Detecting Pornographic Images by Localizing Skin ROIs

In this study, a novel algorithm for recognizing pornographic images based on the analysis of skin color regions is presented. The skin color information essentially provides Regions of Interest (ROIs). It is demonstrated that the convex hull of these ROIs provides semantically useful information for pornographic image detection. Based on these convex hulls, the authors extract a small set of low-level visual features that are empirically proven to possess discriminative power for pornographic image classification. In this study, the authors consider multi-class pornographic image classification, where the “nude” and “benign” image classes are further split into two specialized sub-classes, namely “bikini”/”porn” and “skin”/”non-skin”, respectively. The extracted feature vectors are fed to an ensemble of random forest classifiers for image classification. Each classifier is trained on a partition of the training set and solves a binary classification problem. In this sense, the model allows for seamless coarse-to-fine-grained classification by means of a tree-structured topology of a small number of intervening binary classifiers. The overall technique is evaluated on the AIIA-PID challenge of 9,000 samples of pornographic and benign images. The technique is shown to exhibit state-of-the-art performance against publicly available integrated pornographic image classifiers.