eXtended Reality (XR) is an emerging umbrella term for all the immersive technologies (Virtual Reality, Augmented Reality, and Mixed Reality) able to extend the user senses by blending the virtual and real worlds and creating a fully immersive experience.
Artificial Intelligence (AI) is a broad area of computer science that makes machines seem like they have human intelligence. The goal of AI is to mimic the human brain and create systems that can function intelligently and independently.
The XR&AI Summer School is focused on the synergy of these technologies for cultural and territorial heritage, medicine and industry.
The School is open to Ph.D. and Master students, post-doctoral researchers, and academic or industrial researchers.
The XR&AI Summer School is a satellite event of the International Conference on eXtended Reality (XR Salento).
Take a look back at the previous schools:
- Pasquale Arpaia – Università di Napoli Federico II, Napoli (Italy)
- Lucio Tommaso De Paolis – Università del Salento, Lecce (Italy)
- Ugo Erra – Università della Basilicata, Potenza (Italy)
- Emanuele Frontoni – Università di Macerata (Italy)
- Nicola Masini – CNR-ISPC, Potenza (Italy)
- Roberto Pierdicca – Università Politecnica delle Marche, Ancona (Italy)
- Andrea Tarallo - Università di Napoli Federico II, Napoli (Italy)
- Antonio Emmanuele Uva – Polytechnic University of Bari (Italy)
- Raffaele Vitulli – Cluster Basilicata Creativa, Matera (Italy)
- Primo Zingaretti – Università Politecnica delle Marche, Ancona (Italy)
- Nicola Felice Capece - Università della Basilicata, Potenza (Italy)
- Carola Gatto - Università del Salento, Lecce (Italy)
- Gabriele Gilio - Università della Basilicata, Potenza (Italy)
- Nicla Notarangelo - Università della Basilicata, Potenza (Italy)
- Silke Miss – XRtechnology srl, Lecce (Italy)
- Ileana Riera Panaro – Università del Salento, Lecce (Italy)
Rita Cucchiara, Università di Modena e Reggio Emilia
Generative AI: from image to text and return
Joaquim Jorge, University of Lisboa
Challenges and Opportunities for AI&XR in Health Applications
Andrea Bottino, Dipartimento di Automatica e Informatica, Politecnico di Torino
AI for procedure learning in XR
Joseph L. Gabbard, Virginia Tech, US
Implications of AI for next-generation AR user interfaces
LECTURE 1 - Generative AI: from image to text and return
Generative Artificial Intelligence is one of the hot topics in Machine Learning in the last few years: the transition from Recurrent Networks to Attention-based Encoder-Decoder Neural Architectures allowed effective textual/visual generation both synthetically and semantically. Large Scale Models provided pre-trained networks interesting not only as research prototypes but also as well-known products ( Bert, GPTx, DallE-x etc). Indeed, prompt -based generation is a new and powerful paradigm where research is still open. The seminar considers the specific problem of visual data description by synthetic text (often know as image/video captioning) and the many variants oriented to a broad description, controllable text generation as well as a precise measure of a human-like soundness. Recent scientific results in CVPR-ICCV venues and possible applications in content description, human-robot interaction and virtual environment understanding will be discussed. Finally, the reverse problem will be addressed too, i.e., the image fake generation from textual prompt with specific reference to the challenging problem of fake image detection.
