Interactive multi-user video retrieval systems

Abstract

In this paper we present two interactive multi-user systems for video search and browsing. The first is composed by web applications which allows multiuser interaction in a distributed environment; such applications are based on the Rich Internet Application paradigm, designed to obtain the levels of responsiveness and interactivity typical of a desktop application. The second system implements a multiuser collaborative application within a single location, exploiting multi-touch devices. Both systems use the same backend, based on a service oriented architecture (SOA) that provides services for automatic and manual annotation, and an ontology-based video search and browsing engine. Ontology-based browsing let users to inspect the content of video collections; user queries are expanded through ontology reasoning. User-centered field trials of the systems, conducted to assess the user experience and satisfaction, have shown that the approach followed to design these interfaces is extremely appreciated by professional archivists and people working on multimedia.

Multilayer Video Semantic Feature Extraction Method

“…The feature of the method is to assume that the output random variables constituting a Markov random field [8,10,16] (Markov Random Field, MRF) . Conditional Random Fields is very suitable for time series of random variables labeling problem [2] , and the semantics extraction of video data just satisfies these data features and applications. Thus, our object research work is mainly based on conditional random fields to build video semantic feature extraction method.…”

A social network for video annotation and discovery based on semantic profiling

“…Several methods have been proposed and they can be divided mainly in: i) supervised methods, where a set of classifiers is trained to detect scenes, objects and events, typically using methods based on the Bag-of-VisualWords (BoVW) approach [12,15]; ii) unsupervised methods that exploit the plethora of user generated annotations of multimedia content to annotate images [7,10] and more recently also to suggest and localize tags in video shots [1,2,6]. However, the performance of these systems allows to deploy them in a semiautomatic context, along with tools for manual annotation so that professional users can create semantic annotations based on ontologies [3]; even so, tools made to be deployed to general users have to reduce the cost of learning the use of ontologies such as LSCOM [8], designed for use in professional or scientific contexts, to be substituted by folksonomies created without need of explicitly specifying the relations between the concepts. Moreover, there is no need to force users to produce thorough annotations of the video content.…”

Models and operators for extension of active multimedia documents via annotations

“…subtitles) is traversed [Kurihara, 2012]. On the side of novel operations, there are interaction techniques for controlling media via gestures, which prove useful to support the prominent use of touch-based devices such as tabletops [Bertini et al, 2011] and smart phones [Huber et al, 2010]. Another line of novel operations consists on direct manipulation of objects in the scene, for instance for replaying a specific scene by dragging an object along its original trajectory [Santosa et al, 2013].…”

“…Graph-based visualizations of low-level features, whose links are computed via a distance metric, has been demonstrated by Delest et al [2006]. Systems that count on annotations with semantic relationships also can take advantage of graph-visualizations to allow navigation of related media elements, as demonstrated by Bertini et al [2011]. On the issue of intra-media visualization, hierarchical video browsing has been demonstrated [Del Fabro et al, 2010] as a way to incrementally explore segments of interest in a video by traversing a tree visualization;…”