Artūras Nakvosas, a doctoral student at Vilnius University's Faculty of Mathematics and Informatics (VU MIF), outperformed rivals from around the world in a prestigious international artificial intelligence (AI) competition organised by the International Joint Conference on Biometrics (IJCB 2026), taking place in Rome.
"The win was a very pleasant surprise – I was genuinely astonished," says Nakvosas, reflecting on his victory. His academic supervisor is Prof. Dr Virginijus Marcinkevičius (VU MIF).
The researcher's achievement reflects the strong research potential of the VU Faculty of Mathematics and Informatics in the fields of data science and biometric AI.
Two categories – two awards
The competition was held in two categories – the Full Data Track, in which participants could use the entire provided dataset, and the Limited Data Track.
Nakvosas, who works in the Smart Technologies Research Group at the VU MIF Institute of Data Science and Digital Technologies (VU MIF DMSTI), took first place in the Full Data Track category of the "Adapting Foundation Models for Face Recognition" (AFMFR-2026) competition – one of ten competitions organised by the conference. In the Limited Data Track category, Nakvosas came second.
Competitors had to adapt state-of-the-art AI models for the task of facial recognition, using exclusively synthetic training data provided specifically for the competition within a privacy-preserving environment.
Teams were assessed on how well their adapted models performed across a range of test datasets – a technically very demanding aspect of the task. Teams of researchers from all over the world took part in the competition. The winning team, "DMSTI-Neurotechnology," had just one member – Nakvosas.
Over the past decade, facial recognition technologies have advanced considerably, yet their development is increasingly hampered by limited access to large datasets of real faces due to privacy, legal and ethical requirements. As a result, researchers are paying growing attention to synthetic facial images and pre-trained AI models, which make it possible to build privacy-preserving and efficient facial recognition systems.
About IJCB
IJCB is a top-tier scientific conference in the field of biometrics. It showcases research in facial recognition, fingerprint and iris biometrics, gait and voice analysis, as well as discussions on the application of artificial intelligence to identity verification and privacy protection.
IJCB is organised in collaboration with two international scientific organisations – IEEE (the Institute of Electrical and Electronics Engineers, USA) and IAPR (the International Association for Pattern Recognition) – and is regarded as the foremost global platform enabling biometrics researchers to present their latest research and compete in challenges.
As the winner, Nakvosas has been invited to become a co-author of the official competition paper. The paper has been published in the IEEE/IAPR IJCB 2026 proceedings and will be presented at the conference in Rome on 1–4 September 2026.
17 August 2026