Kinematic sets for real-time robust articulated object tracking

Comport, Andrew I.; Marchand, Éric; Chaumette, François

doi:10.1016/j.imavis.2005.10.005

Cited by 20 publications

(23 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Early articulated model-based tracking techniques relied on tracking 2D features such as image edges on a CPU (Drummond and Cipolla 2001;Comport et al 2007). Recently introduced depth cameras along with highly parallel algorithms optimized for modern GPUs have enabled new algorithms for tracking complex 3D objects in real time.…”

Section: Electronic Supplementary Materialsmentioning

confidence: 99%

DART: dense articulated real-time tracking with consumer depth cameras

2015

View full text Add to dashboard Cite

This paper introduces DART, a general framework for tracking articulated objects composed of rigid bodies connected through a kinematic tree. DART covers a broad set of objects encountered in indoor environments, including furniture and tools, and human and robot bodies, hands and manipulators. To achieve efficient and robust tracking, DART extends the signed distance function representation to articulated objects and takes full advantage of highly parallel GPU algorithms for data association and pose optimization. We demonstrate the capabilities of DART on different types of objects that have each required dedicated tracking techniques in the past.

show abstract

Section: Electronic Supplementary Materialsmentioning

confidence: 99%

DART: dense articulated real-time tracking with consumer depth cameras

2015

View full text Add to dashboard Cite

show abstract

“…The only articulated tracking work validated on several different kinematic objects, of which we are aware, is by Comport [2]. Comport tracks an articulated object by optimizing a single hypothesis at each time step.…”

Section: B Articulated Object Trackingmentioning

confidence: 99%

Articulated Pose Estimation via Over-parametrization and Noise Projection

Brookshire

Teller

2014

Robotics: Science and Systems X

View full text Add to dashboard Cite

Abstract-We describe an algorithm to estimate the pose of a generic articulated object. Our algorithm takes as input a description of the object and a potentially incomplete series of observations; it outputs an on-line estimate of the object's configuration. This task is challenging because: (1) the distribution of object states is often multi-modal; (2) the object is not assumed to be under our control, limiting our ability to predict its motion; and (3) rotational joints make the state space highly non-linear.The proposed method represents three principal contributions to address these challenges. First, we use a particle filter implementation which is unique in that it does not require a reliable state transition model. Instead, the method relies primarily on observations during particle proposal, using the state transition model only at singularities. Second, our particle filter formulation explicitly handles missing observations via a novel proposal mechanism. Although existing particle filters can handle missing observations, they do so only by relying on good state transition models. Finally, our method evaluates noise in the observation space, rather than state space. This reduces the variability in performance due to choice of parametrization, and effectively handles non-linearities caused by rotational joints.We compare our method to a baseline implementation without these techniques and demonstrate, for a fixed error, more than an order-of-magnitude reduction in the number of required particles, an increase in the number of effective particles, and an increase in frame rate. Source code for the method is available at http://rvsn.csail.mit.edu/articulated.

show abstract

“…Ce critère est minimisé à chaque instant en prenant comme initialisation les paramètres du modèle à l'instant précédent. Dans le cas où le capteur (la caméra) est immobile, seuls les paramètres du modèle sont à déterminer à chaque instant [83][84][85][86]. Dans le cas où le capteur est mobile, il faut déterminer la position de l'objet dans l'image [87] ou la position du capteur [59].…”

Section: Mise En Correspondance Modèle/imageunclassified

Analyse de mouvement : une revue

Laguitton

Toumoulin

2009

IRBM

View full text Add to dashboard Cite

Reçu le 21 décembre 2008 ; accepté le 22 décembre 2008 Disponible sur Internet le 5 février 2009Résumé L'analyse des séquences temporelles d'images a fait l'objet de nombreux travaux depuis plus de 20 ans. Elle permet d'accéder à des informations importantes sur les objets contenus dans les scènes observées que ce soit à des fins de détection, d'estimation, de suivi ou de reconnaissance. Elle conduit aussi à de nombreuses applications dans des domaines très divers allant de la télévision à la météorologie, de la robotique à la surveillance ou encore au médical. Cette importance reconnue et la diversité des problématiques qu'elle recouvre ont motivé cet état de l'art qui se positionne dans un cadre générique plutôt qu'applicatif de façon à donner aux lecteurs une vue d'ensemble. Après avoir introduit les éléments clés à considérer (objets, observations, objectifs), trois grandes familles de méthodes sont analysées et discutées : les approches différentielles, les techniques bayesiennes et les méthodes de mise en correspondance. Un dernier paragraphe s'attache à faire le lien avec l'imagerie médicale en soulignant les spécificités qu'elle recouvre. AbstractImage sequence analysis has been a major topic for years. It allows accessing important information on the objects of interest for detection, estimation, tracking and recognition tasks. Many applications are concerned from video to weather forecasting, robotics to in-door monitoring up to medicine. The place that it has today and the diversity of the problems to address have motivated this state-of-the-art. A generic approach has been adopted in order to provide a full view on the recent advances reported in the literature at the methodological level. After introducing the key elements to be considered (objects, observations, objectives), three main classes of methods are analysed and discussed: the differential approaches, the Bayesian techniques and the matching methods. A last section is focused on the links with medical imaging with the aim to emphasize the special features to deal with.

show abstract

Kinematic sets for real-time robust articulated object tracking

Cited by 20 publications

References 28 publications

DART: dense articulated real-time tracking with consumer depth cameras

DART: dense articulated real-time tracking with consumer depth cameras

Articulated Pose Estimation via Over-parametrization and Noise Projection

Analyse de mouvement : une revue

Contact Info

Product

Resources

About