Search

Proceedings of the Special Session on Multimodal Security and Surveillance Analytics 2014, held during the International Conference on Signal Processing and Multimedia Applications (SIGMAP 2014) in Vienna, People detection is a task that has generated a great interest in the computer vision and specially in the surveillance community. One of the main problems of this task in crowded scenarios is the high number of occlusions deriving from persons appearing in groups. In this paper, we address this problem by combining individual body part detectors in a statistical driven way in order to be able to detect persons even in case of failure of any detection of the body parts, i.e., we propose a generic scheme to deal with partial occlusions. We demonstrate the validity of our approach and compare it with other state of the art approaches on several public datasets. In our experiments we consider sequences with different complexities in terms of occupation and therefore with different number of people present in the scene, in order to highlight the benefits and difficulties of the approaches considered for evaluation. The results show that our approach improves the results provided by state of the art approaches specially in the case of crowded scenes, This work has been done while visiting the Communication Systems Group at the Technische Universität Berlin (Germany) under the supervision of Prof. Dr.-Ing. Thomas Sikora. This work has been partially supported by the Universidad Aut´onoma de Madrid (“Programa propio de ayudas para estancias breves en España y extranjero para Personal Docente e Investigador en Formación de la UAM”), by the Spanish Government (TEC2011-25995 EventVideo) and by the European Community’s FP7 under grant agreement number 261776 (MOSAIC).

Submitted by Johnny Rodrigues (johnnyrodrigues@ufcg.edu.br) on 2018-04-18T15:01:39Z No. of bitstreams: 1 MARCELL MANFRIN BARBACENA - DISSERTAÇÃO PPGCC 2014..pdf: 1468565 bytes, checksum: b94d20ffdace21ece654986ffd8fbb63 (MD5) Made available in DSpace on 2018-04-18T15:01:39Z (GMT). No. of bitstreams: 1 MARCELL MANFRIN BARBACENA - DISSERTAÇÃO PPGCC 2014..pdf: 1468565 bytes, checksum: b94d20ffdace21ece654986ffd8fbb63 (MD5) Previous issue date: 2014 Impulsionadas pela crescente demanda por sistemas de segurança para proteção do indivíduo e da propriedade nos dias atuais, várias pesquisas têm sido desenvolvidas com foco na implantação de sistemas de vigilância por vídeo com ampla cobertura. Um dos problemas de pesquisa em aberto nas áreas de visão computacional e redes de computadores envolvem a escalabilidade desses sistemas, principalmente devido ao aumento do número de câmeras transmitindo vídeos em tempo real para monitoramento e processamento. Neste contexto, o objetivo geral deste trabalho é avaliar o impacto que a redução da taxa de transmissão dos fluxos de vídeos impõe na eficácia dos algoritmos de detecção de pessoas utilizados em sistemas inteligentes de videovigilância. Foram realizados experimentos utilizando vídeos em alta resolução no contexto de vigilância com tomadas externas e com um algoritmo de detecção de pessoas baseado em histogramas de gradientes orientados, nos quais se coletou, como medida de eficácia do algoritmo, a métrica de área sob a curva de precisão e revocação para, em sequência, serem aplicados os testes estatísticos de Friedman e de comparações múltiplas com um controle na aferição das hipóteses levantadas. Os resultados obtidos indicaram que é possível uma redução da taxa de transmissão em mais de 70% sem que haja redução da eficácia do algoritmo de detecção de pessoas. Motivated by the growing demand for security systems to protect persons and properties in the nowadays, several researches have been developed focusing on the deployment of widearea video coverage surveillance systems. One open research problem in the areas of computer vision and computer networks involves the scalability of these systems, mainly due to the increasing number of cameras transmitting real-time video for monitoring and processing. In this context, the aim of this study was to evaluate the impact that transmission data-rate reduction of video streams imposes on the effectiveness of people detection algorithms used in intelligent video surveillance systems. With a proposed experimental design, experiments were performed using high-resolution wide-area external coverage video surveillance and using an algorithm for people detection based on histograms of oriented gradients. As a measure of effectiveness of the people detection algorithm, the metric of area under the precision-recall curve was collected and statistical tests of Friedman and multiple comparisons with a control were applied to evaluate the hypotheses. The results indicated that it is possible to reduce transmission rate by more than 70% without decrease in the effectiveness of the people detection algorithm.

