H2020 AI4Media

H2020 AI4Media “A European Excellence Centre for Media, Society and Democracy”, owner CERTH, Greece, partner University “Politehnica” of Bucharest (led by AI Multimedia Lab), funded by H2020 ICT-48-2020 / Towards a vibrant European network of AI excellence centres, (budget ~12M Eur, 2020-2024).

Strategic Relevance and Scientific Premise

AI4Media is a 4-year-long project funded under the European Union’s Horizon 2020 research and innovation programme, the project aspires to become a Centre of Excellence engaging a wide network of researchers across Europe and beyond, focusing on delivering the next generation of core AI advances and training to serve the Media sector, while ensuring that the European values of ethical and trustworthy AI are embedded in future AI deployments.

The AI4Media consortium is composed of 30 leading partners in the areas of AI and media (9 universities, 9 research centres,12 industrial partners) and 35 associate members. This consortium will establish the networking infrastructure that will unite the currently fragmented European AI landscape in the field of media and will foster deeper and longer-running interactions between academia and industry, including Digital Innovation Hubs. It will also shape a research agenda for media AI research, and pursue research and innovation on cutting-edge technologies at the core of AI research and within specific fields of media-related AI. AI4Media will provide a targeted funding framework through open calls so as to speed up the uptake of innovations developed within the network.

Furthermore, the project envisages a close collaboration with AI4EU platform in the form of integrating the project’s outputs such as modules, services and algorithms into the AI4EU platform, as well as, establish interactions with the AI4EU network. Moreover, AI4Media will also seek to maximize interactions with relevant Digital Innovation Hubs (DIH) across Europe and other national initiatives.

A PhD programme will further enhance links to the industry and will foster the exchange of talent, while working to prevent brain drain. A set of use cases will be developed by the network to demonstrate the impact of the advances in the media sector. The Excellence Centre that will be established during the AI4Media project and the ecosystem that will grow around it will provide a long-term basis for the support of AI excellence in Europe long after the project ends with the aim of ensuring that Ethical AI guided by European values assumes a global leading role in the field of Media.

UPB is a full partner in the AI4Media consortium through its AI Multimedia Lab (part of the CAMPUS Research Institute). The lab brings over 15 years of experience in multimedia video/image processing, content-based retrieval, benchmarking, and related AI technologies. It has authored more than 180 scientific publications, led and co-organized major evaluation campaigns (such as ImageCLEF, MediaEval, and ChaLearn), and contributed to over 24 Romanian and EU-funded research projects as principal investigator or collaborator.

Scope of AI4Media

Besides its economic value, the Media is a crucial sector with a great power that can shape societal values, debates, opinions and could nurture a democratic society or, if things go wrong, lead to polarisation and crises. Europe has a unique, human-centric and trustworthy ethical brand of Artificial Intelligence (AI) to offer to the world, and a field where such AI is needed and can offer a strong advantage to European actors is the Media.

AI technologies hold the promise of disrupting the media industry through:

• advances in content synthesis, analysis, and distribution, and offering new deeper insights into the complex and rapidly evolving social processes that unfold online and offline through the capabilities for large-scale sensing of citizen activities, interests and opinions.
• helping to redefine and radically improve the democratic role of the media by enabling new ways of getting informed, deliberation, political participation and decision making.
• supporting to reshape the relationship between the media and its audiences, making it more responsive, and better reflect the needs of citizens in a data field society.
• creating new exciting opportunities for the realisation of public values, such as media diversity, freedom of expression, inclusiveness, a flourishing public sphere also at the local level.
• finding new ways to inform and engage with the audience that will allow the media to compete in the digital marketplace of ideas with new players, most notably platforms.

AI4Media Junior Fellows Exchange Program

The AI4Media Junior Fellows Exchange Program supports the mobility of early-career researchers seeking to enhance their expertise in artificial intelligence for media and society. The program enables participants to collaborate with leading European AI researchers and media organizations, conducting innovative, industry-relevant research.