Rita Cucchiara is Full Professor of Computer Engineering at the University of Modena and Reggio Emilia (UNIMORE), Department of Engineering "Enzo Ferrari", Italy. She is Director of the Center for Artificial Intelligence Research and Innovation, AIRI, of UNIMORE and coordinates the research lab AImagelab, gathering more than 50 researchers in research areas related to AI, Computer Vision, Machine Learning, Multimedia. Since 2020 she is ELLIS Fellow and Director of the Modena unit of the European ELLIS network. In 2016-2018 she has been President of Italian Association in Computer Vision , Pattern Recognition and Machine Learning (CVPL) and In 2018-2021 Director of the National CINI -AIIS AI.. In 2020 she coordinated the working group on Artificial Intelligence of MUR National Research Plan 2021-27 and collaborates with the Presidency Minister Council of Italy for Artificial Intelligence. Since 2021 she is in the Advisory Board of Max Plank IIS (Tubing, Germany) and of the Computer Vision Center (Barcelona, Spain). Since 2022 she is member of the Director Board of Italian Institute of Technology (Italy), Prometeia spa and ART-ER. She served as GC of ACMMM2020,2023 ECCV2022 and CVPR2024 and PC of ICCV2021, and Associated Editor of IEEE T-PAMI. She has been co-writer of more than 450 scientific papers with H-index 62 and more than 23900 citations (Google Scholar March 2023).
LECTURE 2 - Challenges and Opportunities for AI&XR in Health Applications
Recent advances in VR and AR technology have enabled interactive graphics applications to support healthcare professionals in training, diagnosis, planning, and treatment. This field has progressed enough to warrant a course that can inspire new ideas within the graphics community. Medical images create virtual human anatomy models, allowing for natural interaction and visualization in healthcare scenarios. VR and AR are conceptually different and suited for different types of problems. VR is immersive and suitable for learning anatomy, surgical skills, and analyzing 3D medical data. AR, on the other hand, overlays helpful information onto the physical environment, making it useful for tasks such as communication with patients and training assistants. However, several challenges, such as nonstandard equipment and disorientation, still limit the widespread use of these technologies. This course will cover current advances and challenges in this area, including integrating AI techniques.
Joaquim Jorge holds the UNESCO Chair of Artificial Intelligence & Extended Reality at the University of Lisboa, Portugal. He joined Eurographics in 1986 and ACM/SIGGRAPH in 1989. He is Editor-in-Chief of the Computers and Graphics Journal, Eurographics Fellow, ACM Distinguished Member, and member of IEEE Computer Society Board of Governors. He organized 50+ conferences, including Eurographics 2016 (IPC CO-Chair), IEEE VR 2020/21/22 as co-(papers)chair, and ACM IUI 2012 (IPC co-chair). He served on 210+ program committees and (co)authored over 360 peer-reviewed publications and five books. His research interests include graphics, virtual reality, and advanced HCI techniques applied to health technologies.
LECTURE 3 - AI for procedure learning in XR
In many fields, procedures define work and activities. These procedures consist of well-defined sequences of instructions that are followed in a specific order and according to specific rules. For example, maintenance work in industry is defined by procedures that ensure that work is performed according to maintenance strategy, policies, and programs. However, these procedures can be complex and difficult to remember, requiring learning and training for new employees and refresher training for existing personnel. This is also true in other areas, such as learning safety and medical procedures, training employees, and improving decision-making processes. Procedures are used in almost everything we do, such as reading, writing, calculating, driving, cooking, and classifying concepts.
To teach procedural skills well, teachers need to provide activity-specific instructions that describe the logical connections between activities...according to the form of the procedure. Extended reality technologies (XR) can be used to create immersive and interactive learning environments that help learners acquire procedural knowledge and skills through self-training and self-assessment. XR technologies can simulate realistic environments that help learners practice procedures in a safe and controlled environment. For example, a trainee can practice complex medical procedures in a virtual environment before performing them on real patients. In addition, XR can enhance the learning experience by providing multimedia, interactive information, context awareness, and adaptive information delivery. Learners can interact with virtual objects, receive audio and visual feedback, and track their performance and progress in real time, allowing them to identify and correct errors.
The integration of artificial intelligence (AI) systems can significantly improve self-learning systems by making them more adaptable and tailored to learners' needs. In particular, AI enables the implementation of adaptive learning approaches that constantly monitor and analyze learners' performance and adapt content and learning methods to their needs and progress. Such adaptive learning approaches can be supported by a virtual instructor with conversational capabilities. This instructor has two main functions: providing vocal and visual instructions to deliver learning content, and answering to direct requests from learners with explanations and clarifications. This interaction is critical to maintaining motivation and enhancing the learning experience. Finally, AI can be used to control realistic NPCs that actively participate in complex learning scenarios.