This thesis is performed in industrial context and presents a whole framework for people detection and tracking in a camera network. It addresses the main process steps: people detection, people tracking in mono-camera context, and people re-identification in multi-camera context. High performances and real-time processing are considered as strong constraints. People detection aims to localise and delimits people in video sequences. The proposed people detection is performed using a cascade of classifiers trained using LogitBoost algorithm on region covariance descriptors. A state of the art approach is strongly optimized to process in real time and to provide better detection performances. The optimization scheme is generalizable to many other kind of detectors where all possible weak classifiers cannot be reasonably tested. People tracking in mono-camera context aims to provide a set of reliable images of every observed person by each camera, to extract his visual signature, and it provides some useful real world information for re-identification purpose. It is achieved by tracking SIFT features using a specific particle filter in addition to a data association framework which infer object tracking from SIFT points one, and which deals with most of possible cases, especially occlusions. Finally, people re-identification is performed using an appearance based approach by improving a state of the art approach, providing better performances while keeping the real-time processing advantage. A context-aware part is introduced to robustify the visual signature against people orientations, ensuring better re-identification performances in real application case.; Cette thèse CIFRE est effectuée dans un contexte industriel et présente un framework complet pour la détection, le suivi mono-caméra et de la ré-identification de personnes dans le contexte multi-caméras. Les performances élevés et le traitement en temps réel sont les deux contraintes critiques ayant guidé ce travail. La détection de personnes vise à localiser/délimiter les gens dans les séquences vidéo. Le détecteur proposé est basé sur une cascade de classifieurs de type LogitBoost appliqué sur des descripteurs de covariances. Une approche existante a fortement été optimisée, la rendant applicable en temps réel et fournissant de meilleures performances. La méthode d'optimisation est généralisable à d'autres types de détecteurs d'objets. Le suivi mono-caméra vise à fournir un ensemble d'images de chaque personne observée par chaque caméra afin d'extraire sa signature visuelle, ainsi qu'à fournir certaines informations du monde réel pour l'amélioration de la ré-identification. Ceci est réalisé par le suivi de points SIFT à l'aide d'une filtre à particules, ainsi qu'une méthode d'association de données qui infère le suivi des objets et qui gère la majorité des cas de figures possible, notamment les occultations. Enfin, la ré-identification de personnes est réalisée avec une approche basée sur l'apparence globale en améliorant grandement une approche existante, obtenant de meilleures performances tout en étabt applicable en temps réel. Une partie "conscience du contexte" est introduite afin de gérer le changement d'orientation des personnes, améliorant les performances dans le cas d'applications réelles.

International audience; We introduce a generative model for learning person and costume specific detectors from labeled examples. We demonstrate the model on the task of localizing and naming actors in long video sequences. More specifically, the actor's head and shoulders are each represented as a constellation of optional color regions. Detection can proceed despite changes in view-point and partial occlusions. We explain how to learn the models from a small number of labeled keyframes or video tracks, and how to detect novel appearances of the actors in a maximum likelihood framework. We present results on a challenging movie example, with 81% recall in actor detection (coverage) and 89% precision in actor identification (naming).; Nous présentons un modèle génératif pemettant l'apprentissage de détecteurs de personnes et de costumes à partir d'exemples. Nous appliquons notre modèle au problème de la détection, de la localisation et de l'identification d'acteurs dans de longues séquences vidéo. Nous représentons la tete et les épaules de chaque acteur comme une constellation de régions de couleurs. Toutes les régions sont facultatives, ce qui nous permet de rendre la méthode robuste aux changements de points de vues et aux occultations partielles. Nous décrivons comment le modèle peut être appris à partir d'un petit nombre d'exemple, et décrivons un algorithm rapide de détection. Notre méthode permet de detecter et reconnaitre les 8 acteurs du film "La corde" d'Alfred Hitchcock dans 81 % des cas, avec une précision de 89 %.

