UniCa UniCa News Notizie Fusion of multiple classifier for video analysis

Fusion of multiple classifier for video analysis

Pattern Recognition: seminario di Ingrid Visentini dell’Università di Udine
12 maggio 2010

Mercoledì 12 maggio 2010, ore 15
Sardegna Ricerche - Parco scientifico e tecnologico di Pula
Edificio 2, sala conferenze   
 
FUSION OF MULTIPLE CLASSIFIERS FOR VIDEO ANALYSIS
  

Seminario di INGRID VISENTINI (Università di Udine)  nell’ambito delle attività del gruppo di ricerca in Pattern Recognition
 
 
ABSTRACT  The talk will be focused on classifiers fusion in video applications. More in detail, the fusion of classifiers will be exploited to build an on-the-fly model to represent the appearance of an object. The combination of experts will be constantly updated, to learn variations in the appearance of the target through the frames; the ensemble will be composed of online trained classifiers, that are used to detect and follow an object in video streams, improving the real-time tracking accuracy.
The problem of how multiple classifiers can be fused together, that is the choice of the rule to merge their outputs to reach a final consensus, will be addressed. Three criteria to improve the performance of a classifier ensemble will be proposed; to test them, the classification problem as a means to get a better target tracking performance on real-world multimedia video sequences will be examined

BIO  Ingrid Visentini received the MSc with full marks in Computer Science from the University of Udine, Italy, in 2006. Since January 2007, she has been a Ph.D. student in Communication Multimedia at the same University and she is a fellow of the AVIRES Laboratory at the Mathematics and Computer Science Department. She spent several time abroad, at the University of Alicante, Spain (2005), and as a Ph.D. Visiting student at the Centre for Vision and Speech Signal Processing, University of Surrey, U.K. (2008).
Her main interests are in the field of machine learning, in particular: adaptive and online learning as main topic, classifiers combination for target detection, tracking via classification in complex environments, automatic video annotation and surveillance intelligence.
She serves as a reviewer for several International Journals and Conferences. She cooperated with local and national industries and research centres, and participated in several national and international projects. She is currently working on Information Fusion topics, as multisensor fusion, sensor management and situation assessment within the “DAFNE” EDA project.
 
INFO  Prof. Giorgio Giacinto
Dipartimento di Ingegneria Elettrica ed Elettronica
Piazza D’Armi - 09123 Cagliari - Tel: 070 675 5752 Fax: 070 675 5782
URL: http://www.diee.unica.it/giacinto/

Ultime notizie

Questionario e social

Condividi su:
Impostazioni cookie