Out of vocabulary queries for word graph-based keyword spotting

[EN] In this master thesis several approaches are presented to support out of vocabulary queries in a Word Graph (WG)-based Keyword Spotting (KWS) application for handwritten text lines. Generally, KWS assigns a score that estimates how likely is that a given keyword is present in a certain line ima...

Descripción completa

Detalles Bibliográficos
Autor: Puigcerver I Pérez, Joan
Tipo de recurso: tesis de maestría
Fecha de publicación:2014
País:España
Institución:Universitat Politècnica de València (UPV)
Repositorio:RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Idioma:inglés
OAI Identifier:oai:riunet.upv.es:10251/53360
Acceso en línea:https://riunet.upv.es/handle/10251/53360
Access Level:acceso abierto
Palabra clave:Detección de palabras clave
Reconocimiento automático de escritura
Recuperación de la información
Word graphs
Filler
Keyword Spotting
Handwritting text recognition
Information retrieval
Smoothing
LENGUAJES Y SISTEMAS INFORMATICOS
Máster Universitario en Inteligencia Artificial, Reconocimiento de Formas e Imagen Digital-Màster Universitari en Intel·ligència Artificial, Reconeixement de Formes i Imatge Digital
Descripción
Sumario:[EN] In this master thesis several approaches are presented to support out of vocabulary queries in a Word Graph (WG)-based Keyword Spotting (KWS) application for handwritten text lines. Generally, KWS assigns a score that estimates how likely is that a given keyword is present in a certain line image. WGbased KWS offers very fast search times but assumes a closed vocabulary and assigns null scores to any word not included in such vocabulary. This work tries to provide to the WG-based KWS the flexibility of non-restricted searches and the speed achieved by the usage of WG.