Home
Uni-Logo
 

Learning to Generate Chairs, Tables and Cars with Convolutional Networks

IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(4): 692-705, Apr 2017
Abstract: We train generative 'up-convolutional' neural networks which are able to generate images of objects given object style, viewpoint, and color. We train the networks on rendered 3D models of chairs, tables, and cars. Our experiments show that the networks do not merely learn all images by heart, but rather find a meaningful representation of 3D models allowing them to assess the similarity of different models, interpolate between given views to generate the missing ones, extrapolate views, and invent new objects not present in the training set by recombining training instances, or even two different object classes. Moreover, we show that such generative networks can be used to find correspondences between different objects from the dataset, outperforming existing approaches on this task.
Publisher's link

Other associated files : Chairs_PAMI.pdf [9.3MB]  

Images and movies

 

BibTex reference

@Article{DTB17,
  author       = "A. Dosovitskiy and J. T. Springenberg and M. Tatarchenko and T. Brox",
  title        = "Learning to Generate Chairs, Tables and Cars with Convolutional Networks",
  journal      = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
  number       = "4",
  volume       = "39",
  pages        = "692-705",
  month        = "Apr",
  year         = "2017",
  url          = "http://lmb.informatik.uni-freiburg.de/Publications/2017/DTB17"
}

Other publications in the database