Recognizing Actions in Videos Under Domain Shift
Elisa Ricci, University of Trento, Italy
Challenges and Opportunities of AI in Medical Imaging
Daniele Ravi, University of Hertfordshire, United Kingdom
Perceptive Deep Learning in Advanced Driver Assistance Systems
Francesco Rundo, STMicroelectronics srl, Italy
Recognizing Actions in Videos Under Domain Shift
Elisa Ricci
University of Trento
Italy
Brief Bio
Prof. Elisa Ricci (PhD, University of Perugia 2008) is an Associate Professor at Department of Information Engineering and Computer Science (DISI) at the University of Trento and the head of the Deep Visual Learning research group at Fondazione Bruno Kessler. She has published over 160 papers on international venues. Her research interests are mainly in the areas of computer vision, robotic perception and multimedia analysis. At UNITN she is the Coordinator of the Doctoral Programme in Information Engineering and Computer Science. She is an Associate Editor of IEEE Trans. on Multimedia, Computer Vision and Image Understanding and Pattern Recognition. She was the Program Chair of ACM MM 2020 and the Diversity Chair of ACM MM 2022. She is the recipient of the ACM MM 2015 Best Paper award and ICCV 2021 Honorable mention award.
Abstract
Action recognition, which consists in automatically recognizing the action being performed in a video sequence, is a fundamental task in computer vision and multimedia. Supervised action recognition has been widely studied because of the growing need for automatically categorizing video content that are being generated everyday. However, it is nearly impossible for human annotators to keep pace with the enormous volumes of online videos, and thus supervised training becomes infeasible. A cheaper way of leveraging the massive pool of unlabelled data is by exploiting an already trained model to infer the labels on such data and then re-using them to build an improved model. Such an approach is also prone to failure because the unlabelled data may belong to a data distribution that is different from the annotated one. This is often referred to as the domain-shift problem. To address the domain-shift, recently Unsupervised Video Domain Adaptation (UVDA) methods have been proposed. However, these methods typically make strong and unrealistic assumptions. In this talk I will present some recent works of my research group on UVDA, showing that, thanks to recent advances in deep architectures and to the advent of foundation models, it is possible to deal with more challenging and realisting settings and recognize out-of-distribution classes.
Challenges and Opportunities of AI in Medical Imaging
Daniele Ravi
University of Hertfordshire
United Kingdom
![]()
Brief Bio
Dr. Daniele Raví is a senior lecturer at the University of Hertfordshire and an expert researcher in the field of artificial intelligence (AI) applied to medical imaging, image-guided surgery, disease progression modeling, and smart sensing. He holds a BSc and MSc in Computer Science, as well as a PhD in computer vision from the University of Catania. He also has postdoctoral experience from Imperial College London and University College London and has gained industrial experience at ST Microelectronics and two innovative startups. Dr. Raví has a strong publication record, with numerous journal papers in top-tier journals, articles in international conferences, and a patent. He has contributed to several research projects funded by the EU, EPSRC, Wellcome Trust, and industry, and recently secured grants totaling over £500,000 for his research projects. These projects focus on developing and commercializing AI pipelines for healthcare applications and have a significant impact on industry and society, including cost savings and improved outcomes for patients.
Abstract
As the field of medical imaging continues to rapidly evolve, we are faced with new challenges and exciting opportunities for data analytics in health informatics and medical intervention. With the emergence of AI as a powerful tool for machine learning, we are witnessing a transformative shift in the future of automated data analysis. The predictive power and ability of AI to generate automatically optimized high-level features and semantic interpretation from input data are revolutionizing the healthcare infrastructure, making it more efficient, cost-effective, and less prone to errors. In this keynote speech, we will delve into the world of medical imaging and explore the challenges and opportunities it presents. Join me as I share some of the advanced AI solutions I have developed to tackle complex problems that have the potential to transform how we identify diseases, improve patient treatments, and monitor patient outcomes. From aiding in the early detection of diseases such as cancer and Alzheimer's to assisting surgeons during interventions, and developing smart sensors to monitor patient health, we will examine the proposed solutions and results obtained for various clinical applications.
Perceptive Deep Learning in Advanced Driver Assistance Systems
Francesco Rundo
STMicroelectronics srl
Italy
http://www.st.com
![]()
Brief Bio
Francesco Rundo received the degree in computer science engineering and the Ph.D. in Applied Mathematics for Technology from the University of Catania. He is currently Senior Technical Staff Leader of the Artificial Intelligence Team at the Automotive R&D division of STMicroelectronics of Catania. He coordinates the R&D developments about artificial intelligence solutions/deliverables at the Automotive R&D division of STMicroelectronics, Catania. He is a member of the Computer Science Ph.D. Scientific Board, Department of Mathematics and Computer Science, University of Catania. He is also a member of the Computer Science Ph.D. Scientific Board, National Ph.D.
Program of Artificial Intelligence. He has coauthored more than 100 scientific contributions in international journals, conference proceedings , special Issues series, posters, abstracts, and lectures. He is also co-inventor of several international patents in the field of deep learning applied to automotive and industrial fields. He coordinates as WP leader or Task leader as well as research member, such AI-based developments within several funded research programs both national and EU (both completed and in the initial stage/under development: ASTONISH, ADAS+, SATURN, REACTION, EDGE-AI, HICONNECTS, NEUROKIT2E, ARCHIMEDES, R-PODID, etc..). His main research interests include advanced bio-inspired models, advanced and perceptual deep learning, embedded systems for deep learning algorithms, advanced deep learning modeling, and mathematical modeling for automotive, industrial, and healthcare applications. He is a member of several international conference program committees. He serves as a reviewer and guest editor for several special issues organized by such key-editors in the field of computer science. He served as Associate Editor for IET Image Processing journal. He serves as Associate Editor for IET Networks ,Applied Computational Intelligence and Soft Computing. He serves as research-topic editor for Frontiers in Computer Science and Frontiers in Neuroinformatics, and as Topic Editor for Electronics and Drones journals.
Abstract
The contribution will cover theory and applications of perceptive deep learning approaches to driver assistance systems, from intelligent car-driver monitoring to real-time vision algorithms for intelligent driving scenario understanding as well as for driving risk-assessment. One of the challenges of modern Artificial Intelligence systems in automotive domain is linked to the development of sustainable and deployable architectures over automotive-grade hardware. The results achieved in this field will be showed highlighting the advantages of the perceptual deep learning solutions which integrate classic AI methodologies with approaches for selecting the data to be processed (saliency processing), for compensating noise in input data (adversarial compensation) and for domain adaptation. Furthermore, driving assistance techniques based on bio-inspired methods will be discussed. Some use-case applications delivered within EU scientific research programs will be treated.