Search

The international conference on Intelligent Human Computer Interaction (IHCI) is a forum for the presentation of research results and technological advances at the crossroads of human-computer interaction, artificial intelligence, signal processing, and computer vision. Its brings together engineers and scientists from around the world focussing on theoretical, practical, and applicational aspects of the field

International audience; Learn how to use GPU for accelerating 3D registration with Kinect or similar devices in order to capture highly-detailed facial performance in real time or at an interactive speed. We describe the energy-based approach that we borrowed from Hao Li et al. paper published at SGP 2008. We also explain why we can benefit from GPU computation power and achieve higher quality and more detail at interactive speeds. Finally, we elaborate on how real-time performance can be achieved by improving our CUDA-based implementation

International audience; Proceedings of the 6th international conference on Intelligent Human Computer Interaction, IHCI 2014, that was held in Evry, France, on December 8-10, 2014. IHCI is a forum for the presentation of technological advances and research results at the crossroads of human-computer interaction, artificial intelligence, signal processing and computer vision. It brings together engineers and scientists from around the world

National audience; Dans cet article nous comparons deux approches pour la reconnaissance de gestes en 2D et en 3D. La première exploite les points d'intérêt spatio-temporels où chaque séquence est décrite par des histogrammes de flux optique locaux. La reconnaissance est effectuée par une machine à vecteur support (SVM). La deuxième approche consiste à utiliser la trajectoire 3D de la main acquise à l'aide d'un capteur Kinect et la reconnaissance est effectuée par déformation temporelle dynamique (DTW). Nous comparons les deux approches en étudiant leurs limitations, l'efficacité de chacune en termes de taux de reconnaissance et de temps de calcul

International audience; We propose a novel probabilistic tracking algorithm based on an elastic blob ensemble (EBE) which is applicable to track flexible objects, like human upper body composed of head, torso, and hips. It outputs a coarse motion cue in the form of the object's location and orientation together with the location of the blobs. The main assumption is that the orientation of the whole object does not change much between neighboring frames. Hence, a discrete solution space is created in the current frame around the blobs' positions from the previous frame. Our model then promotes solutions whose orientations are close to the prior orientation, which match the modeled to the observed appearance well, and which follow modeled spatial configuration. It combines the strengths of three popular approaches to visual tracking: mean-shift tracker, particle filtering, and pictorial structures. As a result, the proposed framework leads to a robust and fast tracker of a person at the rate of 15-30 fps on a regular desktop PC

Dans une collaboration à distance médiatisée par ordinateur, la communication entre personnes est renforcée par le contact visuel. Toutefois, la vidéoconférence ne permet pas restituer les actions sur les objets partagés. La perception mutuelle des participants et de leurs actions peut être renforcée au moyen d'un environnement virtuel en réseau. Des représentations virtuelles des utilisateurs distants (avatars) et des objets qu'ils partagent y sont rassemblées de sorte que les actions des utilisateurs sur les objets sont utilisées pour animer les avatars qui miment leurs actions. Les gestes de communication sont eux acquis par vision artificielle et restitués virtuellement en temps réel. Un système d'acquisition par vision artificielle des gestes et expressions du visage en 3D peut être développé à partir d'une unique webcam. Ce système à bas débit et bas coût permet d'envisager des applications grand public de téléprésence virtuelle en 3D.

International audience; Particle filtering is known as a robust approach for motion tracking by vision, at the cost of heavy computation in the high dimensional pose space. In this work, we describe a number of heuristics that we demonstrate to jointly improve robustness and real-time for motion capture. 3D human motion capture by monocular vision without markers can be achieved in real-time by registering a 3D articulated model on a video. First, we search the high-dimensional space of 3D poses by generating new hypotheses (or particles) with equivalent 2D projection by kinematic flipping. Second, we use a semi-deterministic particle prediction based on local optimization. Third, we deterministically resample the probability distribution for a more efficient selection of particles. Particles (or poses) are evaluated using a match cost function and penalized with a Gaussian probability pose distribution learned off-line. In order to achieve real-time, measurement step is parallelized on GPU using the OpenCL API. We present experimental results demonstrating robust real-time 3D motion capture with a consumer computer and webcam

Nous nous intéressons à l’acquisition 3D des gestes humains par vision monoscopique en temps réel sans marqueurs. Notre approche procède par recalage d’un modèle 3D articulé du corps sur une séquence vidéo qui consiste à rechercher itérativement la position du modèle et les angles d’articulation qui maximisent la correspondance entre des caractéristiques du modèle 3D projeté et des primitives de l’image. Nous avons précédemment décrit une mise en œuvre à la cadence vidéo d’un recalage initial sur les régions colorées suivi d’un recalage plus précis sur les contours. Dans ce travail, nous comparons expérimentalement l’erreur résiduelle en fonction du temps de calcul pour chacune de ces primitives de recalage et nous proposons un compromis en fonction de la puissance de calcul disponible.

A challenge for 3D motion capture by monocular vision is 3D-2D projection ambiguities that may bring incorrect poses during tracking. In this paper, we propose improving 3D motion capture by learning human gesture models from a library of gestures with variants. This library has been created with virtual human animations. Gestures are described as Gaussian Process Dynamic Models (GPDM) and are used as constraints for motion tracking. Given the raw input poses from the tracker, the gesture model helps to correct ambiguous poses. The benefit of the proposed method is demonstrated with results.

