ArchiNervi. Two historical funds, one digital archive

The Project

ArchiNervi. Two historical funds, one digital archive is a project by CSAC of the University of Parma and the MAXXI Foundation, realised thanks to the financing of the MIC 2021 Culture Fund, two institutions that share the mission of collecting and valorising their respective collections of contemporary visual arts and design and architecture funds.

Both institutions have acquired for their collections a portion of the archive produced by architect Pier Luigi Nervi since 1920, with a commitment to promoting research and knowledge activities. In this framework, the project aims at enhancing the value of this heritage made up of the two strongly complementary funds, by creating ArchiNervi, a new unified digital archive, which does not exist to date, and a portal through which to provide stories of architecture, building sites, techniques and in-depth studies on issues related to the role of the project concerning the preservation of the territory and architecture. The project is intended as a prototype in which different actions (cataloguing and digitisation, communication, publishing, innovative ways of searching for images) are conducted to produce knowledge and test ways of collaboration between institutions.

Actions

The digital archive
The actions carried out jointly for the realisation of ArchiNervi were:

• the definition of guidelines for the creation of the joint database;
• the revision and standardisation of the respective inventories and cataloguing of the two funds;
• the definition of criteria for the digitisation campaign;
• the identification of a sample of projects to be catalogued and digitised analytically based on shared criteria;
• the definition of the archive architecture and cataloguing levels;
• the definition of a maintenance and restoration campaign for graphic and design materials following the digitisation plan;

Access to the digital archive: the portal and the Lodovico Media Library
Access to this digital archive is through a portal that provides a thorough mapping of the projects realised by the studio, enriched by a storytelling programme articulated in Stories of Architecture and Construction Sites and in in-depth studies on the many themes that this archive solicits. A further channel of access to the cataloguing is possible through the Lodovico Media Library, in which the cataloguing carried out in the course of the project will converge, allowing queries within a broader ecosystem of data and images, thus opening up to a public that is not necessarily specialised.

Communication
To valorise this new resource, a communication campaign was set up jointly by the two institutions, and several videos were created to tell the story of the project through social media channels.A book on the project at the end of the work will present ArchiNervi and the results of this joint work on the two historical funds. The book is the first summary of this undertaking by the various specialists involved to highlight the multiplicity of skills brought together.

Advanced search via images
A further achievement, of an experimental nature, is the realisation of a proof of concept concerning the use of Computer Vision and Deep Learning techniques for advanced similarity-based image search applied to design documentation. The problem to be solved with innovative AI techniques is that of searching for images in a large historical archive of architecture and engineering by speeding up as much as possible what is now a manual process. The AI algorithm was tested separately on photographs, drawings, and in cross-domain mode. To improve and refine the search, it was decided to divide the images into several subsections to have a larger search space. The model used an ad-hoc modified DINO Vision Transformer, extracts descriptors for each of these sub-images (plus the whole image), and saves these descriptors to disk. This way they are ready for all future research. The Vision Transformer is a powerful model used successfully in various fields including medicine, autonomous driving, security, and in the humanities, document analysis, and research.

A graphical user interface has been created to allow the user to crop images to search within the archive.