This paper describes an original approach which jointly addresses two fundamental issues of video partitioning which represent the early important stage of any content-based video indexing system. These two issues are the detection of shot changes, and the labeling of the shot con guration related to the camera movement in terms of static shot, panning, traveling, zooming,... They are both derived from the computation, at each time instant, of the dominant motion in the image represented by a 2D a ne model, and from the variation of the size of its associated support. The successive steps of the method rely on statistical techniques ensuring robustness and e ciency. In particular, it can cope with scenes containing moving objects. Results on a real documentary video are reported and validate the proposed approach. Unité de recherche INRIA Rennes IRISA, Campus universitaire de Beaulieu, 35042 RENNES Cedex (France) Téléphone : 02 99 84 71 00 -International : +33 2 99 84 71 00 R sum : Nous d crivons une approche traitant conjointement deux probl mes fondamentaux du partitionnement de vid o en plan, la d tection des changements de plan et l' tiquetage d'une con guration de plan reli un mouvement de cam ra en termes de plan xe, traveling, zoom... Il s'agit d' une premi re tape importante pour tout syst me d'indexation de vid o par le contenu. La m thode propos e exploite, pour les deux aspects du probl me, l'estimation du mouvement dominant dans l'image, repr sent par un mod le a ne 2D, et de l' volution temporelle du support associ . Les tapes successives de cette m thode reposent sur des techniques statistiques, qui en assurent la robustesse et l'e cacit . Mots-cl : D coupage en plan, Mouvement de cam ra, Indexation vid o

No abstract

Usage of mobile devices (phones, digital cameras) raises the need for organizing large personal image collections. In accordance with studies on user needs, we propose a statistical criterion and an associated optimization technique, relying on geo-temporal image metadata, for building and tracking a hierarchical structure on the image collection. In a mixture model framework, particularities of the application and typical data sets are taken into account in the design of the scheme (incrementality, ability to cope with non-Gaussian data, with both small and large samples). Results are reported on real data sets.

No abstract

No abstract

International audienceAggregating statistical representations of classes is an important task for current trends in scaling up learning and recognition, or for addressing them in distributed in- frastructures. In this perspective, we address the problem of merging probabilistic Gaus- sian mixture models in an efficient way, through the search for a suitable combination of components from mixtures to be merged. We propose a new Bayesian modelling of this combination problem, in association to a variational estimation technique, that handles efficiently the model complexity issue. A main feature of the present scheme is that it merely resorts to the parameters of the original mixture, ensuring low computational cost and possibly communication, should we operate on a distributed system. Experimental results are reported on real dat

No abstract

No abstract