000253086 001__ 253086
000253086 005__ 20190317000932.0
000253086 0247_ $$2doi$$a10.1109/ICCVW.2017.106
000253086 037__ $$aCONF
000253086 245__ $$aGraph-Based Classification of Omnidirectional Images
000253086 260__ $$c2017-10-29
000253086 269__ $$a2017-10-29
000253086 336__ $$aConference Papers
000253086 520__ $$aOmnidirectional cameras are widely used in such areas as robotics and virtual reality as they provide a wide field of view. Their images are often processed with classical methods, which might unfortunately lead to non-optimal solutions as these methods are designed for planar images that have different geometrical properties than omnidirectional ones. In this paper we study image classification task by taking into account the specific geometry of omnidirectional cameras with graph-based representations. In particular, we extend deep learning architectures to data on graphs; we propose a principled way of graph construction such that convolutional filters respond similarly for the same pattern on different positions of the image regardless of lens distortions. Our experiments show that the proposed method outperforms current techniques for the omnidirectional image classification problem.
000253086 6531_ $$aOmnidirectional camera, deep learning, deep learning on graphs
000253086 700__ $$aKhasanova, Renata
000253086 700__ $$0241061$$aFrossard, Pascal
000253086 7112_ $$dOctober 22-29, 2017$$cVenice, Italy$$aIEEE International Conference on Computer Vision Workshops (ICCVW)
000253086 773__ $$tProceedings of ICCV Workshops$$q860-869
000253086 8560_ $$fpascal.frossard@epfl.ch
000253086 8564_ $$uhttps://infoscience.epfl.ch/record/253086/files/ICCVW2017.pdf$$zPREPRINT$$s1394414
000253086 8564_ $$uhttps://infoscience.epfl.ch/record/253086/files/ICCVW2017.pdf?subformat=pdfa$$zPREPRINT$$s3304449$$xpdfa
000253086 909C0 $$xU10851$$pLTS4$$0252393
000253086 909CO $$ooai:infoscience.epfl.ch:253086$$qGLOBAL_SET$$pconf$$pSTI
000253086 960__ $$arenata.khasanova@epfl.ch
000253086 961__ $$apierre.devaud@epfl.ch
000253086 973__ $$rREVIEWED$$aEPFL
000253086 980__ $$aCONF
000253086 981__ $$aoverwrite