The EBU Data Technology Seminar (DTS) is a reference event for operators in the radio and television sector for the in-depth analysis and discussion of technologies related to Artificial Intelligence and data analysis.

It took place in Geneva from 12 to 14 March 2024, and its centrality is confirmed by the participation (more than 130 participants in person, more than a hundred online).

The event program is available to everyone on the website https://tech.ebu.ch/events/dts2024, with presentations ranging from content verification to ontologies used in production, and addressing issues such as model transparency and use of generative artificial intelligence to enrich content.

This year too, Rai CRITS contributed to the success of the event, through participation in Paolo Casagranda’s Program Committee, four presentations, a demo station, as well as the moderation of four sessions by Paolo Casagranda, Maurizio Montagnuolo, and Lorenzo Channel.

Moderators of EBU DTS 2024

• Maurizio Montagnuolo presented “Advancing Media Intelligence: The Face Management Framework (FMF) for automated annotation and diversity analysis for Public Service Media”, describing a system to manage the automatic annotation of television personalities, thanks to the use of public databases.
• Fake news was the topic of Lorenzo Canale‘s interactive presentation “Unveiling AI Solutions for Disinformation Combat: Insights from the IDMO Project”.
• The specialization of Large Language Models was addressed by Stefano Scotta in “Adapt LLM to specific tasks: a convenient alternative to the “GPT-for-all” approach?”, which indicated how limited specialized models can obtain more useful results in tests specific compared to generic ones.
• Roberto Iacoviello followed the technological demonstration based on the reconstruction of 3D models of faces and objects starting from 2D images, part of the H2020 XReco project.

EBU Data Technology Seminar, 2024