master Assignment
Heritage Biometrics: does every painter paint himself?
Type: Master EE/CS/AM
Student: Unassigned
Duration: TBD (min. 6 months/fulltime)
If you are interested please contact:
Background:
From the early Renaissance to the present day, art historians often encounter striking facial similarities in the work of major artists, such as Jan van Eyck, Hans Memling, Rembrandt, Van Dyck and others. One particular theory is that artists fused their likeness from their own self-portraits into the faces of their sitters. The Data Management and Biometrics group has 20 years of experience in 2D and 3D face recognition systems. Recently, the Biometrics group worked on face recognition as a tool to identify Roman emperors. By using transfer learning, it utilizes existing deep-learning facial recognition models, augmented with images of Roman imperial portraits, to provide a new empirical foothold in the debate of Roman emperor recognition.
Goals:
The goal of this master thesis is delineating a roadmap to assess face similarity across portraits. In particular, the student will be responsible to collect images of disputed and undisputed artists’ self-portrait and portraits of characters painted by the same artists, to implement algorithms for face-to-face matching of painted portraits, design tools to learn which facial features are most to least discriminative.
Supervisor and main contact: Dzemila Sero (assistant professor DMB)
Co-supervisor: Luuk Spreeuwers (associate professor DMB), Simon Abrahams (art historian)
Your profile:
You are a graduate student with a strong experience in machine learning and computer vision. The project requires critical thinking and exploring new directions, so you will also have the opportunity to go beyond current approaches.
Why join?
- Work at the intersection of art history and computer vision to make an impact in the new field of Heritage Biometrics.
- Collaborate with a team of interdisciplinary experts of engineers, professionals in decorative arts and computer scientists.
Who Should Apply?
Students with a strong background in machine learning and computer vision. Affinity or interest in the world of decorative arts is expected.
References:
- Ramesh, Darshan Srirangachar, et al. "Facial recognition as a tool to identify Roman emperors: towards a new methodology." Humanities and Social Sciences Communications 9.1 (2022): 1-10.
- Lestriandoko, Nova Hadi, et al. "The Role of Facial Hair on Roman Emperors' Face Recognition." 2024 12th International Workshop on Biometrics and Forensics (IWBF). IEEE, 2024.