On March 5th Marcel Grimmer defended his PhD thesis on "Deep Generative Networks in Face Recognition - With Application to Face Image Quality Assessment". Congratulations to the excellent work.

From the abstract:

The widespread adoption of face recognition systems for authentication spans from convenience-based applications (e.g., smartphone unlocking) to security-critical applications (e.g., border control). This trend stems from the transition from traditional machine learning to deep learning-based recognition models, which have significantly improved recognition accuracy and robustness against various sources of variation. However, the effectiveness of deep neural networks largely depends on the availability of sufficient training and evaluation data. One possible solution to meet the high demand for data is to exploit recent advances in deep generative models, which can generate highly realistic facial images that are indistinguishable from images upon human inspection. Therefore, this thesis examines how large-scale data collection can be circumvented by using deep generative models for evaluating face recognition systems and deriving quality assessment measures. The specific contributions of this thesis can be summarised as follows: - An investigation into how deep generative models can be used to create synthetic mated samples by editing facial attributes while preserving identity. - A survey on deep learning-based face age editing techniques, relevant datasets, evaluation strategies, and open challenges. - Development of a diffusion-based face age editing technique that outperforms current state-of-the-art methods in identity preservation, photorealism, and age pattern diversity. - Generation of synthetic evaluation datasets to estimate task-specific metrics, such as equal error rates or average similarity decreases. - Development of quality component measures according to ISO/IEC 29794- 5, for predicting a sample’s face recognition utility with respect to individual quality elements, facilitated by deep generative models.

You can find the full thesis at: 
https://ntnuopen.ntnu.no/ntnu-xmlui/handle/11250/3179398

One of the valuable contributions of this thesis is the neutral expression quality component, which was adopted by ISO/IEC 29794-5 and is contained in the reference implementation Open Source Face Image Quality (OFIQ): https://github.com/BSI-OFIQ/OFIQ-Project.

(Photo by NTNU. From left to right: Patrick Bours, Alice O'Toole, Marcel Grimmer, Christoph Busch, Adam Herout)

Assessment committee:

1st opponent: Professor Alice O'Toole, The University of Texas at Dallas, USA.

2nd opponent: Professor Adam Herout, Brno University of Technology, Czech Republic

Internal member: Professor Patrick Bours, NTNU, Norway

Congratulations!!!