Home
Uni-Logo
 

Descriptor Matching with Convolutional Neural Networks: a Comparison to SIFT

Technical Report 1405.5769, arXiv(1405.5769), May 2014
Abstract: Latest results indicate that features learned via convolutional neural networks outperform previous descriptors on classification tasks by a large margin. It has been shown that these networks still work well when they are applied to datasets or recognition tasks different from those they were trained on. However, descriptors like SIFT are not only used in recognition but also for many correspondence problems that rely on descriptor matching. In this paper we compare features from various layers of convolutional neural nets to standard SIFT descriptors. We consider a network that was trained on ImageNet and another one that was trained without supervision. Surprisingly, convolutional neural networks clearly outperform SIFT on descriptor matching.


Other associated files : 1405.5769v1.pdf [3MB]  

Images and movies

 

BibTex reference

@TechReport{FDB14,
  author       = "P. Fischer and A. Dosovitskiy and T. Brox",
  title        = "Descriptor Matching with Convolutional Neural Networks: a Comparison to SIFT ",
  institution  = "arXiv",
  number       = "1405.5769",
  month        = "May",
  year         = "2014",
  url          = "http://lmb.informatik.uni-freiburg.de/Publications/2014/FDB14"
}

Other publications in the database