[ES] El objetivo de esta tesina es realizar un estudio de algoritmos y paquetes software existentes en la plataforma ROS (Robot Operating System) para la detección de personas, y comprobar el funcionamiento de alguno de ellos. El estudio se centrará principalmente en algoritmos que se basen en visión artificial y sensores de rango para sistemas de tipo ADAS (Advanced Driving Assistance Systems), en los que la detección de personas se utiliza para evitar accidentes. Se ha probado uno de los detectores existentes (Histrogram of Oriented Gradients o HOG) y se ha combinado con una técnica de detección de objetos verticales, reduciendo el tiempo de ejecución, [EN] In this project a survey of people detection algorithms and software in ROS framework is made. The survey is focused on algorithms using computer vision and range sensors for Advanced Driving Assistance Systems (ADAS), in wich people detection is used to avoid traffic accidents. One of the state-of-the-art methods (Histogram of Oriented Gradients or HOG) has been tested separatedly and with a vertical object detection algorithm, reducing the execution time

We present a wireless sensor network (WSN) for monitoring indoor air quality, which is crucial for people’s comfort, health, and safety because they spend a large percentage of time in indoor environments. A major concern in such networks is energy efficiency because gas sensors are power-hungry, and the sensor node must operate unattended for several years on a battery power supply. A system with aggressive energy management at the sensor level, node level, and network level is presented. The node is designed with very low sleep current consumption (only 8 μA), and it contains a metal oxide semiconductor gas sensor and a pyroelectric infrared (PIR) sensor. Furthermore, the network is multimodal ; it exploits information from auxiliary sensors, such as PIR sensors about the presence of people and from the neighbor nodes about gas concentration to modify the behavior of the node and the measuring frequency of the gas concentration. In this way, we reduce the nodes’ activity and energy requirements, while simultaneously providing a reliable service. To evaluate our approach and the benefits of the context-aware adaptive sampling, we simulate an application scenario which demonstrates a significant lifetime extension (several years) compared to the continuously-driven gas sensor. In March 2012, we deployed the WSN with 36 nodes in a four-story building and by now the performance has confirmed models and expectations.

In surveillance applications, head and body orientation of people is of primary importance for assessing many behavioral traits. Unfortunately, in this context people are often encoded by a few, noisy pixels so that their characterization is difficult. We face this issue, proposing a computational framework which is based on an expressive descriptor, the covariance of features. Covariances have been employed for pedestrian detection purposes, actually a binary classification problem on Riemannian manifolds. In this paper, we show how to extend to the multiclassification case, presenting a novel descriptor, named weighted array of covariances, especially suited for dealing with tiny image representations. The extension requires a novel differential geometry approach in which covariances are projected on a unique tangent space where standard machine learning techniques can be applied. In particular, we adopt the Campbell-Baker-Hausdorff expansion as a means to approximate on the tangent space the genuine (geodesic) distances on the manifold in a very efficient way. We test our methodology on multiple benchmark datasets, and also propose new testing sets, getting convincing results in all the cases.

In most cases today when a specific person's whereabouts is monitored through video surveillance it is done manually and his or her location when not seen is based on assumptions on how fast he or she can move. Since humans are good at recognizing people this can be done accurately, given good video data, but the time needed to go through all data is extensive and therefore expensive. Because of the rapid technical development computers are getting cheaper to use and therefore more interesting to use for tedious work. This thesis is a part of a larger project that aims to see to what extent it is possible to estimate a person of interest's time dependent 3D position, when seen in surveillance videos. The surveillance videos are recorded with non overlapping monocular cameras. Furthermore the project aims to see if the person of interest's movement, when position data is unavailable, could be predicted. The outcome of the project is a software capable of following a person of interest's movement with an error estimate visualized as an area indicating where the person of interest might be at a specific time. This thesis main focus is to implement and evaluate a people detector meant to be used in the project, reduce noise in position measurement, predict the position when the person of interest's location is unknown, and to evaluate the complete project. The project combines known methods in computer vision and signal processing and the outcome is a software that can be used on a normal PC running on a Windows operating system. The software implemented in the thesis use a Hough transform based people detector and a Kalman filter for one step ahead prediction. The detector is evaluated with known methods such as Miss-rate vs. False Positives per Window or Image (FPPW and FPPI respectively) and Recall vs. 1-Precision. The results indicate that it is possible to estimate a person of interest's 3D position with single monocular cameras. It is also possible to follow the movement, to some extent, were position data are unavailable. However the software needs more work in order to be robust enough to handle the diversity that may appear in different environments and to handle large scale sensor networks.

