Advanced Image-based Biometrics
With the advances of computer vision with deep learning biometrics gained a lot of extra traction. This is evident from the papers published in the top-notch conferences like CVPR, ICCV, ECCV, FG, IJCB, journals like PAMI, TIFS, TIP, IVC, KBS, and also in a number of competitions organised each year. The research conducted in the field of biometrics improves security and privacy issues in society - two of very relevant questions that even the general public is very interested in. Based on the student doctoral topic, we will link it to selected biometrical modality to address open research questions, for instance, in: segmentation and recognition models, bias assessment, generation of synthetic datasets, privacy-preserving biometrics, deidentification, input quality assessment and enhancement, deepfake detection.
Restrictions/Prerequisites: Familiarity with the basic computer vision and deep learning approaches. Maximum number of students: 3.
