Over the past two decades, the nature of child pornography in terms of generation, distribution and possession of images drastically changed, evolving from basically covert and offline exchanges of content to a massive network of contacts and data sharing. Nowadays, the internet has become not only a transmission channel but, probably, a child pornography enabling factor by itself.
The use of deep convolution neural networks for sexually exploitative imagery of children (SEIC) is challenging, since those models require large amounts of training data. To bypass that problem, in this paper, the authors proposed a data-driven solution in which: (1) transfer the network parameters and configurations trained on ImageNet to the target problem detection (1-tiered solution) and (2) perform a 2-tiered transfer learning procedure, in which knowledge is transferred from the a network trained over ImageNet to the problem of adult content detection and fine-tune the network for detecting child pornography content in an image problem.
The proposed method outperform different existing solutions and seem to represent an important step forward when dealing with child pornography content detection. The solutions are encapsulated in a sandbox virtual machine ready for deployment by experts and practitioners.