Detalhes bibliográficos
| Ano de defesa: |
2018 |
| Autor(a) principal: |
SANTOS, Daniel de Matos Luna dos
 |
| Orientador(a): |
BARROS FILHO, Allan Kardec Duailibe
|
| Banca de defesa: |
BARROS FILHO, Allan Kardec Duailibe
,
SANTANA, Ewaldo Eder Carvalho
,
FREIRE, Raimundo Carlos Silvério
,
CAVALCANTE, André Borges
|
| Tipo de documento: |
Dissertação
|
| Tipo de acesso: |
Acesso aberto |
| Idioma: |
por |
| Instituição de defesa: |
Universidade Federal do Maranhão
|
| Programa de Pós-Graduação: |
PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA DE ELETRICIDADE/CCET
|
| Departamento: |
DEPARTAMENTO DA ENGENHARIA DA ELETRICIDADE/CCET
|
| País: |
Brasil
|
| Palavras-chave em Português: |
|
| Palavras-chave em Inglês: |
|
| Área do conhecimento CNPq: |
|
| Link de acesso: |
https://tedebc.ufma.br/jspui/handle/tede/2163
|
Resumo: |
Convolutional Neural Networks (CNN) has been successfully used for positioning operations on standalone platforms, for environments whose scenario complexity and image pre-processing capabilities are decisive factors for the success of the classification (repositioning attitudes). The objective of the present study is to develop an autonomous approximation system with the base classification of images by a CNN. The results show the superior CNN (accuracy 82%) to a method that uses Decision threshold and Markers (accuracy 51.8%), developed an initial test of the approach system. For the generation of the database based on the virtual model and its insertion different operating scenarios was used an image processing technique characterized as \Background Subtraction", where from a control threshold, the desired object) was extracted from the \Background"(pixels of the scenario), and later inserted in another \Background" (pixels related to the desired scenario), associated with the angle values of the respective model. The final results obtained include a tool for generating a database (applied to machine learning methods) and automating the repositioning process. |
