Special Session on Learning-Based Visual QoE Estimation Methods

Abstract:

Understanding the Quality of Experience (QoE) for visual media has been very important for optimal compression and content delivery for diverse video formats, and it has been a hot topic of research for the last decades. Researchers mostly studied this problem from a signal processing perspective using image and video processing tools, while learning-based methods have been increasing momentum recently. The popularisation of deep learning-based methods affected the whole signal and image processing community as a disruptive force, and visual QoE estimation is no different than others. Use of learning-based methods and especially deep learning methods open a new path for understanding the human visual system in the perception process and other QoE parameters. The objectives of this special session are twofold: first, to develop new metrics reaching beyond the performance of the legacy signal processing approaches for visual QoE estimation, and second, to understand the stages of human perception for visual media better utilising the learning-based methods and different analysis methods such as ablation studies.

Topics of interest:

This special session is aimed at addressing the new approach of employing learning-based (especially deep learning-based) methods in visual (and more generally audio-visual) QoE estimation. This special session covers the following conference topics of QoMEX 2020:

• Novel assessment and evaluation methodologies
• Immersive experiences and technologies
• Human behaviour analysis and QoE
• Visual saliency for QoE
• Visual and audio-visual QoE

Special Session Organisers:

• Dr. Emin Zerman – Trinity College Dublin, Dublin, Ireland
• Dr. Giuseppe Valenzise – L2S, CNRS – CentraleSupélec – Université Paris Sud, Gif-sur-Yvette, France
• Dr. Cagri Ozcinar – Trinity College Dublin, Dublin, Ireland

Special Session on Quality of Experience for Security-Oriented Applications

Abstract:

Security applications are very diverse. They include both image security and privacy such as watermarking or fingerprinting, the security of goods and people as in the case of video protection or anti-spoofing, but also military applications related to the detection and characterization of targets. The growth of such applications could be explained by the affordability of capturing and editing technologies during the last decade, but also because of the new security issues increasing the vulnerability of the systems. Whatever the type of application, the quality of the experience remains a significant problem in order to provide the guarantee with the usability of the content. This special session will bring experts, practitioners and researchers together to present and discuss all type of procedures and methods for security-oriented applications

Topics of interest:

• QoE for robust and resilient multimedia watermarking and data hiding.
• QoE for piracy protection of user-generated content.
• QoE for cybersecurity applications.
• QoE for CCTV, video surveillance and video protection applications.
• QoE for Biometric applications.
• QoE for military applications.
• Utility and usefulness for detection, recognition and identification tasks.
• Subjective quality assessment for security-oriented applications.
• Databases for security-oriented applications.

Special Session Organisers:

• Chaker Larabi, XLIM CNRS, University of Poitiers, France
• Mikolaj Leszczuk, AGH University of Science and Technology, Poland
• Nadia Kanwal, Athlone Insutitute of Technology, Ireland

Special Session on Catering for More User Diversity in Research on Perceptual Quality and QoE