This thesis is performed in industrial context and presents a whole framework for people detection and tracking in a camera network. It addresses the main process steps: people detection, people tracking in mono-camera context, and people re-identification in multi-camera context. High performances and real-time processing are considered as strong constraints. People detection aims to localise and delimits people in video sequences. The proposed people detection is performed using a cascade of classifiers trained using LogitBoost algorithm on region covariance descriptors. A state of the art approach is strongly optimized to process in real time and to provide better detection performances. The optimization scheme is generalizable to many other kind of detectors where all possible weak classifiers cannot be reasonably tested. People tracking in mono-camera context aims to provide a set of reliable images of every observed person by each camera, to extract his visual signature, and it provides some useful real world information for re-identification purpose. It is achieved by tracking SIFT features using a specific particle filter in addition to a data association framework which infer object tracking from SIFT points one, and which deals with most of possible cases, especially occlusions. Finally, people re-identification is performed using an appearance based approach by improving a state of the art approach, providing better performances while keeping the real-time processing advantage. A context-aware part is introduced to robustify the visual signature against people orientations, ensuring better re-identification performances in real application case., Cette thèse CIFRE est effectuée dans un contexte industriel et présente un framework complet pour la détection, le suivi mono-caméra et de la ré-identification de personnes dans le contexte multi-caméras. Les performances élevés et le traitement en temps réel sont les deux contraintes critiques ayant guidé ce travail. La détection de personnes vise à localiser/délimiter les gens dans les séquences vidéo. Le détecteur proposé est basé sur une cascade de classifieurs de type LogitBoost appliqué sur des descripteurs de covariances. Une approche existante a fortement été optimisée, la rendant applicable en temps réel et fournissant de meilleures performances. La méthode d'optimisation est généralisable à d'autres types de détecteurs d'objets. Le suivi mono-caméra vise à fournir un ensemble d'images de chaque personne observée par chaque caméra afin d'extraire sa signature visuelle, ainsi qu'à fournir certaines informations du monde réel pour l'amélioration de la ré-identification. Ceci est réalisé par le suivi de points SIFT à l'aide d'une filtre à particules, ainsi qu'une méthode d'association de données qui infère le suivi des objets et qui gère la majorité des cas de figures possible, notamment les occultations. Enfin, la ré-identification de personnes est réalisée avec une approche basée sur l'apparence globale en améliorant grandement une approche existante, obtenant de meilleures performances tout en étabt applicable en temps réel. Une partie "conscience du contexte" est introduite afin de gérer le changement d'orientation des personnes, améliorant les performances dans le cas d'applications réelles.

This paper describes a system capable of detecting and tracking various people using a new approach based on colour, stereo vision and fuzzy logic. Initially, in the people detection phase, two fuzzy systems are used to filter out false positives of a face detector. Then, in the tracking phase, a new fuzzy logic based particle filter (FLPF) is proposed to fuse stereo and colour information assigning different confidence levels to each of these information sources. Information regarding depth and occlusion is used to create these confidence levels. This way, the system is able to keep track of people, in the reference camera image, even when either stereo information or colour information is confusing or not reliable. To carry out the tracking, the new FLPF is used, so that several particles are generated while several fuzzy systems compute the possibility that some of the generated particles correspond to the new position of people. Our technique outperforms two well known tracking approaches, one based on the method from Nummiaro et al. [1] and other based on the Kalman/meanshift tracker method in Comaniciu and Ramesh [2]. All these approaches were tested using several colour-with-distance sequences simulating real life scenarios. The results show that our system is able to keep track of people in most of the situations where other trackers fail, as well as to determine the size of their projections in the camera image. In addition, the method is fast enough for real time applications.

