University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Evaluating the Efficacy of a New Synthetic AI-Generated Dataset for Training Face Recognition Models
Elhabashy, Adham (2025) Evaluating the Efficacy of a New Synthetic AI-Generated Dataset for Training Face Recognition Models.
|
PDF
4MB |
| Abstract: | The increasing demand for large-scale facial datasets to train deep learning-based face recognition (FR) systems has raised critical concerns regarding privacy, consent, and data collection ethics. Synthetic face datasets, particularly those generated by diffusion models have proven to be a promising solution by offering diversity, scalability, and identity control without relying on real individuals. This study investigates the effectiveness of FLUXSynID—a high-resolution, diffusion-based document-style synthetic dataset for training face recognition models. We conducted a comprehensive evaluation under three experimental scenarios: full-data training, sequential learning in data-scarce settings, and hybrid training with mixed real and synthetic data. The results show that models trained on synthetic faces can match or exceed the performance of models trained on real data, particularly in high-security verification tasks and expressive test conditions. Moreover, strategically combining synthetic and real data improves generalization and bridges performance gaps caused by data imbalance. These findings highlight the viability of synthetic data as both a privacy-preserving alternative and a valuable complement to real-world datasets for modern face recognition. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/107383 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page