Chaabane Djeraba

High performance crawling system

In the present paper, we will describe the design and implementation of a real-time distributed system of Web crawling running on a cluster of machines. The system crawls several thousands of pages every second, includes a high-performance fault manager, is platform independent and is able to adapt transparently to a wide range of configurations without incurring additional hardware expenditure. We will then provide details of the system architecture and describe the technical choices for very high performance crawling. Finally, we will discuss the experimental results obtained, comparing them with other documented systems

A framework for mining meaningful usage patterns within a semantically enhanced web portal

Semantic Web (SW) is a new trend in the evolution of the current Web aimed at extending its basic functionalities by providing computer-readable semantic meta-data about the Web content. The meta-data is typically organized into a domain ontology where key concepts and relations from the domain appear. The benefits of such a representation are manifold: a more topical information seeking process, better content adaptation and higher interoperability even on the current, still largely syntactical, Web, to name only a few. As the SW is, arguably, the future of the Web, it is only too natural that Web mining, i.e., the application of data mining techniques to web-related data, tackles the processing semantically annotated data. In this context, we study the detecting of typical navigation scenarios on an ontology-powered Web portal, i.e., an instance of usage mining on the SW. In the present paper, we tackle the fundamental aspects of the underlying mining problem and clarify the impact a fully-fledged ontology has on the data and pattern languages. Indeed, current ontology-aware mining approaches tend to limit their scope to the core conceptual hierarchy (taxonomy) of an ontology whereas in a realistic settings there will be a lot more knowledge in the ontology, in particular, on semantic relations between domain concepts, the way they instantiate into links between content objects, etc. We show that reflecting domain relations in the navigation patterns results in a new pattern structure that combines elements from sequential, generalized and graph pattern mining and therefore requires a dedicated mining strategy. After characterizing the underlying pattern space, we describe a dedicated level-wise mining method and present some empirical evidence of its viability.

Visual Gaze Estimation by Joint Head and Eye Information

In this paper, we present an unconstrained visual gaze estimation system. The proposed method extracts the visual field of view of a person looking at a target scene in order to estimate the approximate location of interest (visual gaze). The novelty of the system is the joint use of head pose and eye location information to fine tune the visual gaze estimated by the head pose only, so that the system can be used in multiple scenarios. The improvements obtained by the proposed approach are validated using the Boston University head pose dataset, on which the standard deviation of the joint visual gaze estimation improved by 61:06% horizontally and 52:23% vertically with respect to the gaze estimation obtained by the head pose only. A user study shows the potential of the proposed system.

DLBP: A novel descriptor for depth image based face recognition

This paper presents a novel descriptor for face depth images, generalizing the well-known Local Binary Pattern (LBP), in order to enhance its discriminative power for smooth depth images. The proposed descriptor is based on detecting shape patterns from face surfaces and enables accurate and fast description of shape variation in depth images. It is in the same form as conventional LBP, so patterns can be readily combined to form joint histograms to represent depth faces. The descriptor is computationally very simple, rapid and it is totally training-free. When we associate our descriptor in a face recognition scheme based on nearest neighbor classifier, it shows its discriminative power in depth based face recognition comparing to the conventional LBP and other extensions proposed for 3D face recognition. Many experiments are conducted on different databases in order to evaluate our method.

Rapid and accurate face depth estimation in passive stereo systems

In this paper, we introduce a novel approach for face depth estimation in a passive stereo vision system. Our approach is based on rapid generation of facial disparity maps, requiring neither expensive devices nor generic face models. It consists in incorporating face properties into the disparity estimation process to enhance the 3D face reconstruction. We propose a model-based method that is independent from the specific stereo algorithm used. Our method is a two-step process. First, an algorithm based on the Active Shape Model (ASM) is proposed to acquire a disparity model specific to the face concerned. Second, using this model as a guidance, the dense disparity is calculated and the depth map is estimated. Besides, an original post-processing algorithm is proposed in order to detect holes and spikes in the generated depth maps caused by wrong matches and uncertainties. It is based on the smoothness property of the face and a local and global analysis of the image. Experimental results are presented to demonstrate the reconstruction accuracy and the speed of the proposed method.

Multi-Modal User Interactions in Controlled Environments

Multi-Modal User Interactions in Controlled Environments investigates the capture and analysis of users multimodal behavior (mainly eye gaze, eye fixation, eye blink and body movements) within a real controlled environment (controlled-supermarket, personal environment) in order to adapt the response of the computer/environment to the user. Such data is captured using non-intrusive sensors (for example, cameras in the stands of a supermarket) installed in the environment. This multi-modal video based behavioral data will be analyzed to infer user intentions while assisting users in their day-to-day tasks by adapting the systems response to their requirements seamlessly. This book also focuses on the presentation of information to the user. Multi-Modal User Interactions in Controlled Environments is designed for professionals in industry, including professionals in the domains of security and interactive web television. This book is also suitable for graduate-level students in computer science and electrical engineering.

A Local Approach for Negative Emotion Detection

Recognizing human facial expression and emotion by computer is an interesting and challenging problem. In this paper, we propose a method for recognizing negative emotions through an appropriate representation of facial features from relevant face regions displayed in video streams and still images. A measure that is sensitive to facial movements is used in predefined regions of interest to detect the negative emotions. The experimentation has been performed on a standard dataset and live video streams and has showed promising results.

3D face reconstruction in a binocular passive stereoscopic system using face properties

In this paper, we introduce a novel approach for face stereo reconstruction in passive stereo vision system. Our approach is based on the generation of a facial disparity map, requiring neither expensive devices nor generic face models. It consists of incorporating face properties in the disparity estimation to enhance the 3D face reconstruction. An algorithm based on the Active Shape Model (ASM) is proposed to acquire 3D sparse estimation of the face with a high confidence. Using sparse estimation as guidance and considering the face symmetry and smoothness, the dense disparity is completed. Experimental results are presented to demonstrate the reconstruction accuracy of the proposed method.

Abnormal Event Detection

This chapter presents a system that detects abnormal events extracted from videos in a crowded environment. At this stage, our intention is to stick to simple methods so as to enhance the real-time requirements of the processing. The efforts made and results achieved will lay the groundwork and serve as a benchmark for future work. The selected approach consists of extracting some portions of videos coinciding with sudden changes and abnormal motion variations in the points of interest. It performs calculations on information such as the density, direction and velocity of motion, and classifies the content as normal or abnormal. This approach will help to index abnormal events and will offer users the opportunity of making queries about content. Our approach has been tested on various videos in real-world conditions, namely incidents in airport escalator exits.

Estimation of Visual Gaze

Studying gaze direction is useful for the completion of visual tasks. It is well known that the visual gaze is a product of two contributing factors: head pose and eye location. In general, the visual gaze can be determined from head pose when standing at some distance from the visual target [76]. Starting from this general statement, the present chapter will first detail the human visual system and the steps that are necessary to generate the gaze. An account of the completed works on gaze tracking will then be presented along with selected techniques according to the different types of systems (intrusive or non-intrusive). A few applications will be introduced afterwards to illustrate how data collected from the gaze are processed. Finally, we will examine how the head pose contributes to visual gaze estimation, and also how to estimate the gaze based on eye position only.