This is the author’s version of a work that was accepted for publication in Pattern Recognition Letters. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Pattern Recognition Letters, 33, 2 (2012) DOI: 10.1016/j.patrec.2011.09.038, This paper describes a corpus, dataset and associated ground-truth, for the evaluation of people detection algorithms in surveillance video scenarios, along with the design procedure followed to generate it. Sequences from scenes with different levels of complexity have been manually annotated. Each person present at a scene has been labeled frame by frame, in order to automatically obtain a people detection ground-truth for each sequence. Sequences have been classified into different complexity categories depending on critical factors that typically affect the behavior of detection algorithms. The resulting corpus, which exceeds other public pedestrian datasets in the amount of video sequences and its complexity variability, is freely available for benchmarking and research purposes under a license agreement., This work has been partially supported by the Cátedra UAM-Infoglobal (“Nuevas tecnologías de vídeo aplicadas a sistemas de video-seguridad”), by the Ministerio de Ciencia e Innovación of the Spanish Goverment (TEC2011-25995 EventVideo: “Estrategias de segmentación, detección y seguimientos de objetos en entornos complejos para la detección de eventos en videovigilancia y monitorización”) and by the Universidad Autónoma de Madrid (“FPI-UAM: Programa propio de ayudas para la Formación de Personal Investigador”).

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Á. García-Martín, A. Cavallaro, J. M. Martínez, "People-background segmentation with unequal error cost", in 19th IEEE International Conference on Image Processing, ICIP 2012, p. 157 - 160, We address the problem of segmenting a video in two classes of different semantic value, namely background and people, with the goal of guaranteeing that no people (or body parts) are classified as background. Body parts classified as background are given a higher classification error cost (segmentation with bias on background), as opposed to traditional approaches focused on people detection. To generate the people-background segmentation mask, the proposed approach first combines detection confidence maps of body parts and then extends them in order to derive a background mask, which is finally post-processed using morphological operators. Experiments validate the performance of our algorithm in different complex indoor and outdoor scenes with both static and moving cameras., Work partially supported by the Universidad Autónoma de Madrid (“FPI-UAM”) and by the Spanish Goverment (“TEC2011-25995 EventVideo”). This work was done while the first author was visting Queen Mary University of London.

International audience; Actual computer vision algorithms cannot extract semantic information of people activity coming from the large and increasing amount of surveillance cameras installed around the world. Algorithms need to analyse video content at real time frame rate and with a false alarm detection rate as small as possible. Such algorithms can be dedicated and specifically parameterised in certain applications and restrained environment. To make algorithms as useful as possible, they need to tackle many challenging issues in order to correctly analyse human activties. For instance, people are rarely entirely seen in a video because of static (contextual object or they are partly seen by the camera field of view) and dynamic occlusion (e.g. person in front of another). We here present a novel people, head and face detection algorithm using Local Binary Pattern based features and Haar like features which we refer to as couple cell features. An Adaboost training scheme is adopted to train object features. During detection, integral images are used to speed up the process which can reach several frames per second in surveillance videos.

Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. A. Garcia-Martin, A. Hauptmann, and J. M. Martínez "People detection based on appearance and motion models", in 8th IEEE International Conference on Advanced Video and Signal-Based Surveillance, AVSS 2011, p. 256-260, The main contribution of this paper is a new people detection algorithm based on motion information. The algorithm builds a people motion model based on the Implicit Shape Model (ISM) Framework and the MoSIFT descriptor. We also propose a detection system that integrates appearance, motion and tracking information. Experimental results over sequences extracted from the TRECVID dataset show that our new people motion detector produces results comparable to the state of the art and that the proposed multimodal fusion system improves the obtained results combining the three information sources., This work has been partially supported by the Cátedra UAM-Infoglobal ("Nuevas tecnologías de vídeo aplicadas a sistemas de video-seguridad") and by the Universidad Autónoma de Madrid (“FPI-UAM: Programa propio de ayudas para la Formación de Personal Investigador”)