Abstract

Psycho-physical experiments and other empirical studies, involving human participants and addressing topics related to perceptual quality and Quality of Experience (QoE) typically tend to focus on test populations that are relatively easy to reach (e.g., university students, colleagues, acquaintances and friends) and that can easily be "fitted in" certain methodological approaches (e.g., visual quality experiments according to the relevant standards). However, especially when focusing more on higher-level issues in image quality, e.g., emotion and affective states in general, quality of experience, relying primarily on "traditional" subject groups has important implications in terms of the external / ecological validity (and thus the generalizability of results).
Often, this typical approach is due to economical and/or practical constraints, and thus justifiable (and commonly accepted) from that perspective. Still, results from such studies are used to define a kind of "golden standard", even though they do not at all reflect the prevalent diversity in audiences / user segments nor capture these diverse groups' experiences and actual experiences with a range of (multi)media applications and services. In addition, test populations in studies that do include "under-represented" user groups are typically smaller, may rely to a larger extent on qualitative approaches, etc. which makes them harder to publish as they deviate more from the more dominant, traditional approaches (which also has implications for how the field and state of the art evolves). This special session therefore aims to raise this issue by inviting in particular work, empirical studies and contributions, which for example address:
Address user groups / audiences that are typically invisible or under-represented in perceptual quality and QoE studies (and the related challenges). E.g., elderly people, children and teenagers, disabled people.
Bring in perspectives and experiences from other disciplines that are relevant to deepen the understanding of e.g., interpersonal or intercultural differences (e.g., cognitive sciences, applied and social psychology, HCI, UX...) or that can help to yield insights into how to take a more proactive approach to QoE (e.g., design-inspired approaches)
Explicitly focus on user diversity when analysing data and discussing issues related to generalizability of findings, in order to better reflect the diversity in society in the body of knowledge (e.g., in-depth focus on gender-related differences, analyses taking into account diversity in terms of ethnicity and cultural differences; or in terms of different personality types, cognitive style, attitudes, use motivations, …
Reflect at a higher level on strategies for better inclusion of diverse user groups and segments in perceptual quality and QoE studies.

Topics of interest

The overall goal is to cater for more diversity in all aspects, including planning of studies, setup of test panels, running of studies, analysis of results, discussion on implications and limitations of findings. The envisioned outcomes are:

• Experience sharing and mutual learning: how to overcome the constraints in reaching out to user groups and audiences that are typically not included? What can we learn from case studies (pitfalls and promises)?
• Discussion on limitations of the common methodological practices in this respect and their implications
• Identification of strategies and best practices for how to change these common practices and to evolve towards more inclusive and diversity-friendly practices

Special Session Organisers:

• Katrien De Moor - NTNU, Norway
• Kjell Brunnström - RISE Research Institutes of Sweden

Special Session on Quality of Experience of Interaction with and within Connected Vehicles

Abstract

Emerging wireless communication networks and standards such as 4G and 5G have enabled high availability and connectivity for a massive number of users and devices. High-speed wireless networks do not only enable mobile multimedia applications (3D, 4K/8K, live video gaming streaming), they also enable operation of smart vehicles like driverless cars, drones, robotics, remote mining and telepresence robots that offer the opportunity for remote control, telepresence as well as more efficient and convenient modes of transport.
In this context, autonomous vehicles (also known as self-driving or driverless vehicles) represent mobile agents that are fully independent in decision-making. They do not need external intervention and are capable of responding to situations such as emergencies and unexpected traffic situations (e.g., congestion, rain, or black ice). Similarly, smart vehicles are equipped with multi interface cards, as well as sensors, both on board and externally (i.e. MTS, IEEE 802.11p, Bluetooth, radar, ladar, etc.) which assists the driver decisions. Control, communications, and computing technologies of smart vehicles facilitates the development of efficient transport and management applications. Smart and autonomous vehicles require fast and reliable communication networks such as 5G. In addition, smart vehicles can or have to be controlled remotely by an external operator. Remote control will also be enabled by 5G networks that can provide guaranteed network resources and enable very short response times which is crucial in applications such as telepresence and networked multimedia services with excellent QoE. Furthermore, smart and autonomous vehicles represent a new context for experiencing networked communication and multimedia services.
Several challenges and issues regarding QoE for smart vehicles remain unresolved. For example, the combination of navigation with high accuracy and high responsiveness requirements leads to issues at different influencing factor levels, impacting the perceived quality for real-time, autonomous and automated vehicles. Furthermore, we witness an increasing interest from the society in utilising smart vehicles in different contexts which further emphasizes the need to explore QoE of smart vehicles.
The main objective of this special session is to bring together engineers and researchers from academia and industry to contribute, present and discuss novel scenarios, research challenges and solutions regarding quality of multimedia experience in smart, autonomous and automated driving. This special session focus on innovative applications, tools and architectures in all technology areas related to smart vehicles, QoE in connected vehicles, telepresence, remote-controlled driving etc. Prospective authors are encouraged to submit original research papers in all major areas, which include, but not limited to the following:

Topics of interest

• Cooperative Driving and QoE-driven Traffic Management for Smart Vehicles
• Big Data Analytics for Intelligent Transportation Systems in 5G networks
• Quality of Connected Services and Mobility Management on the Internet of Vehicles
• QoE in the context of autonomous driving
• QoE driven security and privacy models in Connected Vehicle Communication systems
• Telepresence and QoE of remote control and navigation of smart vehicles
• Quality of autonomous and automated driving systems
• Principles for the communication of uncertainty due to technical performance constraints
• Congestion control and QoE-driven control in vehicular networks
• Performance evaluation techniques and methodologies for vehicular networks
• User experience with and within smart vehicles (i.e quality of navigation and control)
• Quality of life with smart vehicles’ applications and services
• Multimedia experience and smart vehicles (i.e. Audio, Video, AR, VR)

Special Session Organisers:

• Raimund Schatz, AIT Austrian Institute of Technology / AAU Alpe-Adria Universität Klagenfurt, Austria
• Alcardo Alex Barakabitze, University College Dublin, Ireland
• Hamed Jahromi, University College Dublin, Ireland
• Peter Fröhlich, AIT Austrian Institute of Technology, Austria
• Sergio Cabrero Barros, IBM Research - Ireland

Special Session on Bypassing GDPR: Synthetic Datasets and how to build and apply them

Abstract

The introduction of GDPR has created a significant challenge for researchers in many fields ranging from biometric authentication, machine learning & artificial intelligence, multimedia content management and process to name just a few. GDPR  restricts the data that can be gathered and makes it particularly difficult to engage in research that relies on personal data such as faces or similar biometric identifiers. Fortunately, there has been recent progress with new data generation, augmentation and synthesis methodologies that enable researchers to create complex data samples that are almost indistinguishable from real-world data.

Topics of Interest

In this special session, we invite a range of contributions related to synthetic datasets - how to build and apply them to real-world problems. This special session covers the following conference topics of QoMEX 2020:

• Synthetic Datasets for QoE  
• QoE piracy protection with Synthetic Datasets
• Artificial Intelligence and Synthetic Datasets.
• Assessment and evaluation methodologies for Synthetic Datasets.
• Content creation for Synthetic Datasets.
• Immersive experiences and technologies
• Annotations/Metadata issues for Synthetic Datasets    
• QoE for Biometric applications
• QoE for detection, recognition and identification tasks

Session Organisers

• Prof. Peter Corcoran, NUI Galway

Special Session on Dataset Acquisition, Curation and Validation for Emerging Consumer, HCI Technologies and Devices

Abstract

Many of today’s advanced consumer devices employ new modes of interaction with the user and their environment. As examples consider wearable AR/VR headsets, in-cabin driver monitoring systems (DMS), AI-enabled Internet-of-Things devices,  and a new generation of smartphones with advanced gesture and environmental sensing, All of these devices rely on advanced artificial intelligence to enable their state-of-art capabilities and that AI requires, in turn, large datasets to learn from. In this special session we will look at some of the approaches and methodologies that can be applied to improve the acquisition, curation and validation of such large training datasets. In particular we welcome case studies from industry and research groups that show how such datasets are built and validated in the context of practical products and services.

Topics of Interest

In this special session, we invite a range of contributions related to the acquisition, curation and validation of datasets for consumer and advanced HCI technologies and devices. This special session covers the following conference topics of QoMEX 2020:

• Immersive experiences & technologies
• New interaction paradigms
• Artificial Intelligence and QoE
• Audio/Visual user experience
• Affective Experience
• Reproducible QoE research
• Human behaviour analysis and QoE

Session Organisers

• Prof. Peter Corcoran, NUI Galway
• Joe Lemley, Xperi Corporation