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.

In this article, we propose Multi-Entity Bayesian Networks (MEBNs) as the probabilistic ontological framework for the analysis of the Tsamiko and Salsa dances. More specifically, our analysis has the objective of the dancer assessment with respect to both choreography execution accuracy and the synchronization of the dance movements with the musical rhythm. For this task, we make use of the explicit, expert-provided knowledge on dance movements and their relations to the musical beat. Due to the complexity of this knowledge, the MEBNs were used as the probabilistic ontological framework in which the knowledge is formalized. The reason we opt for MEBNs for this task is that they combine Bayesian and formal (first-order) logic into a single model. In this way, the Bayesian probabilistic part of MEBNs was used to capture, using example data and training, the implicit part of the expert knowledge about dances, i.e., this part of the knowledge that cannot be formalized and explicitly defined accurately enough, while the logical maintains the explicit knowledge representation in the same way ontologies do. Moreover, we present in detail the MEBN models we built for Tsamiko and Salsa, using expert-provided explicit knowledge. Last, we conduct experiments that demonstrate the effectiveness of the proposed MEBN-based methodology we employ to achieve our analysis objectives. The results of the experiments demonstrate the superiority of MEBNs to conventional models, such as BNs, in terms of the dancer assessment accuracy.

This survey article is concerned with the emergence of vision augmentation AI tools for enhancing the situational awareness of first responders (FRs) in rescue operations. More specifically, the article surveys three families of image restoration methods serving the purpose of vision augmentation under adverse weather conditions. These image restoration methods are: (a) deraining; (b) desnowing; (c) dehazing ones. The contribution of this article is a survey of the recent literature on these three problem families, focusing on the utilization of deep learning (DL) models and meeting the requirements of their application in rescue operations. A faceted taxonomy is introduced in past and recent literature including various DL architectures, loss functions and datasets. Although there are multiple surveys on recovering images degraded by natural phenomena, the literature lacks a comprehensive survey focused explicitly on assisting FRs. This paper aims to fill this gap by presenting existing methods in the literature, assessing their suitability for FR applications, and providing insights for future research directions.

Abstract. In this work we present InFeRno, an intelligent web pornography elimination system, classifying web pages based solely on their visual content. The main characteristics of our system include: (i) a powerful vector space with a small but sufficient number of features that manage to improve the discriminative ability of the SVM classifier; (ii) an extra class (bikini) that strengthens the performance of the classifier; (iii) an overall classification scheme that achieves high accuracy at considerably lower runtime costs compared to current state-of-the-art systems; and (iv) a full-fledged implementation of the proposed system capable of being integrated with ICAP-aware web proxy cache servers.