In computer science, contextual information can be used both to reduce computations and to increase accuracy. This paper discusses how it can be exploited for people surveillance in very cluttered environments in terms of perspective (i.e., weak scene calibration) and appearance of the objects of interest (i.e., relevance feedback on the training of a classifier). These techniques are applied to a pedestrian detector that uses a LogitBoost classifier, appropriately modified to work with covariance descriptors which lie on Riemannian manifolds. On each detected pedestrian, a similar classifier is employed to obtain a precise localization of the head. Two novelties on the algorithms are proposed in this case: polar image transformations to better exploit the circular feature of the head appearance and multispectral image derivatives that catch not only luminance but also chrominance variations. The complete approach has been tested on the surveillance of a construction site to detect workers that do not wear the hard hat: in such scenarios, the complexity and dynamics are very high, making pedestrian detection a real challenge.

In this paper, we propose to study the problem of localization of a dense set of people with a network of heterogeneous cameras. We propose to recast the problem as a linear inverse problem. The proposed framework is generic to any scene, scalable in the number of cameras and versatile with respect to their geometry, e.g. planar or omnidirectional. It relies on deducing an \emph {occupancy vector}, i.e. the discretized occupancy of people on the ground, from the noisy binary silhouettes observed as foreground pixels in each camera. This inverse problem is regularized by imposing a sparse occupancy vector, i.e. made of few non- zero elements, while a particular dictionary of silhouettes linearly maps these non-empty grid locations to the multiple silhouettes viewed by the cameras network. This constitutes a linearization of the problem, where non- linearities, such as occlusions, are treated as additional noise on the observed silhouettes. Mathematically, we express the final inverse problem either as Basis Pursuit DeNoise or Lasso convex optimization programs. The sparsity measure is reinforced by iteratively re-weighting the $\ell_1$-norm of the occupancy vector for better approximating its $\ell_0$ ``norm'', and a new kind of ``repulsive'' sparsity is used to adapt further the Lasso procedure to the occupancy reconstruction. Practically, an adaptive sampling process is proposed to reduce the computation cost and monitor a large occupancy area. Qualitative and quantitative results are presented on a basketball game. The proposed algorithm successfully detects people occluding each other given severely degraded extracted features, while outperforming state-of-the-art people localization techniques.

Air quality monitoring in indoor environments is of great significance for comfort and health, especially nowadays that people spend more than 80% of the day indoor. We propose a flexible wireless system able to detect polluted air and dangerous situations in a complex and large environment. It is important for ambient intelligent systems to be unobtrusive and to optimize the power consumption of the platforms in order to be able to live on batteries for several years. We present a system with aggressive energy management that involves three levels: sensor level, node level and network level. The sensor board we designed is a wireless sensor network (WSN) node, with very low sleep current consumption (only 8 A). It contains two modalities — a gas sensor and a Pyroelectric InfraRed (PIR) sensor. The network is multimodal: it uses information from the PIR sensor and neighbor nodes to detect the presence of people and to modulate the duty cycle of the node and the Metal Oxide Semiconductor (MOX) gas sensor. In this way we reduce the nodes’ activity and energy requirements, providing a reliable service at the same time. We simulate the benefits of the context-aware adaptive duty-cycling of the gas sensor activity and we demonstrate a significant lifetime extension compared to the continuously driven gas sensor (several years vs. several days).

Air quality monitoring in indoor environments is of great significance for comfort and health, especially nowadays that people spend more than 80% of the day indoor. We propose a flexible wireless system able to detect polluted air and dangerous situations in a complex and large environment. It is important for ambient intelligent systems to be unobtrusive and to optimize the power consumption of the platforms in order to be able to live on batteries for several years. We present a system with aggressive energy management that involves three levels: sensor level, node level and network level. The sensor board we designed is a wireless sensor network (WSN) node, with very low sleep current consumption (only 8 µA). It contains two modalities — a gas sensor and a Pyroelectric InfraRed (PIR) sensor. The network is multimodal: it uses information from the PIR sensor and neighbor nodes to detect the presence of people and to modulate the duty cycle of the node and the Metal Oxide Semiconductor (MOX) gas sensor. In this way we reduce the nodes' activity and energy requirements, providing a reliable service at the same time. We simulate the benefits of the context-aware adaptive duty-cycling of the gas sensor activity and we demonstrate a significant lifetime extension compared to the continuously driven gas sensor (several years vs. several days).

