Ph.D. started in: 2017
Expected year of graduation: 2020
COINS consortium member: Norwegian University of Science and Technology
Supervised by: Christoph Busch
Research area: Authentication and Biometrics
Project title: Automated Authentication of Audiovisual Contents: A Biometric Approach
Project description: Thanks to recent advances in computer-generated imagery and speech synthesis, it is possible to produce convincing real-life like videos of talking faces.
However, as with any other technology, there is the potential to use this possibility for malicious deeds such as generation of fake news contents.
The goal of this project is firstly to stimulate further research on detecting false presentations of talking faces by collecting a dataset of existing technologies for fake content generation along with a multimodal baseline system.
The similarity of this task to presentation attack detection and biometric authentication encourages us to adapt these technologies for the baseline system, evaluate them on the collected data, and to do further research on improving them and developing new ones.
As an example, presentation attack detection techniques based on quality measures and features like local binary patterns have shown good generalization for detection of false presentations of human faces and voices.
Existance of such systems would bring back the trust on audiovisual content needed for their use as a reliable source of information in social media or any other type of digital communication.
- Ali Khodabakhsh, Marius Pedersen, Christoph Busch (2019). Subjective Versus Objective Face Image Quality Evaluation For Face Recognition
- Ali Khodabakhsh, Raghavendra Ramachandra, Christoph Busch (2019). Subjective Evaluation of Media Consumer Vulnerability to Fake Audiovisual Content
- Erwin Haasnoot, Ali Khodabakhsh, Chris Zeinstra, Luuk Spreeuwers, Raymond N.J. Veldhuis (2018). FEERCI: A Package for Fast Non-Parametric Confidence Intervals for Equal Error Rates in Amortized O(m log n)
- Ali Khodabakhsh, Erwin Haasnoot, Patrick Bours (2018). Predicted Templates: Learning-curve Based Template Projection for Keystroke Dynamics
- Ali Khodabakhsh, Raghavendra Ramachandra, Christoph Busch (2018). A Taxonomy of Audiovisual Fake Multimedia Content Creation Technology
- Ali Khodabakhsh, Raghavendra Ramachandra, Kiran B. Raja, Pankaj Shivdayal Wasnik, Christoph Busch (2018). Fake Face Detection Methods: Can They Be Generalized?
- COINS Finse winter school, Finse, Norway, 2019
- COINS summer school, Metochi, Greece, 2018
- COINS Ph.D student seminar, Longyearbyen, Svalbard, Norway, 2018
- NISK 2018, Longyearbyen, Svalbard, Norway, 2018