Articles by Marianna Figuera

This paper focuses on collaborative methods and open source tools aimed to analyze and query 3D photogrammetric models of ancient architectures. The processing of virtual models led to the constitution of a training dataset of around 1300 wall facing stones from four archaeological sites in Crete. Through a purposely-conceived add-on of the open source software Blender, some algorithms expressed in Python are able to extract archaeologically significant features and to perform processes of Machine Learning and data mining. The resulting data are imported into a dedicated DB managed through a web application based on the open source framework Django. This workflow addresses some peculiar challenges of the application of Artificial Intelligence to archaeological heritage: the lack of training dataset, particularly related to architecture; the lack of best practices for geometry processing and analysis of 3D data; the use of poorly predictive data in semi-automatic processes; the sharing of data into the scientific community; the importance of the open source technology and open data.

The article presents the results of a wider research carried out by a multidisciplinary group (archaeologists and engineers) of the University and the CNR-ISPC of Catania in the South-Western Quarter of the Minoan Palace of Phaistos (Crete). The article focuses on two digital survey campaigns carried out respectively in 2014, laser scanning, and 2019, Structure from Motion. Starting from the point cloud by laser scanner, the most recent, low cost and user-friendly photogrammetric tools (GoPro camera and software Agisoft Metashape) have been used during the 2019 campaign in order to update and to improve the previous dataset, which was used as a grid for georeferencing and scaling the new virtual model. Special attention was addressed to the comparison of the two datasets and to the reuse of the first one for georeferencing and scaling the second one. Furthermore, the research has been focused on the opportunity to exploit the obtained virtual model both for scientific purposes and for the outreach. The lack of accessibility of the South-Western Quarter of Phaistos Palace to the visitors attributes a special interest to this output. The virtual environment thus realized constituted an ideal starting point for the development of an educational fruition project based on a Serious Game approach. The cooperation of archaeologists and engineers in the development of the Phaistos game ensures a gaming experience not only pleasant but also provided with a strong educational profile.

A specific relational database has been created for the management of small finds found on the Cretan sites of Phaistos and Ayia Triada in Crete. Such artifacts have been often underestimated in past archaeological studies, because of the perception that they are less relevant objects in comparison to other categories such as pottery. The database GEAR has been created to improve the recognition of their potential. This article presents the possible solutions used in the database to overcome some methodological problems: specifically two of the methodological aspects that have been dealt with are standardization and management of the reference sources and typological attributions. In the first case, a standard typological terminology (recommended by the ICCD) is used to assure data normalization; in the second case, the preservation of the subjectivity and uncertainty of archaeological attributions is obtained with the application of fuzzy logic and its concept of ‘probability of belonging’.