The impacts of plastic surgery on face recognition systems have been investigated in the past decade by many researchers. Diverse well-known face recognition approaches, e.g. based on PCA or LBP, have been bench-marked mostly on the web-collected IIITD plastic surgery face database. Generally, significant performance drops were reported when comparing facial images taken before and after plastic surgeries. On the one side, some researchers reported problems with said plastic surgery database, i.e. the presence of low quality images. On the other side, the applied methods no longer reflect the state-of-the-art in face recognition. This calls for evaluating the impact of plastic surgery on state-of-the-art deep face recognition systems anew considering high quality imagery of most relevant plastic surgeries.This work introduces the new Hochschule Darmstadt (HDA) plastic surgery database of facial images taken before and after surgery. This database vastly complies with the quality requirements defined by the International Civil Aviation Organization (ICAO) for electronic travel documents and comprises face images of the five most frequently applied facial plastic surgeries. The HDA plastic surgery database, the IIITD plastic surgery database, and a non-surgery database, i.e. ICAO-compliant subsets of the FRGCv2 and FERET datasets, are used for comparative verification and identification evaluations which are conducted using the commercial Cognitec FaceVACS system and the open-source ArcFace system. The obtained results suggest that the impact of plastic surgery on deep face recognition systems is less significant than that observed for previously benchmarked methods.

Plastic surgery: An obstacle for deep face recognition? / Rathgeb, C.; Dogan, D.; Stockhardt, F.; De Marsico, M.; Busch, C.. - 2020-:(2020), pp. 3510-3517. (Intervento presentato al convegno 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2020 tenutosi a usa) [10.1109/CVPRW50498.2020.00411].

Plastic surgery: An obstacle for deep face recognition?

De Marsico M.;
2020

Abstract

The impacts of plastic surgery on face recognition systems have been investigated in the past decade by many researchers. Diverse well-known face recognition approaches, e.g. based on PCA or LBP, have been bench-marked mostly on the web-collected IIITD plastic surgery face database. Generally, significant performance drops were reported when comparing facial images taken before and after plastic surgeries. On the one side, some researchers reported problems with said plastic surgery database, i.e. the presence of low quality images. On the other side, the applied methods no longer reflect the state-of-the-art in face recognition. This calls for evaluating the impact of plastic surgery on state-of-the-art deep face recognition systems anew considering high quality imagery of most relevant plastic surgeries.This work introduces the new Hochschule Darmstadt (HDA) plastic surgery database of facial images taken before and after surgery. This database vastly complies with the quality requirements defined by the International Civil Aviation Organization (ICAO) for electronic travel documents and comprises face images of the five most frequently applied facial plastic surgeries. The HDA plastic surgery database, the IIITD plastic surgery database, and a non-surgery database, i.e. ICAO-compliant subsets of the FRGCv2 and FERET datasets, are used for comparative verification and identification evaluations which are conducted using the commercial Cognitec FaceVACS system and the open-source ArcFace system. The obtained results suggest that the impact of plastic surgery on deep face recognition systems is less significant than that observed for previously benchmarked methods.
2020
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2020
Biometric s; face recognition; plastic surgery
04 Pubblicazione in atti di convegno::04b Atto di convegno in volume
Plastic surgery: An obstacle for deep face recognition? / Rathgeb, C.; Dogan, D.; Stockhardt, F.; De Marsico, M.; Busch, C.. - 2020-:(2020), pp. 3510-3517. (Intervento presentato al convegno 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPRW 2020 tenutosi a usa) [10.1109/CVPRW50498.2020.00411].
File allegati a questo prodotto
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1466622
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 8
  • ???jsp.display-item.citation.isi??? 5
social impact