Many works address the problem of object detection by means of machine learning with boosted classifiers. They exploit sliding window search, spanning the whole image: the patches, at all possible positions and sizes, are sent to the classifier. Several methods have been proposed to speed up the search (adding complementary features or using specialized hardware). In this paper we propose a statistical-based search approach for object detection which uses a Monte Carlo sampling approach for estimating the likelihood density function with Gaussian kernels. The estimation relies on a multi-stage strategy where the proposal distribution is progressively refined by taking into account the feedback of the classifier (i.e. its response). For videos, this approach is plugged in a Bayesian-recursive framework which exploits the temporal coherency of the pedestrians. Several tests on both still images and videos on common datasets are provided in order to demonstrate the relevant speedup and the increased localization accuracy with respect to sliding window strategy using a pedestrian classifier based on covariance descriptors and a cascade of Logitboost classifiers.

The goal of this thesis is to research detection and tracking of a single person with just one camera and recognize and track human poses in order to improve the performance of an interactive spatial game that uses the horizontal position of the person and the performed poses for its control. Of course the results achieved with this research are not attached to this application only but to any application that can be imagined for using human poses detection as input. The use of one single camera restricts considerably the depth information and it is more challenging for detection and tracking when self occlusions occur but, on the other hand, it has the advantage that you only need a simple camera to run applications of this kind so it is interesting for commercial use on laptops or personal computers with a web camera. These types of games controlled by human poses can develop physical aptitudes and coordination while having fun playing with them and also re lieves you from the load of holding a mouse or stick; it is just you and the screen.

Contextual information can be used both to reduce computations and to increase accuracy and this paper presents how it can be exploited for people surveillance in terms of perspective (i.e. weak scene calibration) and appearance of the objects of interest (i.e. relevance feedback on the training of a classifier). These techniques are applied to a pedestrian detector that exploits covariance descriptors through a LogitBoost classifier on Riemannian manifolds. The approach has been tested on a construction working site where complexity and dynamics are very high, making human detection a real challenge. The experimental results demonstrate the improvements achieved by the proposed approach.

We propose to evaluate our sparsity driven people localization framework on crowded complex scenes. The problem is recast as a linear inverse problem. It relies on deducing an occupancy vector, i.e. the discretized occupancy of people on the ground, from the noisy binary silhouettes observed as foreground pixels in each camera. This inverse problem is regularized by imposing a sparse occupancy vector, i.e. made of few non-zero elements, while a particular dictionary of silhouettes linearly maps these non-empty grid locations to the multiple silhouettes viewed by the cameras network. The proposed approach is (i) generic to any scene of people, i.e. people are located in low and high density crowds, (ii) scalable to any number of cameras and already working with a single camera, (iii) unconstraint on the scene surface to be monitored. Qualitative and quantitative results are presented given the PETS 2009 dataset. The proposed algorithm detects people in high density crowd, count and track them given severely degraded foreground silhouettes.

