Add abstract
Want to add your dissertation abstract to this database? It only takes a minute!
Search abstract
Search for abstracts by subject, author or institution
Want to add your dissertation abstract to this database? It only takes a minute!
Search for abstracts by subject, author or institution
[en] AN EVALUATION OF BIMODAL RECOGNITION SYSTEMS BASED ONVOICE AND FACIAL IMAGES
by SEBASTIN SANTAMARINA ABEL
Institution: | Pontifical Catholic University of Rio de Janeiro |
---|---|
Year: | 2017 |
Keywords: | [pt] FUSAO DE ESCORES BASEADA EM DENSIDADE; [pt] FUSAO DE ESCORES BASEADA EM TRANSFORMACAO; [pt] FUSAO DE ESCORES BASEADAS EMCLASSIFICADORES; [pt] GMMUBM; [pt] I-VECTOR; [pt] LBP |
Posted: | 02/01/2018 |
Record ID: | 2159656 |
Full text PDF: | http://www.maxwell.vrac.puc-rio.br/Busca_etds.php?strSecao=resultado&nrSeq=29315 |
[pt] Esta dissertao tem como objetivo avaliar osmtodos de fuso de escores mais importantes na combinao de doissistemas uni-modais de reconhecimento em voz e imagens faciais.Para cada sistema uni-modal foram implementadas duas tcnicas declassificao: o GMM/UBM e o I-Vetor/GPLDA para voz e o GMM/UBM eum classificador baseado em LBP para imagens faciais. Estessistemas foram combinados entre eles, sendo 4 combinaes testadas.Os mtodos de fuso de escores escolhidos se dividem em trsgrupos: Fuso baseada em densidade, fuso baseada em transformaoe fuso baseada em classificadores, e foram testadas algumasvariantes para cada grupo. Os mtodos foram avaliados em modo deverificao, usando duas bases de dados, uma base virtual formadapor duas bases uni-modais e outra base bimodal. O resultado de cadatcnica bimodal empregada foi comparado com os resultados dastcnicas uni-modais, percebendo-se ganhos significativos naacurcia de reconhecimento. As tcnicas de fuso baseadas emdensidade mostraram os melhores resultados entre todas as outrastcnicas, mais apresentaram uma maior complexidade computacionalpor causa do processo de estimao da densidade. [en] The main objective of this dissertation is tocompare the most important approaches for score-level fusion of twounimodal systems consisting of facial and independent speakerrecognition systems. Two classification methods for each biometricmodality were implemented: a GMM/UBM and an I-Vector/GPLDAclassifiers for speaker independent recognition and a GMM/UBM andLBP-based classifiers for facial recognition, resulting in fourdifferent multimodal combination of fusion explored. Thescore-level fusion methods investigated are divided inDensity-based, Transformation-based and Classifier-based groups andfew variants on each group are tested. The fusion methods weretested in verification mode, using two different databases, onevirtual database and a bimodal database. The results of eachbimodal fusion technique implemented were compared with theunimodal systems, which showed significant recognition performancegains. Density-based techniques of fusion presented the bestresults among all fusion approaches, at the expense of highercomputational complexity due to the density estimationprocess.Advisors/Committee Members: RAUL QUEIROZ FEITOSA.
Want to add your dissertation abstract to this database? It only takes a minute!
Search for abstracts by subject, author or institution
Electric Cooperative Managers' Strategies to Enhan...
|
|
Bullied!
Coping with Workplace Bullying
|
|
The Filipina-South Floridian International Interne...
Agency, Culture, and Paradox
|
|
Solution or Stalemate?
Peace Process in Turkey, 2009-2013
|
|
Performance, Managerial Skill, and Factor Exposure...
|
|
The Deritualization of Death
Toward a Practical Theology of Caregiving for the ...
|
|
Emotional Intelligence and Leadership Styles
Exploring the Relationship between Emotional Intel...
|
|
Commodification of Sexual Labor
Contribution of Internet Communities to Prostituti...
|
|
The Census of Warm Debris Disks in the Solar Neigh...
|
|
Risk Factors and Business Models
Understanding the Five Forces of Entrepreneurial R...
|
|