This course on procedure learning in XR will discuss best practices for creating effective XR/AI-based learning programs and experiences that meet the needs of diverse learners and achieve specific learning objectives.
Andrea Bottino is Associate Professor at the Department of Control and Computer Engineering of the Politecnico di Torino, where he heads the Computer Graphics and Vision research group of the same university. His current research interests include Computer Vision, Machine Learning, Human Computer Interaction, Serious Games, and Virtual and Augmented Reality.
LECTURE 4 - Implications of AI for next-generation AR user interfaces
As we see augmented reality (AR) applications move from research labs to commercial applications, the need for usable AR-based systems has become more and more evident. Despite the fact that AR technology fundamentally changes the way we visualize, use, and interact with computer-based information, we are only just recently starting to see real investment in the design and development of user interface (UI) designs that truly leverage AR’s promise. While traditional UX methods can be applied to determine what information should be presented to users, these approaches do not tell us how and when information should be presented to the user. Given the infancy of the AR UI design field, we mostly see smartphone, tablet and even desktop UI metaphors migrating to AR, in what we can consider first-gen AR user interfaces.
With recent advancements in computer vision and machine learning, we have a unique opportunity to consider intelligent next-generation AR user interfaces. In industrial settings for example, we can consider these next-gen AR interfaces to adapt to, the worker, their work progress and quality, as well as the overall work environment. In this talk, I discuss the implications for AI for next-generation AR user interfaces. Specifically, I provide some specific examples of how AI-powered AR user interfaces could adapt to users’ contexts, and touch on both the promise and challenges associated with creating and fielding these next-gen UIs. I conclude with open questions regarding issues of inclusion, privacy and agency.
Dr. Joseph L. Gabbard is an Associate Professor of Human Factors at Virginia Tech. He holds Ph.D., M.S. and B.S. degrees in Computer Science (HCI) and a B.A. in Sociology from Virginia Tech. Dr. Gabbard’s work centers on human-computer interaction in augmented and virtual reality. For nearly 25 years, Gabbard has been researching new methods of design and evaluation for interactive virtual and augmented reality systems. Currently, Dr. Gabbard directs the COGnitive Engineering for Novel Technologies (COGENT) Lab, conducting basic and applied HCI and human factors research. His work focuses on the application of principles and theories from several disciplines to the design of augmented reality user interfaces, information presentation and user interaction. His research has been funded by the National Science Foundation, National Institute of Standards and Technology, Microsoft, Meta, the National Institutes of Health, the Office of Naval Research, and several automotive manufacturers.
Fabio Galeandro, National Archaeological Museum and Archaeological Park of Egnazia
Fabio Galeandro is an archaeological officer of the Puglia Regional Directorate Museums. Director of the "Giuseppe Andreassi" National Archaeological Museum and Archaeological Park of Egnazia (Fasano - BR) he is also director of the National Archaeological Museum and Castle of Gioia del Colle (BA) and, until april 2023, of the Archaeological Park of Monte Sannace (Gioia del Colle - BA). From 2012 to 2018 he was archaeologist official at the Archaeological Park of Pompeii where he was responsible for Regio VI, the necropolis and Suburban Villas, the Catalog and Inventory Office, the functional area of the demo-ethno-anthropological heritage and the protection of the territory of the Municipality of Pompeii.
Classic archaeologist, he graduated and specialized at the University of Bari, where he was a research fellow (Assegnista di Ricerca) for four years and has a Ph.D. in Archaeological and Historical-Artistic Sciences from the University of Naples Federico II. Engaged in university and post-university teaching activities, author of numerous abstracts in specialized journals, miscellaneous, and of a monograph, he was and is the planner, sole responsible and director of the works of multiple interventions concerning enhancement, digitization, restoration, accessibility of the cultural heritage.