The work presented in this manuscript deals with people detection and activity analysis in images sequences. This work has been done in the PRISME institut within the framework of the CAPTHOM project of the French Cluster S2E2. After a state of the art on video analysis and a comparative study of several video surveillance tools, we present the people detection method proposed within the framework of the CAPTHOM project. This method is based on three steps : change detection, mobile objects tracking and classification. Each steps is described in this thesis. The system was assessed on a wide videos dataset. Then, we present methods used to obtain other high-level information concerning the activity of detected persons. A criterion for characterizing their activity is presented. Then, a multi-sensors stereovision system combining an infrared and a daylight camera is used to increase performances of the people detection system but also to localize persons in the 3D space and so build the moving cartography. Finally, an abnormal events detection method based on statistics about spatio-temporal foreground pixel distribution is presented. These proposed methods offer robust and efficient solutions on high-level information extraction from images sequences., Les travaux présentés dans ce manuscrit traitent de la détection de personnes dans des séquences d’images et de l’analyse de leur activité. Ces travaux ont été menés au sein de l’institut PRISME dans le cadre du projet CAPTHOM du pôle de compétitivité S2E2. Après un état de l’art sur l’analyse de séquences d’images pour l’interprétation automatique de scènes et une étude comparative de modules de vidéo-surveillance, nous présentons la méthode de détection de personnes proposée dans le cadre du projet CAPTHOM. Celle-ci s’articule autour de trois étapes : la détection de changement, le suivi d’objets mobiles et la classification. Chacune de ces étapes est décrite dans ce manuscrit. Ce système a été évalué sur une large base de vidéos correspondant à des scénarios de cas d’usage de CAPTHOM établis par les partenaires du projet. Ensuite, nous présentons des méthodes permettant d’obtenir, à partir du flux vidéo d’une ou deux caméras, d’autres informations de plus haut-niveau sur l’activité des personnes détectées. Nous présentons tout d’abord une mesure permettant de quantifier leur activité. Ensuite, un système de stéréovision multi-capteurs combinant une caméra infrarouge et une caméra visible est utilisé pour augmenter les performances du système de détection mais aussi pour permettre la localisation dans l’espace des personnes et donc accéder à une cartographie de leurs déplacements. Finalement, une méthode de détection d’événements anormaux, basée sur des statistiques de distributions spatiales et temporelles des pixels de l’avant-plan est détaillée. Les méthodes proposées offrent un panel de solutions performantes sur l’extraction d’informations haut-niveau à partir de séquences d’images.

A novel approach is presented to locate dense crowd of people in a network of fixed cameras given the severely degraded background subtracted silhouettes. The problem is formulated as a sparsity constrained inverse problem using an adaptive dictionary constructed on- line. The framework has no constraint on the number of cameras neither on the surface to be monitored. Even with a single camera, partially occluded and grouped people are correctly detected and segmented. Qualitative results are presented in indoor and outdoor scenes.

The detection of humans in very complex scenes can be very challenging, due to the performance degradation of classical motion detection and tracking approaches. An alternative approach is the detection of human-like patterns over the whole image. The present paper follows this line by extending Tuzel et al.'s technique [1] based on covariance descriptors and LogitBoost algorithm applied over Riemannian manifolds. Our proposal represents a significant extension of it by: (a) exploiting motion information to focus the attention over areas where motion is present or was present in the recent past; (b) enriching the human classifier by additional, dedicated cascades trained on positive and negative samples taken from the specific scene; (c) using a rough estimation of the scene perspective, to reduce false detections and improve system performance. This approach is suitable in multi-camera scenarios, since the monolithic block for human-detection remains the same for the whole system, whereas the parameter tuning and set-up of the three proposed extensions (the only camera-dependent parts of the system), are automatically computed for each camera. The approach has been tested on a construction working site where complexity and dynamics are very high, making human detection a real challenge. The experimental results demonstrate the improvements achieved by the proposed approach.

6 pages en double colonne; International audience; We propose a general framework that focuses on automatic individual/multiple people motion-shape analysis and on suitable features extraction that can be used on action/activity recognition problems under real, dynamical and unconstrained environments. We have considered various athletic videos from a single uncalibrated, possibly moving camera in order to evaluate the robustness of the proposed method. We have used an easily expanded hierarchical scheme in order to classify them to videos of individual and team sports. Robust, adaptive and independent from the camera motion, the proposed features are combined within Transferable Belief Model (TBM) framework providing a two level (frames and shot) video categorization. The experimental results of 97% individual/team sport categorization accuracy, using a dataset of more than 250 videos of athletic meetings indicate the good performance of the proposed scheme.

This paper presents a probability method for detecting people in a static image. We simplify people as a torso and four limbs. The torso is fitted by a quadrangle, and each limb is fitted by one or two quadrangles depending on its pose. In order to find people, we should try to find a combination of quadrangles that must satisfy some geometric and topological constrains. Firstly, we try to detect and fit rectangle regions in the image, then we search some combinations of rectangles under certain constrains. After we get a combination, we calculate its probability of being people. If the probability is above threshold, we adjust the vertexes of rectangles so as to get a compact people model.