Objective: In cases of suspected child pornography, the age of the victim represents a crucial factor for legal prosecution. The conventional methods for age estimation provide unreliable age estimates, particularly if teenage victims are concerned. In this pilot study, the potential of age estimation for screening purposes is explored for juvenile faces. In addition to a visual approach, an automated procedure is introduced, which has the ability to rapidly scan through large numbers of suspicious image data in order to trace juvenile faces. Methods: Age estimations were performed by experts, non-experts and the Demonstrator of a developed software on frontal facial images of 50 females aged 10-19 years from Germany, Italy, and Lithuania. To test the accuracy, the mean absolute error (MAE) between the estimates and the real ages was calculated for each examiner and the Demonstrator. Results: The Demonstrator achieved the lowest MAE (1.47 years) for the 50 test images. Decreased image quality had no significant impact on the performance and classification results. The experts delivered slightly less accurate MAE (1.63 years). Throughout the tested age range, both the manual and the automated approach led to reliable age estimates within the limits of natural biological variability. Conclusions: The visual analysis of the face produces reasonably accurate age estimates up to the age of 18 years, which is the legally relevant age threshold for victims in cases of pedo-pornography. This approach can be applied in conjunction with the conventional methods for a preliminary age estimation of juveniles depicted on images.

The juvenile face as a suitable age indicator in child pornography cases : a pilot study on the reliability of automated and visual estimation approaches / M. Ratnayake, Z. Obertová, M. Dose, P. Gabriel, H.M. Bröker, M. Brauckmann, A. Barkus, R. Rizgeliene, J. Tutkuviene, S. Ritz Timme, L. Marasciuolo, D. Gibelli, C. Cattaneo. - In: INTERNATIONAL JOURNAL OF LEGAL MEDICINE. - ISSN 0937-9827. - 128:5(2014 Sep), pp. 803-808. [10.1007/s00414-013-0875-y]

The juvenile face as a suitable age indicator in child pornography cases : a pilot study on the reliability of automated and visual estimation approaches

L. Marasciuolo;D. Gibelli
Penultimo
;
C. Cattaneo
Ultimo
2014

Abstract

Objective: In cases of suspected child pornography, the age of the victim represents a crucial factor for legal prosecution. The conventional methods for age estimation provide unreliable age estimates, particularly if teenage victims are concerned. In this pilot study, the potential of age estimation for screening purposes is explored for juvenile faces. In addition to a visual approach, an automated procedure is introduced, which has the ability to rapidly scan through large numbers of suspicious image data in order to trace juvenile faces. Methods: Age estimations were performed by experts, non-experts and the Demonstrator of a developed software on frontal facial images of 50 females aged 10-19 years from Germany, Italy, and Lithuania. To test the accuracy, the mean absolute error (MAE) between the estimates and the real ages was calculated for each examiner and the Demonstrator. Results: The Demonstrator achieved the lowest MAE (1.47 years) for the 50 test images. Decreased image quality had no significant impact on the performance and classification results. The experts delivered slightly less accurate MAE (1.63 years). Throughout the tested age range, both the manual and the automated approach led to reliable age estimates within the limits of natural biological variability. Conclusions: The visual analysis of the face produces reasonably accurate age estimates up to the age of 18 years, which is the legally relevant age threshold for victims in cases of pedo-pornography. This approach can be applied in conjunction with the conventional methods for a preliminary age estimation of juveniles depicted on images.
age estimation; automated age estimation; child pornography; juvenile pornography
Settore MED/43 - Medicina Legale
2-giu-2013
Article (author)
File in questo prodotto:
File Dimensione Formato  
art_10.1007_s00414-013-0875-y.pdf

accesso riservato

Tipologia: Publisher's version/PDF
Dimensione 143.17 kB
Formato Adobe PDF
143.17 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

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/2434/220706
Citazioni
  • ???jsp.display-item.citation.pmc??? 3
  • Scopus 29
  • ???jsp.display-item.citation.isi??? 24
social impact