Papers by Roberto Brunelli
Efficient Image Retrieval by Examples
Advances in Visual Information Management, 2000
Google, Inc. (search). ...
Estimation of Head Pose
Computers in the Human Interaction Loop, 2009
In building proactive systems for interacting with users by analyzing and recognizing scenes and ... more In building proactive systems for interacting with users by analyzing and recognizing scenes and settings, an important task is to deal with people’s occupations: Not only do their locations or identities become important, but their looking direction and orientation are crucial cues to determine everybody’s intentions and actions. The understanding of interaction partners or targeted objects is relevant in deciding
Machine Vision and Applications, 1995
Abstract|The paper describes a multisensorial person identi cation system: visual and acoustic cu... more Abstract|The paper describes a multisensorial person identi cation system: visual and acoustic cues are used jointly for person identi cation. A simple approach, based on the fusion of the lists of scores produced independently by a speaker recognition system and a face recognition system, is presented. Experiments are reported which s h o w that integration of visual and acoustic information enhances both performance and reliability of the separate systems. Finally two n e t work architectures, based on radial basis function theory, a r e proposed to describe integration at di erent levels of abstraction.
Journal of Visual Communication and Image Representation, 1999
Today a considerable amount of video data in multimedia databases requires sophisticated indices ... more Today a considerable amount of video data in multimedia databases requires sophisticated indices for its effective use. Manual indexing is the most effective method to do this, but it is also the slowest and the most expensive. Automated methods have then to be developed. This paper surveys several approaches and algorithms that have been recently proposed to automatically structure audio-visual data, both for annotation and access. C 1999 Academic Press
Journal of Computational and Applied Mathematics, 1995
In this paper a nondeterministic minimization algorithm is presented. A common feature of random ... more In this paper a nondeterministic minimization algorithm is presented. A common feature of random search algorithms is that little or no use is made of information on the local structure of the function to be minimized. While this can be justified when the function has a very complicated microstructure, it results in an unnecessary loss of efficiency when the landscape is smooth but anisotropic. To overcome this deficiency, we propose a random minimization algorithm with adaptive memory: the algorithm decides by itself how much of the information gathered through the process of minimizing the function can be successfully used to guide the search. Extensive experiments (minimization of quadratic forms, computation of the minimum eigenvalue of positive definite quadratic forms of high dimensionality, eigenvalue computation in Hilbert spaces and fitting of data by superposition of Gaussians) show that efficiency is increased and that the algorithm is able to adapt quickly to the current landscape.
IEEE Transactions on Pattern Analysis and Machine Intelligence, 1995
IEEE Transactions on Multimedia, 2000
A currently relevant research field in information sciences is the management of non-traditional ... more A currently relevant research field in information sciences is the management of non-traditional distributed multimedia databases. Two related key issues are achieving an efficient content-based query by example retrieval and a fast response time. This paper presents the architecture of a distributed image retrieval system which provides novel solutions to these key issues. In particular, a way to quantify the effectiveness of low level visual descriptors in database query tasks is presented. The results are also used to improve the system response time, an important issue when querying very large databases. A new mechanism to adapt system query strategies to user behavior is also introduced to improve the effectiveness of relevance feedback and overall system response time. Finally, the issue of browsing multiple distributed databases is considered and a solution proposed using multidimensional scaling techniques.
Graphical Models and Image Processing, 1996
Abstract| In this paper we present a system for browsing large mug-shot databases and the creatio... more Abstract| In this paper we present a system for browsing large mug-shot databases and the creation identikits of photographic quality. The two functions are interrelated: the available database provides direct feedback to the user building the identikit and the identikit itself can be used as an access key to the image database. SpotIt! provides a virtually unlimited set of alternative features that can be browsed e ciently in the appropriate context, interactive holistic feature modi cation coupled to syntactic access to a feature database, and quantitative, automatic computation of face similarities, providing real-time feedback of the system which constantly shows the most promising matches to the identikit being built.
