On the detection of images containing child-pornographic material

Abstract

The vast increase in the use of social networks and other internet-based communication tools contributed to the escalation of the problem of exchanging child pornographic material over the internet. The problem of dissemination of child pornographic material could be addressed using dedicated image detection algorithms capable of rating the inappropriateness level of images exchanged through computer networks so that images with inappropriate content involving children are blocked. However, the complexity of the image detection task coupled with the nonexistence of suitable datasets, inhibit the development of efficient algorithms that can be used for detecting offensive images containing children. To deal with the problem, we propose a methodological approach that can be used for supporting the development of child pornography detectors through the generation of synthetic datasets and through the decomposition of the task into a set of simpler tasks for which training data is available. Preliminary results show the promise of the proposed approach.

Metadata-Based Detection of Child Sexual Abuse Material

“…Multiple approaches to the CSAM identification problem using computer vision-based methods have appeared in the literature in the past years [28], [16], [32]. Although these approaches appear promising, they lack evaluation in realworld big data unbalanced data sets; data sets with ethnicity diversity, and time efficiency and scalability evaluation.…”

“…Since PhotoDNA's first development, computer vision models have undergone a revolution resulting in novel machine learning based models for pornography and CSAM detection [19], [16], [32], [22]. The current approaches either combine a computer vision model to extract image descriptors [28], train computer vision models on pornography data [9], perform a combination of age estimation and pornography detection [16] or synthetic data [32].…”

“…Since PhotoDNA's first development, computer vision models have undergone a revolution resulting in novel machine learning based models for pornography and CSAM detection [19], [16], [32], [22]. The current approaches either combine a computer vision model to extract image descriptors [28], train computer vision models on pornography data [9], perform a combination of age estimation and pornography detection [16] or synthetic data [32]. However, due to legal restrictions in maintaining a database of CSAM images, all current works are based on either unrealistic images [32], or validated by authorities in very small [28], [16], [9] data sets that hardly represents the true data distribution in the internet [6].…”

Mitigando os Efeitos de GAN em Classificação de Imagens com CNN

“…Por sua vez, [Yiallourou et al 2017] produziram 88 imagens PI baseando-se em questionários. Posteriormente, treinam um modelo de regressão linear com estas imagens, sendo que obtiveram ACC = 76,14%.…”

“…No primeiro, 2 peritos desenvolveram o software NuDetective [Polastro e Eleuterio 2010], similar ao trabalho de [Sae-Bae et al 2014]. No segundo, apresentaram o uso de método de regressão linear [Yiallourou et al 2017]. No terceiro, propõem-se uma abordagem a partir da extração das características discriminadoras de imagens, para no final ser empregado o classificador SVM [Vitorino et al 2016], assim como [Ulges e Stahl 2011], que também utilizam descritor e SVM.…”