A study of the effect of noise and occlusion on the accuracy of convolutional neural networks applied to 3D object recognition

Por favor, use este identificador para citar o enlazar este ítem: http://hdl.handle.net/10045/72633
Registro completo de metadatos
Registro completo de metadatos
Campo DCValorIdioma
dc.contributorInformática Industrial y Redes de Computadoreses_ES
dc.contributorRobótica y Visión Tridimensional (RoViT)es_ES
dc.contributor.authorGarcia-Garcia, Alberto-
dc.contributor.authorGarcia-Rodriguez, Jose-
dc.contributor.authorOrts-Escolano, Sergio-
dc.contributor.authorOprea, Sergiu-
dc.contributor.authorGomez-Donoso, Francisco-
dc.contributor.authorCazorla, Miguel-
dc.contributor.otherUniversidad de Alicante. Departamento de Tecnología Informática y Computaciónes_ES
dc.contributor.otherUniversidad de Alicante. Departamento de Ciencia de la Computación e Inteligencia Artificiales_ES
dc.contributor.otherUniversidad de Alicante. Instituto Universitario de Investigación Informáticaes_ES
dc.identifier.citationComputer Vision and Image Understanding. 2017, 164: 124-134. doi:10.1016/j.cviu.2017.06.006es_ES
dc.identifier.issn1077-3142 (Print)-
dc.identifier.issn1090-235X (Online)-
dc.description.abstractIn this work, we carry out a study of the effect of adverse conditions, which characterize real-world scenes, on the accuracy of a Convolutional Neural Network applied to 3D object class recognition. Firstly, we discuss possible ways of representing 3D data to feed the network. In addition, we propose a set of representations to be tested. Those representations consist of a grid-like structure (fixed and adaptive) and a measure for the occupancy of each cell of the grid (binary and normalized point density). After that, we propose and implement a Convolutional Neural Network for 3D object recognition using Caffe. At last, we carry out an in-depth study of the performance of the network over a 3D CAD model dataset, the Princeton ModelNet project, synthetically simulating occlusions and noise models featured by common RGB-D sensors. The results show that the volumetric representations for 3D data play a key role on the recognition process and Convolutional Neural Network can be considerably robust to noise and occlusions if a proper representation is chosen.es_ES
dc.description.sponsorshipThis work has been supported by the Spanish Government DPI2013-40534-R grant for the SIRMAVED project, also supported with FEDER funds. This work has also been funded by the grant “Ayudas para Estudios de Máster e Iniciación a la Investigación” from the University of Alicante.es_ES
dc.rights© 2017 Elsevier Inc.es_ES
dc.subjectDeep learninges_ES
dc.subject3D object recognitiones_ES
dc.subjectConvolutional neural networkses_ES
dc.subject.otherArquitectura y Tecnología de Computadoreses_ES
dc.subject.otherCiencia de la Computación e Inteligencia Artificiales_ES
dc.titleA study of the effect of noise and occlusion on the accuracy of convolutional neural networks applied to 3D object recognitiones_ES
Aparece en las colecciones:INV - I2RC - Artículos de Revistas
INV - RoViT - Artículos de Revistas

Archivos en este ítem:
Archivos en este ítem:
Archivo Descripción TamañoFormato 
Thumbnail2017_Garcia-Garcia_etal_CompVisImageUnderst_final.pdfVersión final (acceso restringido)2,74 MBAdobe PDFAbrir    Solicitar una copia
Thumbnail2017_Garcia-Garcia_etal_CompVisImageUnderst_preprint.pdfPreprint (acceso abierto)3,62 MBAdobe PDFAbrir Vista previa

Todos los documentos en RUA están protegidos por derechos de autor. Algunos derechos reservados.