Centre de documentation
scientifique

Automatic training data generation in deep learning-aided semantic segmentation of heritage buildings / Arnadi Murtiyoso in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol V-2-2022 (2022 edition)  

Public [article]  inISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences > vol V-2-2022 (2022 edition) . - pp 317 - 324 Titre : Automatic training data generation in deep learning-aided semantic segmentation of heritage buildings Type de document : Article/Communication Auteurs : Arnadi Murtiyoso, Auteur ; Francesca Matrone, Auteur ; M.C. Martini, Auteur ; Andrea Lingua, Auteur ; Pierre Grussenmeyer, Auteur ; Roberto Pierdicca, Auteur Année de publication : 2022 Article en page(s) : pp 317 - 324 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] apprentissage profond [Termes IGN] données d'entrainement (apprentissage automatique) [Termes IGN] monument historique [Termes IGN] segmentation sémantique [Termes IGN] semis de points Résumé : (auteur) In the geomatics domain the use of deep learning, a subset of machine learning, is becoming more and more widespread. In this context, the 3D semantic segmentation of heritage point clouds presents an interesting and promising approach for modelling automation, in light of the heterogeneous nature of historical building styles and features. However, this heterogeneity also presents an obstacle in terms of generating the training data for use in deep learning, hitherto performed largely manually. The current generally low availability of labelled data also presents a motivation to aid the process of training data generation. In this paper, we propose the use of approaches based on geometric rules to automate to a certain degree this task. One object class will be discussed in this paper, namely the pillars class. Results show that the approach managed to extract pillars with satisfactory quality (98.5% of correctly detected pillars with the proposed algorithm). Tests were also performed to use the outputs in a deep learning segmentation setting, with a favourable outcome in terms of reducing the overall labelling time (−66.5%). Certain particularities were nevertheless observed, which also influence the result of the deep learning segmentation. Numéro de notice : A2022-430 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Article DOI : 10.5194/isprs-annals-V-2-2022-317-2022 Date de publication en ligne : 17/05/2022 En ligne : https://doi.org/10.5194/isprs-annals-V-2-2022-317-2022 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100736 [article]

Knowledge-based data enrichment for HBIM: Exploring high-quality models using the semantic-web / Ramona Quattrini in Journal of Cultural Heritage, vol 28 (November–December 2017)  

Public [article]  inJournal of Cultural Heritage > vol 28 (November–December 2017) Titre : Knowledge-based data enrichment for HBIM: Exploring high-quality models using the semantic-web Type de document : Article/Communication Auteurs : Ramona Quattrini, Auteur ; Roberto Pierdicca, Auteur ; Christian Morbidoni, Auteur Année de publication : 2017 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Termes IGN] flux de travaux [Termes IGN] interopérabilité [Termes IGN] modélisation 3D du bâti BIM [Termes IGN] monument historique [Termes IGN] ontologie [Termes IGN] RDF [Termes IGN] visualisation 3D [Termes IGN] web sémantique [Vedettes matières IGN] Géovisualisation Résumé : (auteur) In the last decade, the paradigm Historical Building Information Modeling (HBIM) was investigated to exploit the possibilities offered by the application of BIM to historical buildings. In the Cultural Heritage domain, the BIM-oriented approach can produce 3D models that are data collector populated by both geometrical and non-geometrical information related to various themes: historical documents, monitoring data, structural information, conservation or restoration state and so on. The realization of a 3D model fully interoperable and rich in its informative content could represent a very important change towards a more efficient management of the historical real estate. The work presented in these pages outlines a novel approach to solve this interoperability issue, by developing and testing a workflow that exploits the advantages of BIM platforms and Semantic-Web technologies, enabling the user to query a repository composed of semantically structured and rich HBIM data. The presented pipeline follows four main steps: (i) the first step consists on modeling an ontology with the main information needs for the domain of interest, providing a data structure that can be leveraged to inform the data-enrichment phase and, later, to meaningfully query the data. (ii) Afterwards, the data enrichment was performed, by creating a set of shared parameters reflecting the properties in our domain ontology. (iii) To structure data in a machine-readable format, a data conversion was needed to represent the domain (ontology) and analyze data of specific buildings respectively; this step is mandatory to reuse the analysis data together with the 3D model, providing the end-user with a querying tool. (iv) As a final step in our workflow, we developed a demonstrative data exploration web application based on the faceted browsing paradigm and allowing to exploit both structured metadata and 3D visualization. This research demonstrates how is possible to represent a huge amount of specialized information models with appropriate LOD and Grade in BIM environment and then guarantee a complete interoperability with IFC/RDF format. Relying on semantically structured data (ontologies) and on the Linked Data stack appears a valid approach for addressing existing information system issues in the CH domain and constitutes a step forward in the management of repositories and web libraries devoted to historical buildings. Numéro de notice : A2017-231 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE Nature : Article DOI : 10.1016/j.culher.2017.05.004 En ligne : https://doi.org/10.1016/j.culher.2017.05.004 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=85144 [article]

Integrating elevation data and multispectral high-resolution images for an improved hybrid Land Use/Land Cover mapping / Mirco Sturari in European journal of remote sensing, vol 50 n° 1 (2017)  

Public [article]  inEuropean journal of remote sensing > vol 50 n° 1 (2017) . - pp 1 - 17 Titre : Integrating elevation data and multispectral high-resolution images for an improved hybrid Land Use/Land Cover mapping Type de document : Article/Communication Auteurs : Mirco Sturari, Auteur ; Emanuele Frontoni, Auteur ; Roberto Pierdicca, Auteur ; Adriano Mancini, Auteur ; Eva Savina Malinverni, Auteur ; Anna Nora Tassetti, Auteur ; Primo Zingaretti, Auteur Année de publication : 2017 Article en page(s) : pp 1 - 17 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications photogrammétriques [Termes IGN] base de données d'occupation du sol [Termes IGN] carte d'occupation du sol [Termes IGN] classification hybride [Termes IGN] données altimétriques [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] image multibande [Termes IGN] intégration de données [Termes IGN] occupation du sol Résumé : (Auteur) The combination of elevation data together with multispectral high-resolution images is a new methodology for obtaining land use/land cover classification. It represents a step forward for both the accuracy and automation of LULC applications and allows users to setup thematic assignments through rules based on feature attributes and human expert interpretation of land usage. The synergy between different types of information means that LiDAR can give new hints at both the segmentation and hybrid classification steps, leading to a joint use of multispectral, spatial and elevation data. The output is a thematic map characterized by a custom-designed legend that is able to discriminate between land cover classes with similar spectral characteristics (level 3 of the CLC legend). Experimental results from a hilly farmland area with some urban structures (Musone river basin, Ancona, Italy) are used to highlight how the proposed methodology enhances land cover classification in heterogeneous environments. Numéro de notice : A2017-043 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.1080/22797254.2017.1274572 En ligne : http://doi.org/10.1080/22797254.2017.1274572 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=84213 [article]