TU Darmstadt / ULB / TUbiblio

Unsupervised Face Recognition using Unlabeled Synthetic Data

Boutros, Fadi ; Klemt, Marcel ; Fang, Meiling ; Kuijper, Arjan ; Damer, Naser (2023)
Unsupervised Face Recognition using Unlabeled Synthetic Data.
17th International Conference on Automatic Face and Gesture Recognition. Waikoloa Beach, USA (05.-08.01.2023)
doi: 10.1109/FG57933.2023.10042627
Konferenzveröffentlichung, Bibliographie

Kurzbeschreibung (Abstract)

Over the past years, the main research innovations in face recognition focused on training deep neural networks on large-scale identity-labeled datasets using variations of multi-class classification losses. However, many of these datasets are retreated by their creators due to increased privacy and ethical concerns. Very recently, privacy-friendly synthetic data has been proposed as an alternative to privacy-sensitive authentic data to comply with privacy regulations and to ensure the continuity of face recognition research. In this paper, we propose an unsupervised face recognition model based on unlabeled synthetic data (USynthFace). Our proposed USynthFace learns to maximize the similarity between two augmented images of the same synthetic instance. We enable this by a large set of geometric and color transformations in addition to GAN-based augmentation that contributes to the USynthFace model training. We also conduct numerous empirical studies on different components of our USynthFace. With the proposed set of augmentation operations, we proved the effectiveness of our USynthFace in achieving relatively high recognition accuracies using unlabeled synthetic data. The training code and pretrained model are publicly available under https://github.com/fdbtrs/Unsupervised-Face-Recognition-using-Unlabeled-Synthetic-Data.

Typ des Eintrags: Konferenzveröffentlichung
Erschienen: 2023
Autor(en): Boutros, Fadi ; Klemt, Marcel ; Fang, Meiling ; Kuijper, Arjan ; Damer, Naser
Art des Eintrags: Bibliographie
Titel: Unsupervised Face Recognition using Unlabeled Synthetic Data
Sprache: Englisch
Publikationsjahr: 16 Februar 2023
Verlag: IEEE
Buchtitel: 2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)
Veranstaltungstitel: 17th International Conference on Automatic Face and Gesture Recognition
Veranstaltungsort: Waikoloa Beach, USA
Veranstaltungsdatum: 05.-08.01.2023
DOI: 10.1109/FG57933.2023.10042627
Kurzbeschreibung (Abstract):

Over the past years, the main research innovations in face recognition focused on training deep neural networks on large-scale identity-labeled datasets using variations of multi-class classification losses. However, many of these datasets are retreated by their creators due to increased privacy and ethical concerns. Very recently, privacy-friendly synthetic data has been proposed as an alternative to privacy-sensitive authentic data to comply with privacy regulations and to ensure the continuity of face recognition research. In this paper, we propose an unsupervised face recognition model based on unlabeled synthetic data (USynthFace). Our proposed USynthFace learns to maximize the similarity between two augmented images of the same synthetic instance. We enable this by a large set of geometric and color transformations in addition to GAN-based augmentation that contributes to the USynthFace model training. We also conduct numerous empirical studies on different components of our USynthFace. With the proposed set of augmentation operations, we proved the effectiveness of our USynthFace in achieving relatively high recognition accuracies using unlabeled synthetic data. The training code and pretrained model are publicly available under https://github.com/fdbtrs/Unsupervised-Face-Recognition-using-Unlabeled-Synthetic-Data.

Freie Schlagworte: Biometrics, Face recognition, Privacy enhancing technologies, Computer vision, Deep learning
Fachbereich(e)/-gebiet(e): 20 Fachbereich Informatik
20 Fachbereich Informatik > Graphisch-Interaktive Systeme
20 Fachbereich Informatik > Mathematisches und angewandtes Visual Computing
Hinterlegungsdatum: 15 Mär 2023 08:02
Letzte Änderung: 04 Jul 2023 11:26
PPN: 50927692X
Export:
Suche nach Titel in: TUfind oder in Google
Frage zum Eintrag Frage zum Eintrag

Optionen (nur für Redakteure)
Redaktionelle Details anzeigen Redaktionelle Details anzeigen