The members of the project team organized the international conference FG 2015 (11th IEEE International Conference on Automatic Face and Gesture Recognition; http://www.fg2015.org), in the following roles: Aleš Leonardis (General Chair), Matej Kristan (Evaluation Chair), Danijel Skočaj (Publication Chair), Luka Čehovin (Web chair). The conference took place in Cankarjev dom in Ljubljana between 4th and 8th of May. The conference is the premier international forum for research in image and video-based face, gesture, and body movement recognition. It is a very established conference with a long tradition; since 1995 it is organized biannually. It attracts the most established researchers from the fields of image and video analysis for face, gestures, and gait recognition; in FG 2015 there were 219 registered participants. Its broad scope includes advances in computer vision, pattern recognition, computer graphics, and machine learning. By organizing this event, the members of the project team strengthen their position in the international research community and increased the visibility of the Slovenian research and organizational potentials and skills. The conference also had very good media coverage that contributed to the dissemination of our research to the general public in Slovenia. The members of the project team were also very active in organizing other international research meetings. From 3-5. 2. 2016 they organized workshop Computer Vision Winter Workshop CVWW 2016 in Rimske toplice. Every year, together with colleagues from abroad, they organisee in conjunction with one of the largest conferences in the field of computer vision, VOT - Visual Object Tracking Workshop; VOT 2014 was held at ECCV in Zurich, Switzerland, VOT 2015 at ICCV in Santiago, Chile, VOT 2016 at ECCV in Amsterdam, Netherlands, and VOT 2017 at the ICCV in Venice, Italy.
B.01 Organiser of a scientific meeting
COBISS.SI-ID: 11014740The members of the project team also gave several invited lectures, such as at MTA SZTAKI in Budapest, Hungary, where the project and its initial results were presented. This is an additional way to disseminate the project results and to exchange the ideas with other researchers addressing similar problems. In addition to the invited talk we had several discussions about our research work and possible collaborations on the research topics that have been addressed in the framework of this project.
B.04 Guest lecture
COBISS.SI-ID: 1536824771The area of artificial intelligence and, in particular, machine learning, has recently become very successful on the wings of deep learning methods. Therefore, it is also increasingly exposed to the public, and the messages given in mass media are not always completely correct. Therefore, it is important that researchers in this field appear in public, explaining the developments in this field and predicting trends, rejecting fears and raising interest in this research field. The members of the project team often appear in the media, where we discuss these issues, as well as present our results, including the results of this project. In addition to the one-hour broadcast Akcent on the national television RTV SLO, which was attended by the project leader, the members of the project team appeared in the media also at other occasions on TV channels Kanal A and Pop TV, radio Val 202, daily newspaper Dnevnik, etc.
F.18 Transfer of new know-how to direct users (seminars, fora, conferences)
COBISS.SI-ID: 35486557With the development of machine learning methods, the power of the developed methods and the managers that operate with them, is undoubtedly increasing, which is even more important when considering the increase of the amount of data available to these methods. It is therefore urgent to question the ethical, moral and legal aspects of this development. In this article, we discuss some of these issues in the field of automatic decision-making where automatic decisions in different areas of our lives are being made. We divide methods for addressing such problems into those that operate on the principle of predefined rules and those that decide according to the learned model. The latter type of approaches is based on learning from the data generated by human decision makers, which has certain implications that we need to be aware of. As socially responsible researchers in the field of artificial intelligence and machine learning, we should be aware of these implications and dilemmas ad we have to properly address them as well.
B.03 Paper at an international scientific conference
COBISS.SI-ID: 1537320643During the project, the members of the project team put a lot of emphasis on mentoring undergraduate, postgraduate and doctoral students introducing them into the world of research. The research topics that were the subject of research on the project, such as Convolutional neural networks, were also included in the lectures and thus enriched the pedagogical process with the latest trends. During the execution of the project, three doctoral students were awarded a PhD degree under the mentorship of the members of the project team: Peter Uršič, Luka Čehovin and Barry Ridge. The first two of them also actively participated in the project. A number of diploma theses related to the work on the project, were also done. The students were also directly involved in the work on the project and thus, under the supervision and guidance of experienced members of the project team, were involved in research work. Four of them decided to stay in academia and joined the laboratory, and continue to successfully work on other research projects.
D.09 Tutoring for postgraduate students
COBISS.SI-ID: 11483476