The University Politehnica of Bucharest (UPB) plays a key role in the AI4Media Junior Fellows Exchange Program, participating both as a sending and host institution. This dual role reflects UPB’s strong commitment to advancing AI research and fostering international academic collaboration.

UPB’s involvement is highlighted by the achievements of its researchers:

• Alexandra Andrei, a PhD student at UPB supervised by Prof. Bogdan Ionescu, participated in an exchange at the Swiss Digital Center (HES‑SO Valais Wallis) in Switzerland. Her research focused on generating realistic synthetic histopathology images using Generative Adversarial Networks (GANs), addressing the challenge of balancing data privacy with utility.
• Ana‑Maria Claudia Drăgulinescu, affiliated with UPB’s AI Multimedia Lab and supervised by Prof. Bogdan Ionescu, completed a two-month exchange at the Université Côte d’Azur (UCA) and the I3S Laboratory in Sophia‑Antipolis, France. Her work centered on multimedia network control using Column Generation optimization and Graph Neural Networks, resulting in a continued collaboration and plans for joint publications.

AI4Media Benchmarking Platform

UPB’s AI Multimedia Lab has also delivered technologies and prototypes tailored to AI4Media’s goals. Notably, it developed the AI4Media-Bench benchmarking suite and the AI4Media EaaS platform (AI4Media Ecosystem-as-a-Service prototype). These assets support systematic evaluation and sharing of AI tools within the media-focused AI research community.

A few words about this technology

The AI4Media benchmarking platform provides support for benchmarking organisers and AI developers by helping in the creation, maintenance and management of benchmarking competitions. Thus, this platform can be integrated in organisations that wish to ensure ethical considerations like data protection, fairness and robustness in performance assessment, and reproducibility for AI models. The platform is deployed as an online Evaluation-as-a-Service (EaaS) tool, providing functionality for creating and deploying benchmarking competitions, offering common data, data splits, metrics for evaluating the performance of AI models, concept definitions, as well as a complete history and logging of the submitted runs, automatic scripts for detecting errors at submission time, and integration with 3rd party cloud providers for hosting and running competition-related scripts and services, as well as distributed computing for running the AI models in a reproducible environment. Furthermore, we wish to encourage the development of environmentally friendly AI models and provide methods of integrating metrics that can measure the computational complexity of the submitted AI models.

Who can benefit

The AI4Media benchmarking platform is geared towards integration in research organisations and companies that are interested in benchmarking the performance of AI models for a series of tasks in a fair and reproducible way. These entities can either organise their benchmarking competition for internal use (choosing an optimal AI model for deployment in a product or for internal use) or as a publicly available platform, in open benchmarking tasks, available to the worldwide research community.

Impact and added value for the media industry

We summarise the impact of the platform with three key points and differentiators, as follows: follows: (i) providing a comprehensive and easy-to-implement API collection that can aid competition organisers in deploying computational complexity-related metrics, while also providing an implemented time complexity metric that organisers can either use as-is or use as an implementation example for their own metrics; (ii) using both API integration and containerization-based integration for submitting participant methods, thus offering several options for competition organisers to check and implement reproducibility for the proposed AI models; (iii) offering an important EU-based benchmarking platform, that can be focused towards common AI goals for the European Community.

New developments of the technology

Future developments for this platform will be geared towards expanding the key differentiators and advantages that this platform brings to the AI benchmarking landscape. Concretely, we will look into expanding the possibilities of integrating GPU processing power into the platform either via more 3rd party cloud infrastructure options, or via integrating the infrastructure that the implementing organisation has available, as well as further development on the computational complexity metrics.

For Further Information

To learn more about POLITEHNICA Bucharest’s activities in the project, explore collaboration opportunities, or inquire about technical specifications, please contact:

📩 Prof. Bogdan IONESCU
Email: bogdan.ionescu@upb.ro