presentations.
Segmentation protocols in the digital twins of monumental heritage: a methodological development
The paper shows an advancement of the research that the authors have been carrying out in recent years in semantic structuring of digital architectural representations field, with a focus on the issue of uncertainty of annotations. The studies carried out in this regard have shown how the domain experts specialization determines a vision and interpretation of the same architectural object that we could define “categorized”. The interest was, then, in verifying which categories of experts have a greater degree of agreement in classifying and segmenting architectural elements, to highlight which specializations contribute the most in enriching the semantic reasoning about such forms. Aiming to broaden this reasoning, the research was deepened with annotation sessions concerning architecture examples that didn’t correspond to the classical orders rule but included wider fields of historical heritage (from sacred to fortified architecture). The aim is to verify whether the uncertainty of annotation is actually ascribable to a specific segment of the historical heritage, for example, the classical world, or whether the question is broader and as such in needs deeper thinking.
The Role of the Graphic Element in the Context of Playful Games for Cultural Heritage
This paper aims to investigate the conditioning that the use of playful games requires the role of graphic element for disseminating and promoting the Cultural Heritage. Within the CHROME project, which has, among the various objectives, the definition of an innovative strategy to promote the three Charterhouses of Campania, it comes up with the idea to plan a playful game placed in one of the three monasteries. Its purpose is to provide a first knowledge, both in relation to the spatiality of a Carthusian monastery and to the life of a monk of the Order, exploiting the playful dimension of the games. Since that the proposed location for the game is a monastic complex whose modeling is gained by range-based and image-based survey processes, the project shows the definition of a methodology to generate digital three-dimensional models, whose geometric genesis is at the same time both topologically coherent and enjoyable on the selected technology platform. Once obtained the scene in which the narration develops, it must qualify as a visual device able to activate the sensory involvement, the share and the exploration. For this reason, some expedients (illumination techniques, framing, distortions, sonorous scenes) have been studied to stimulate the player and to communicate cognitive messages related to the game space using the principle “show, don’t tell”.