Perceptual Losses for Real-Time Style Transfer and Super-Resolution

Code

• github.com/libreai/neural-painters-x
  tf★ 79
• github.com/ninatu/mood_challenge
  pytorch★ 16
• github.com/MartinBuessemeyer/Artistic-Texture-Control
  pytorch★ 15
• github.com/ZyoungXu/MoSt-DSA
  pytorch★ 12
• github.com/ashkanpakzad/atn
  pytorch★ 5
• github.com/vieduy/Neural-Style-Transfer
  paddle★ 4
• github.com/ksivaman/super-res
  pytorch★ 3
• github.com/anjalipemmaraju/styletransfernetwork
  pytorch★ 1
• github.com/CYetlanezi/Proyecto-Opti
  tf★ 0
• github.com/vijishmadhavan/ArtLine
  pytorch★ 0

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.

Tasks

Benchmark Results

Reproductions