Home
Uni-Logo
 

Generating Images with Perceptual Similarity Metrics based on Deep Networks

Advances in Neural Information Processing Systems (NIPS), 2016
Abstract: We propose a class of loss functions, which we call deep perceptual similarity metrics (DeePSiM), allowing to generate sharp high resolution images from compressed abstract representations. Instead of computing distances in the image space, we compute distances between image features extracted by deep neural networks. This metric reflects perceptual similarity of images much better and, thus, leads to better results. We demonstrate two examples of use cases of the proposed loss: (1) networks that invert the AlexNet convolutional network; (2) a modified version of a variational autoencoder that generates realistic high-resolution random images.
Downloads

Other associated files : inverting_GAN_nips2016_final.pdf [2.6MB]   inverting_GAN_nips2016_supp_final.pdf [13.2MB]  

Images and movies

 

BibTex reference

@InProceedings{DB16c,
  author       = "A. Dosovitskiy and T. Brox",
  title        = "Generating Images with Perceptual Similarity Metrics based on Deep Networks",
  booktitle    = "Advances in Neural Information Processing Systems (NIPS)",
  month        = " ",
  year         = "2016",
  url          = "http://lmbweb.informatik.uni-freiburg.de/Publications/2016/DB16c"
}

Other publications in the database