Nagrado letos prejmejo:

  • Jovana Videnović, Faculty of Computer and Information Science, University of Ljubljana (UL FRI), for the publication A Distractor-Aware Memory for Visual Object Tracking with SAM2
  • Martin Justin, Faculty of Arts, University of Ljubljana, and Faculty of Arts, University of Maribor (UL FF and UM FF), for the publication More Hope for Conciliationism

In addition to the prizes, the committee selected the following students to receive commendations for their achievements:

  • Tina Šaula, Biotechnical Faculty, University of Ljubljana (UL BF), for the publication Enrichment of the Nutritional Value of Pea Flour Milling Fractions Through Fermentation
  • Vita Movrin, Jožef Stefan Institute (IJS), for the publication Initiation of Epithelial Wound Closure by an Active Instability at the Purse String
  • Vladimir Smrkolj and Aljoša Škorjanc, Faculty of Medicine, University of Ljubljana (UL MF), for the publication GOReverseLookup: A Gene Ontology Reverse Lookup Tool

The Dr Uroš Seljak Prize is being presented for the fourth consecutive year, based on a public call for applications. Its purpose is to encourage scientific research among students and to recognise and reward outstanding achievements in the field of science. Special emphasis is placed on the role of mentors, who receive commemorative plaques alongside the prize winners.

The prize was established at the initiative of Dr Uroš Seljak, Professor in the Department of Physics at the University of California, Berkeley, alumnus of the University of Ljubljana, and mentor at the American Slovenian Educational Foundation (ASEF). In 2021, he received the prestigious Gruber Cosmology Prize, awarded by the Gruber Foundation at Yale University. With the funds from this prize – totalling $150,000 – he established a charitable prize fund that finances the prize bearing his name, with an annual prize amounting to $10,000.

Awarded works

Jovana Videnović’s (UL FRI) research work focuses on the problem of visual object tracking where an algorithm has to recognise and segment the same object throughout an entire video recording based solely on one marked image. Modern algorithms have trouble doing this, particularly in the presence of similar objects (distractors), which disrupt tracking, or when the target temporarily disappears from the field of vision.

Videnović has developed the DAM4SAM method, which is based on an analysis of Meta’s SAM2 model and features an advanced memory system and a mechanism for updating the visual model of an object. She has also designed a new database for analysing tracking algorithms in complex conditions. The DAM4SAM method outperformed existing trackers in ten standard databases and has potential for wider application. Videnović’s research is an important contribution to the development of artificial intelligence, and can provide the basis for new applications in navigation, biomedicine, video production and surveillance systems.

Martin Justin’s (UL FF and UM FF) research work focuses on the philosophical question of how to act when we disagree with someone who has the same knowledge or outlook as us. In the paper “More Hope for Conciliationism”, which was published in the prestigious journal Episteme, he defends the conciliatory principle, which proposes that we reduce trust in our own viewpoints in the event of disagreement.

He showed that many of the criticisms of this principle are based on a confusion between the problem of disagreement and the more general issue of higher-order evidence. His analysis reaffirms validity of the conciliatory principle and is an important contribution to modern epistemology. It opens up space for a more rational treatment of disagreements in science, philosophy and wider society.

Tina Šaula (UL BF) is working on improving the nutritional value of milling fractions of pea flour using lactic acid fermentation. Her interdisciplinary study analysed the impact of the preparation of milling fractions at industrial scale, and of fermentation conditions on nutrient content, anti-nutrients, gamma-aminobutyric acid (GABA), polyamines and biogenic amines.

Her research has shown that milling fractions with smaller particles contain more proteins, minerals and polyamines, while starch-rich fractions with larger particles contain fewer phytates, which increases the biological availability of iron. Fermentation using the Lactiplantibacillus plantarum bacterium resulted in reduced levels of anti-nutrients, such as phytic acid and histamine, and in increased levels of GABA, while spontaneous fermentation led to the formation of nutritionally undesirable biogenic amines.

The project resulted from successful collaboration between the academic sphere and industry, and is an important contribution to circular economy goals. Its scientific excellence was confirmed by publication in the journal Food Chemistry (IF 9,8), where the candidate was lead author.

Vita Movrin’s (IJS) research work focuses on a theoretical analysis of the process of wound closure in epithelial tissue. As the outer lining of organs throughout the body, epithelial tissue provides crucial protection against external influences. The existing models assumed that wounds healed because of increased mechanical tension at the edges. Vita Movrin has, however, developed an entirely new theoretical approach. Her model treats the epithelium as a dynamic autonomous system in which healing is triggered spontaneously as a result of a feedback loop between the deformation of cells and the formation of active tension at points of intercellular contact.

Through an analysis of stability and numerical simulations, she shows that a wound can be the result of dynamic instability and not of pre-assumed forces. This is the first model of its type to open up new possibilities for understanding the mechanical response of tissue, and makes an important contribution to theoretical biology and the development of biomechanical approaches in medicine.

Vladimir Smrkolj and Aljoša Škorjanc have collaborated on developing a new bioinformation tool, GOReverseLookup, which is a conceptual breakthrough in gene analysis. Instead of starting with a known list of genes, researchers can now proceed from a biological phenomenon, such as inflammation or cancer, with the tool then calculating which genes are statistically linked to it. This approach is particularly valuable in the early stages of research, and facilitates the analysis even in the case of rare diseases and less well-researched conditions. The tool incorporates data from several bioinformation databases, and supports the translation of findings between any organisms.

GOReverseLookup has been successfully tested with rheumatoid arthritis, chronic inflammation and tumorigenesis. The tool enables research to be conducted on the from phenomenon to gene principle, includes the advanced configuration of analyses and incorporates orthologous genes from model organisms. With its open access and practical utility, the tool is an important contribution to the development of modern bioinformatics and functional genomics.