Teses e Dissertações
URI Permanente desta comunidade
Navegar
Navegando Teses e Dissertações por Autor "Almeida Junior, Jurandy Gomes de Almeida Junior [UNIFESP]"
Agora exibindo 1 - 1 de 1
Resultados por página
Opções de Ordenação
- ItemExplorando informação semântica na construção de dicionários visuais para recuperação de vídeos por conteúdo(Universidade Federal de São Paulo (UNIFESP), 2016-10-08) Duarte, Leonardo Assuane [UNIFESP]; Almeida Junior, Jurandy Gomes de Almeida Junior [UNIFESP]; Universidade Federal de São Paulo (UNIFESP)Recent advances in technology have allowed the increase of multimedia content, creating huge video libraries, thus motivating the development of systems to efficiently manage those kind of data. In this context, content-based information retrieval systems are a promissing tool to search videos by their content. One of the challenges is how to extract semantic information using the visual information of the video content. The visual dictionary model is commonly used to handle this problem. In this model, visual patterns are grouped to form a semantic unit, called visual word. Then, the video?s visual content can be represented as a bag of visual words, a similar idea used for the representation of textual documents. It is common to create visual dictionaries using information that has no semantic, like corners and edges. The contributions of this work are approaches to include semantic information in the visual dictionary. To accomplish that, two methods were proposed to build visual dictionaries with semantic information for content-based information retrieval using recent solutions in computer vision and machine learning. Results show the proposed methods achieves results comparable or superior to state-of-the-art methods, with the advantage of providing a much more compact representation than existing solutions.