International audience; In this paper we present how work toward building a library of communicative gestures with the aim to use it to train the tracking algorithm. We describe also how we encompass user's communicative distinctiveness in our library

International audience; Virtual worlds are developing rapidly over the Internet. They are visited by avatars and staffed with Embodied Conversational Agents (ECAs). An avatar is a representation of a physical person. Each person controls one or several avatars and usually receives feedback from the virtual world on an audio-visual display. Ideally, all senses should be used to feel fully embedded in a virtual world. Sound, vision and sometimes touch are the available modalities. This paper reviews the technological developments which enable audio-visual interactions in virtual and augmented reality worlds. Emphasis is placed on speech and gesture interfaces, including talking face analysis and synthesis

National audience; Nous nous intéressons à l'acquisition 3D des gestes humains par vision monoscopique en temps réel sans marqueurs. Notre approche procède par recalage d'un modèle 3D articulé du corps sur une séquence vidéo qui consiste à rechercher itérativement la position du modèle et les angles d'articulation qui maximisent la correspondance entre des caractéristiques du modèle 3D projeté et des primitives de l'image. Nous avons précédemment décrit une mise en oeuvre à la cadence vidéo d'un recalage initial sur les régions colorées suivi d'un recalage plus précis sur les contours [7]. Dans ce travail, nous comparons expérimentalement l'erreur résiduelle en fonction du temps de calcul pour chacune de ces primitives de recalage et nous proposons un compromis en fonction de la puissance de calcul disponible.

International audience; This paper presents briefly describes the state of the art of accelerating image processing with graphics hardware (GPU) and discusses some of its caveats. Then it describes GpuCV, an open source multi-platform library for GPU-accelerated image processing and Computer Vision operators and applications. It is meant for computer vision scientist not familiar with GPU technologies. GpuCV is designed to be compatible with the popular OpenCV library by offering GPU-accelerated operators that can be integrated into native OpenCV applications. The GpuCV framework transparently manages hardware capabilities, data synchronization, activation of low level GLSL and CUDA programs, on-the-fly benchmarking and switching to the most efficient implementation and finally offers a set of image processing operators with GPU acceleration available

International audience; Home Healthcare application based on vital and actimetrics parameters telemonitoring : " Télévigilance ". Gesture-based communication in networked virtual environments. Speech Recognition Technologies

This paper presents a state of the art report on using graphics hardware for image processing and computer vision. Then we describe GPUCV, an open library for easily developing GPU accelerated image processing and analysis operators and applications.

This paper presents a comparison of original couplings between color filter array demosaicking methods and wavelet compression (JPEG2000). We focus on an application handling huge microscopy images (64 K/spl times/64 K pixels) for telediagnosis. Whereas coding is usually achieved after interpolation, we also consider demosaicking after decompression in order to optimize image quality for a given size of data. We also study the JPEG2000 stream structure for interactive visualization.

We present a method for 3D human motion capture using a single camera, without markers and without a priori knowledge on gestures. It is based on registering a 3D articulated model on color images with respect to biomechanical constraints. Gestures regularization is discussed as a way to cope with projection ambiguities. Computation is reduced by registering only the moving parts of the body.

National audience

We propose a software architecture for picture processing that allows efficient memory management when algorithms with many operators are applied to large images, and that allows automated parallelization. This architecture relies on image tiling and operators with a call back function that evaluates image tiles on demand. Several tiling strategies with and without overlapping are discussed. The compexity of this evaluation strategy is hidden to application programs. This is shown with a sample program. This architecture is well suited for neighborhood operators such as convolutions and mathematical morphology operators.

A three-dimensional (3D) reconstruction of the vessel lumen from two angiographic views, based on the reconstruction of a series of cross-sections, is proposed. Assuming uniform mixing of contrast medium and background subtraction, the cross-section of each vessel is reconstructed through a binary representation. A priori information about both the slice to be reconstructed and the relationships between adjacent slices are incorporated to lessen ambiguities on the reconstruction. Taking into account the knowledge of normal vessel geometry, an initial solution of each slice is created using an elliptic model-based method. This initial solution is then deformed to be made consistent with projection data while being constrained into a connected realistic shape. For that purpose, properties on the expected optimal solution are described through a Markov random field. To find an optimal solution, a specific optimization algorithm based on simulated annealing is used. The method performs well both on single vessels and on branching vessels possessing an additional inherent ambiguity when viewed at oblique angles. Results on 2D slice independent reconstruction and 3D reconstruction of a stack of spatially continuous 2D slices are presented for single vessels and bifurcations. >

The paper presents a method to model an arterial bifurcation from a pair of X-ray angiographic images. It is the initial step of a reconstruction process aiming at detecting and quantifying abnormal sites located on bifurcations. The method proposed consists of two steps. First, each image is independently segmented to extract the vessels in the images. The algorithm uses dynamic programming first to find the bifurcation centrelines from the original images, and secondly to extract vessel edges from the morphological gradient images, under a constraint of parallelism with the previously detected centrelines. Then, a three-dimensional bifurcation model is built by adapting cylinders around the three-dimensional bifurcation centrelines. These cylinders are obtained as a stack of binary orientable ellipses fitted to the projection densities in the corresponding cross-sections. Results obtained on simulated data, phantom and femoral bifurcations are displayed.

Le travail est consacré à l'extraction de primitives dans des images tridimensionnelles obtenues par une caméra laser à triangulation. Les zones artéfactées de l'image, où l'information fournie est fausse, sont détectées. La recherche des arêtes des objets est abordée. Enfin, les méthodes d'extraction des faces planes sont analysées. En particulier la transformation de Hough a été mise en oeuvre et étudiée par les outils du traitement du signal: son support fréquentiel a été recherché afin d'optimiser son échantillonnage