Paolo Sernani, University of Macerata, Italy
A levelling tutorial to the Unity Editor
Marco Bertini, University of Florence, Italy and Paolo Mazzanti, University of Florence, Italy
Innovative and Sustainable Approaches for User Engagement and Digital Interaction with Cultural Heritage
Luca Melchionna, Machimeria
Conversational automations for cultural institutions. Best practices and hands-on approaches
Nicola Capece, Università della Basilicata, Italy
Exploring the Potential of Hand Tracking and Gesture Recognition through Neural Networks in Unity 3D
LECTURE 1 - A levelling tutorial to the Unity Editor
The practical lecture is a hands-on session about the development of projects in the Unity Editor. The main goal is to level the participants to a common understanding of the main features of Unity for the other practical lectures. The students will get familiar with the best practices for the set-up, development, and building process of projects in Unity.
Paolo Sernani is Assistant Professor at the Department of Law of the University of Macerata, where he is currently in charge of the courses of “Digital Processes and Technologies” and ”Information Systems for Transportation”. He received a Ph.D. degree in Information Engineering at Università Politecnica delle Marche in March 2016, defending a thesis entitled “Design and virtualization of intelligent systems for the management of assistive environments” dealing with the application of multi-agent systems to manage assistive smart environments. After his PhD, he was involved in projects where he used Unity for the development of AR applications. Currently, his main research interests include expert systems, deep learning, and hybrid techniques.
LECTURE 2/1 - Innovative and Sustainable Approaches for User Engagement and Digital Interaction with Cultural Heritage
Within the H2020 ReInHerit Project the Media Integration and Communication Center (MICC) has developed a Toolkit that uses CV and AI technologies with the primary goal of letting visitors to interact with collections and exhibitions in a playful approach, based on gamification and learning-by-doing techniques. Playful, interactive, and immersive experiences can be employed to evoke emotions, inspire creativity, and facilitate participative digital learning scenarios in a "play and interact" visiting style.
In this lecture will be analysed the technical requirements needed to develop this type of applications and the computer vision techniques needed to implement this type of interaction. Participants will be invited to use, test, and explore the Toolkit in a collaborative and interdisciplinary approach, connecting technological and cultural sectors.
Marco Bertini is an associate professor of computer science at the School of Engineering of the University of Florence and director of the Media Integration and Communication Center (MICC) at the same university. His interests regard computer vision, multimedia, pattern recognition, and their application to different domains such as cultural heritage. He has been general co-chair, program co-chair and area chair of several international conferences and workshops on multimedia and computer vision (ACM MM, ICMR, CBMI, etc.), and was associate editor of IEEE Transactions on Multimedia. He has been involved in different roles in more than 10 EU research projects. He is a co-founder of Small Pixels, an academic spin-off working on visual quality improvement based on AI. Presently involved in the ReInHerit H2020 Project.
Personal Page HERE.
Paolo Mazzanti is a researcher at MICC with an interdisciplinary university education: graduated in theoretic philosophy, post-graduated in multimedia content design and in planning and communication of cultural heritage. His research interests focus on emotions and informal learning in museums, user-experience and interaction design, new media and digital tools for user engagement. Winner of @diversity European Competition in 2013, innovative ideas for cultural and creative sectors in Europe. Scientific Co-ordinator of "MuseiEmotivi" training at Competence Center NEMECH - New Media for Cultural Heritage. Involved in national and EU research projects related to Cultural Heritage among which UMETECH, 3DCOFORM, ENRICH, MICHAEL. Member of EU Network NEMO and part of the working group LEM - The Learning Museum. Editor and co-author of the LEM/NEMO report ``Emotions and Learning in Museums”. Presently involved in the ReInHerit H2020 Project. Personal Page HERE.
