Abstract

We consider image transformation problems, where an input image is transformed into an output image. Recent methods for such problems typically train feed-forward convolutional neural networks using a per-pixel loss between the output and ground-truth images. Parallel work has shown that high-quality images can be generated by defining and optimizing perceptual loss functions based on high-level features extracted from pretrained networks. We combine the benefits of both approaches, and propose the use of perceptual loss functions for training feed-forward networks for image transformation tasks. We show results on image style transfer, where a feed-forward network is trained to solve the optimization problem proposed by Gatys et al. in real-time. Compared to the optimization-based method, our network gives similar qualitative results but is three orders of magnitude faster. We also experiment with single-image super-resolution, where replacing a per-pixel loss with a perceptual loss gives visually pleasing results.
To appear in ECCV 2016

Code and Extras

You can find the code on Github, including:
  • Training/test code (uses Torch/Lua)
  • Pretrained models
  • Live webcam demo

Bibtex

@inproceedings{Johnson2016Perceptual,
  title={Perceptual losses for real-time style transfer and super-resolution},
  author={Johnson, Justin and Alahi, Alexandre and Fei-Fei, Li},
  booktitle={European Conference on Computer Vision},
  year={2016}
}

Example Results: Style Transfer

The Starry Night,
Vincent Van Gogh,
1889
Original Gatys et al. Ours
The Great Wave off Kanagawa,
Hokusai,
1829-1832
Original Gatys et al. Ours
Composition VII,
Wassily Kandinsky,
1913
Original Gatys et al. Ours
The Muse,
Pablo Picasso,
1935
Original Gatys et al. Ours