Biological Cybernetics, 1991
Query simplification and strategy selection for image retrieval
Abstract While many systems are currently available supporting the query-by-example paradigm for ... more Abstract While many systems are currently available supporting the query-by-example paradigm for image retrieval, some key issues, such as the effective introduction of relevance feedback and the automatic selection of an optimal search strategy, need further investigation. This paper discusses query splitting as a way to improve the effectiveness of relevance feedback techniques based on feature weighting. The possibility of selecting among a set of image retrieval strategies the one which optimizes speed and quality on ...
We propose a methodology for extracting multimedia information from product catalogues empowered ... more We propose a methodology for extracting multimedia information from product catalogues empowered by the synergetic use and extension of a domain ontology. The use of domain ontologies in this context additionally opens up innovative ways of catalogue use. The method is characterized by incrementally feeding and exploiting the ontology during an information extraction process, implemented by the semantic annotation of the analysed document, and by providing support for detecting existing similar ontologies to enable reuse of (parts of) them.
COMPASS: An image retrieval system for distributed databases
Abstract A currently relevant research field in information sciences is the management of non-tra... more Abstract A currently relevant research field in information sciences is the management of non-traditional distributed multimedia databases. Three related key issues are achieving an effective content-based query-by-example retrieval, a fast response time and adapting image comparison strategies to user needs. This paper presents COMPASS, a distributed image retrieval system which provides novel solutions to these key issues. Statistical methods are used to quantify image descriptor effectiveness, to simplify user queries and ...
Abstract This paper analyzes the use of histograms of low-level image features, such as color and... more Abstract This paper analyzes the use of histograms of low-level image features, such as color and luminance, as descriptors for image retrieval purposes. The discrimination ability of several descriptors, the issues of histogram size and comparison, are considered in a common statistical framework
Proc. of the 3rd Italian Semantic Web Workshop-SWAP 2006, 2006
Abstract—The demand for efficient methods for extracting knowledge from multimedia content has le... more Abstract—The demand for efficient methods for extracting knowledge from multimedia content has led to a growing research community investigating the convergence of multimedia and knowledge technologies. In this paper we describe a methodology for extracting multimedia information from product catalogues empowered by the synergetic use and extension of a domain ontology. The methodology was implemented in the Trade Fair Advanced Semantic Annotation Pipeline of the VIKE-framework. Index Terms—Semantic ...
In Proceedings of the ESWC 2006 Workshop on Mastering the Gap: From Information Extraction to Semantic Representation, Budva, Montenegro, Jun 12, 2006
Abstract. Semantic annotation of content is a crucial building block of making the Semantic Web f... more Abstract. Semantic annotation of content is a crucial building block of making the Semantic Web fly. The (semi-) automatic support of the underlying semantic knowledge supply chain requires contributions from different research disciplines and well-defined pipelines, which step-by-step create such annotations from raw content objects. This paper presents an annotation pipeline that has been designed and implemented as part of the VIKEF project. A clear structuring of the pipeline, the selection of adequate representation formats for the ...
Abstract| A person recognition system that makes use of acoustic and visual features is described... more Abstract| A person recognition system that makes use of acoustic and visual features is described. The system combines features at a score level and is capable of performing either an identi cation or a rejection. The improved performance of the integrated system with respect to the separate subsystems (acoustic and visual) is quanti ed.
Recognition system, particularly for recognising people
Tracking Visitors in a Museum
Cognitive Technologies, 2007
The most basic, low-level information about people is their location in the environment. This inf... more The most basic, low-level information about people is their location in the environment. This information is of the utmost importance in surveillance systems that, among other things, handle protected or dangerous zones since suspicious behaviour may often be detected ...
Joint Bayesian tracking of head location and pose from low-resolution video
This paper presents a visual particle lter for jointly track- ing the position of a person and he... more This paper presents a visual particle lter for jointly track- ing the position of a person and her head pose. The resulting informa- tion may be used to support automatic analysis of interactive people behaviour, by supporting proxemics analysis and providing dynamic in- formation on focus of attention. A pose-sensitive visual likelihood is pro- posed which models the appearance of
This paper presents a real time graphical simulator based on a client-server architecture. The re... more This paper presents a real time graphical simulator based on a client-server architecture. The rendering engine, supported by a specialized client application for the automatic generation of goal oriented motion of synthetic characters, is used to produce realistic image sequences for extensive performance assessment of computer vision algorithms for people tracking.
Uploads
Papers by Roberto Brunelli