---

The digitization of large quantities of analogue data and the massive production of born-digital documents for many years now provide us with large volumes of varied multimedia data (images, maps, text, video, multisensor data, etc.), an important feature of which is that they are cross-domain. "Cross-domain" reflects the fact that these data may have been acquired in very different conditions: different acquisition systems, times and points of view (e.g. a 1962 postcard from the Arc de Triomphe vs. a recent street-view acquisition by mobile mapping of the same monument). These data represent an extremely rich heritage that can be exploited in a wide variety of fields, from SSH to land use and territorial policies, including smart city, urban planning, tourism, creative media and entertainment. In terms of research in computer science, they address challenging problems related to the diversity and volume of the media across time, the variety of content descriptors (potentially including the time dimension), the veracity of the data, and the different user needs with respect to engaging with this rich material and the extraction of value out of the data. These challenges are reflected in research topics such as multimodal and mixed media search, automatic content analysis, multimedia linking and recommendation, and big data analysis and visualisation, where scientific bottlenecks may be exacerbated by the time dimension, which also provides topics of interest such as multimodal time series analysis.

---

Date: 20 October 2021

Venue: Virtual Event, 14:00-18:00 (GMT+8 Time Zone)

This year, like the previous year, we have the pleasure to award a prize of 500 euros for the best article. The best article will be chosen by a jury, and announced at the end of the workshop. The award is co-sponsored by Friedrich-Alexander-Universität Erlangen, the French Mapping Agency (IGN) and the French National Research Agency (ANR, Alegoria project). The workshop will also feature the following two keynote talks.

The event is fully organized online, but for those who plan to attend the conference and workshops in Chengdu, please note that a onsite room has been assigned to SUMAC'21, where you can attend the workshop and present your article (see the main agenda of the ACM Multimedia conference, it is room Langbo room A).

---

Agenda of the day

(All times in GMT+8 Time Zone, refer TIMEBIE for time zone comparison)

14:00 - 14:15

Welcome & Introductory session by organizers

14:15 - 15:00

Keynote 1 ( Mathieu Aubry ) ( Presentation Slides )

15:00 - 15:20

(#1) Built Year Prediction from Buddha Face with Heterogeneous Labels Yiming Qian (Osaka University); Cheikh Brahim EL VAIGH (Irisa/Inria); Yuta Nakashima (Osaka University); Benjamin Renoust (Osaka University); Hajime Nagahara (Osaka University); Yutaka Fujioka (Osaka University)

15:20 - 15:40

(#2) Evaluation of Deep Learning Techniques for Content Extraction in Spanish Colonial Notary Records Nouf Alrasheed (University of Missouri-Kansas City); Shivika Prasanna (University of Missouri-Columbia); Ryan Rowland (University of Missouri-Kansas City); Praveen Rao (University of Missouri-Columbia); Viviana Grieco (University of Missouri-Kansas City); Martín Wasserman (University of Buenos Aires & CONICET)

15:40 - 16:00

(#3) Searching Silk Fabrics by Images Leveraging on Knowledge Graph and Domain Expert Rules Thomas Schleider (EURECOM); Raphael Troncy (EURECOM); Mareike Dorozynski (Leibniz Universitat Hannover"); Franz Rottensteiner ("Leibniz Universitat Hannover, Germany"); Jorge Sebastián Lozano (Universitat de València); Georgia Lo Cicero (University di Palermo); Thibault Ehrhart (EURECOM)

16:00 - 16:15

BREAK

16:15 - 17:00

Keynote 2 ( Jon Hardeberg ) ( Presentation Slides )

17:00 - 17:20

(#4) How to spatialize geographical iconographic heritage Emile Blettery (IGN); Nelson Fernandes (LASTIG/IGN-UGE); Valérie Gouet-Brunet (LASTIG/IGN-UGE)

17:20 - 17:40

(#5) Software and Content Design of a Browser-based Mobile 4D VR Application to Explore Historical City Architecture Sander Muenster (FSU Jena); Jonas Bruschke (FSU Jena); Ferdinand Maiwald (FSU Jena); Constantin Kleiner (FSU Jena)

17:40 - 18:00

Technical break for best paper selection

18:00 - 18:15

Wrap-up and Ceremony Award

---

---

Keynote 1

Mathieu Aubry ( Presentation Slides )

Senior researcher, Imagine team, LIGM lab, ENPC, Écoledes Ponts ParisTech, France)

Deep Learning for Historical Data Analysis

This presentation will give an overview of projects on leveraging deep learning for historical data analysis my group did in the last 3 years, partly in the context of the ANR EnHerit project. I will discuss in particular how deep learning can be used to establish links between artworks and historical documents: repeated patterns discovery in artwork collections, fine artwork alignment, document images segmentation, historical watermarks recognition, generic clustering, and scientific illustration propagation analysis. In all cases, I will show that standard approaches can give useful baseline results when tuned adequately, but that developing dedicated approaches that take into account the specificity of the data and the problem significantly improves the results.

Keynote 2

Jon Hardeberg ( Presentation Slides )

Professor at the Computer Science Department of the Nor-wegian University of Science and Technology, Norway

Analyzing CHANGE in cultural heritage objects through images

Cultural heritage (CH) objects have been constantly undergoing changes/degradation over time. In order to pass the legacy of these objects to future generations, it is important to monitor, estimate and understand these changes as accurately as possible. These investigations will support the conservators to plan necessary treatments in advance or to slow down the specific deterioration processes. The dynamic characteristics of materials vary from one object to another and are influenced by several factors. To detect and predict their changes, accurate documentation and analysis are necessary. Over the years, CH digitization using scientific imaging techniques has become more widespread and has created a massive amount of datasets of different forms in 2D and 3D. Several past projects focused on different aspects of technological developments for better digitization methods. There has been less focus on the processing and analysis of these datasets to make the greatest use of them and to their further exploration for monitoring ‘changes’ in CH artifacts for conservation purposes. The CHANGE project takes cultural heritage digitization to a new level by exploring digital datasets for deeper analysis and interpretation by developing methodologies for the assessment of changes in CH objects by comparing and combining digital datasets captured at different time periods.

BEST PAPER AWARD - SUMAC 2021

''Searching Silk Fabrics by Images Leveraging on Knowledge Graph and Domain Expert Rules''

T. Schleider (EURECOM), R. Troncy (EURECOM), M. Dorozynski (Leibniz Universitat Hannover), F. Rottensteiner (Leibniz Universitat Hannover, Germany), J.S. Lozano (Universitat de València), G. Lo Cicero (University di Palermo), T. Ehrhart (EURECOM)

The award is accompanied by a money prize equivalent to 500 Euros