LECTURE 2/2 - Conversational automations for cultural institutions. Best practices and hands-on approaches
As the full impact of generative AIs shakes cultural institutions to their core, previous lessons learned through experiments with predictive AIs seem particularly valuable because they point at possible solutions being developed in safer and more comfortable contexts. In this presentation, some best practices will be explored and discussed. The examples selected will focus on issues of alignment of the project with the mission, editorial management, definition of the project scope and audience engagement.
Luca Melchionna is an Italian journalist and entrepreneur. Born in Bressanone in 1971, he studied History at the Universities of Bologna and Edinburgh. He worked for 11 years in the MART press office, and later collaborated with Palazzo Grassi-Pinault Foundation, Fondazione Fitzcarraldo, TSM Trento and contributed to the debate on audience development and the management of technological innovations in the museum sector. He is one of the founders, in Trento, of "Machineria - Stories that Work", a company that develops content for the cultural sector - and which has just released a chatbot for the Colosseum Archaeological Park.
LECTURE 3 - Exploring the Potential of Hand Tracking and Gesture Recognition through Neural Networks in Unity 3D
This lesson introduces hand tracking and gesture recognition using neural networks, implemented with Keras and TensorFlow, and integrated into the Unity 3D game engine. Through practical exercises, students will learn the principles of neural network design and training and the implementation of hand tracking and gesture recognition in Unity 3D. By the end of the lesson, students will have a fundamental understanding of the potential of these technologies in interactive systems and the skills to implement them using Keras, TensorFlow, and Unity 3D.
Nicola Capece is a postdoctoral researcher at the University of Basilicata in Italy. His research interests encompass Real-Time and Offline Rendering, Deep Learning, Computer Graphics, eXtended Reality (XR), and Human-Computer Interaction (HCI). He has authored numerous papers published in international journals and conferences. He is involved in several research projects concerning XR and HCI for cultural heritage storytelling, virtual dressing rooms, virtual reality 3D modelling, and hand gesture recognition systems. Additionally, he is an adjunct professor of Software Engineering and Virtual Archeology Lab courses.
For the payment of the school participation fee is mandatory to carry out the school registration.
The XR&AI Summer School 2023 registration fee includes lectures, daily lunches and coffee breaks as well as the pizza party and the guided visit of the town.
The registration fee does not include the accommodation costs.
Early bird registration ends on June 30th, and regular registration is open from July 1st.
The following registration options are provided:
- Early bird registration: € 550,00
- Regular registration: € 600,00
N.B.: The "early bird registration" payment is to be made no later than June 30th.
N.B.: The invoice will be issued with 'reverse charge' for institutions/companies based in EU that are registered in VIES (VAT Information Exchange System).
The payment can be carried out:
by bank transfer
Bank transfer details:
Beneficiary: XRtechnology srl
Bank name: Banca Sella
Reason: XR&AI Summer School 2023 + name of person attending the school
by credit card through the following PayPal link [CLICK HERE]
Please send copy of the payment receipt by email to firstname.lastname@example.org
To receive the invoice for the school registration fee, please download here the application form and return it completed at email@example.com.
For further information, please contact chairs at firstname.lastname@example.org
Casa delle Tecnologie Emergenti
Via S. Rocco, 1
75100 Matera, Italy
From Bari Palese Airport
The nearest airport is Bari Palese (60 km), from where you can get to Matera in the following ways:
- by direct shuttle bus service served by Miccolis (see the “Direct Booking” section);
- by direct shuttle bus service (see the “Bus Tour of Aeroporti di Puglia” section);
- by direct shuttle bus service by Grassani E Garofalo (see the timetable);
- by train with the Nord Barese Railway Line (ferrovienordbarese.it), reaching the station of Bari Centrale. Leaving the station, you get Piazza Aldo Moro: at that point, turn left to reach FAL (Ferrovie Appulo Lucane), where several trains arrive at Matera Centrale Station in about 80 minutes;
- by private transfer booking by sending an e-mail to email@example.com.