The Augmented Artwork Analysis project is organizing a symposium, “Interrogating the visible: images that respond to each other”, aimed at bringing together some of the leading researchers involved in augmented reality and artificial intelligence projects dedicated to the enhancement, analysis and interpretation of works of art as museum heritage.
Without the support of digital technology, we are faced with the difficulty of working on comparatively large corpora of images, and therefore of annotating and cataloguing more or less recurrent features and motifs that have undergone various evolutions. At the same time, the persistence, at least partial, of past theories on a programmatic, or even intuitive, level should not now lead to a “devolution” of the historian’s or semiotician’s work to machines and software. If computer-assisted analysis has been much discussed, it’s because automatic data processing necessarily remains one channel within an interpretative practice that goes beyond machine horizons and embraces additional critical means to achieve these goals.
Basically, we could reverse the usual rhetorical expression to emphasize the complementary dimension of “interpreter-assisted computing”. This “chiasmatic” version of mutual assistance between machine and interpreter may be a good way of correcting some of the errors of the past, but it may not be able to integrate recent advances in available technologies. When we speak of deep learning, we are referring to machine learning structured on architectures with various levels of representation; these levels are hierarchical and can take over from one another, thus supporting non-linear and recursive data processing. Most computations could be reduced to information extraction and pattern recognition, but as soon as deep learning is assumed as a prosthesis by a researcher, it seems to build a new perspective on the phenomena investigated and, although not neutral, it brings knowledge and therefore a significance to the results that goes beyond simple recognition.
The colloquium will be organized into four thematic half-days, with specialist speakers in each field: (i) epistemological approaches: seeing with the machine; (ii) computational approaches: teaching the machine to see; (iii) professional and pedagogical approaches: teaching museum visitors to see; (iv) “augmented” approaches: seeing artistic heritage anew and differently. The last two themes will also be developed through round-table discussions aimed at shifting the focus from the academic to the professional perspective, so as to take into account knowledge from the field as well as museum practices, particularly as regards their collaborations with companies involved in the implementation of augmented reality devices. In view of the presence of three doctoral students funded by the ANR and FNR, with grants supporting the AAA project, the colloquium aims to set aside time for them to present their research in action to the scientific community. Doctoral students will also be asked to act as discussants for the presentations closest to the subjects of their theses. Other doctoral students working in the same field will also be invited to participate. In addition, the colloquium aims to follow up on the work carried out at the Winter School (Rome, Academia Belgica/Villa Medici, January 15-19, 2024) and to extend the discussions.