Skip to content
main
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
package License Project Status: Active
citation pyOpenSci JOSS
code black mypy Lint status via GitHub Actions
tests Test status via GitHub Actions Test coverage
docs Docs status via GitHub Actions Latest documentation hosted on Read the Docs

pystiche logo

pystiche

pystiche (pronounced /ˈpaɪˈstiʃ/ ) is a framework for Neural Style Transfer (NST) built upon PyTorch. The name of the project is a pun on pastiche meaning:

A pastiche is a work of visual art [...] that imitates the style or character of the work of one or more other artists. Unlike parody, pastiche celebrates, rather than mocks, the work it imitates.

pystiche banner

pystiche has similar goals as Deep Learning (DL) frameworks such as PyTorch:

  1. Accessibility
    Starting off with NST can be quite overwhelming due to the sheer amount of techniques one has to know and be able to deploy. pystiche aims to provide an easy-to-use interface that reduces the necessary prior knowledge about NST and DL to a minimum.
  2. Reproducibility
    Implementing NST from scratch is not only inconvenient but also error-prone. pystiche aims to provide reusable tools that let developers focus on their ideas rather than worrying about bugs in everything around it.

Installation

pystiche is a proper Python package and can be installed with pip. The latest release can be installed with

pip install pystiche

Usage

pystiche makes it easy to define the optimization criterion for an NST task fully compatible with PyTorch. For example, the banner above was generated with the following criterion:

from pystiche import enc, loss

mle = enc.vgg19_multi_layer_encoder()

perceptual_loss = loss.PerceptualLoss(
    content_loss=loss.FeatureReconstructionLoss(
        mle.extract_encoder("relu4_2")
    ),
    style_loss=loss.MultiLayerEncodingLoss(
        mle,
        ("relu1_1", "relu2_1", "relu3_1", "relu4_1", "relu5_1"),
        lambda encoder, layer_weight: ops.GramOLoss(
            encoder, score_weight=layer_weight
        ),
        score_weight=1e3,
    ),
)

For the full example, head over to the example NST with pystiche.

Documentation

For

or anything else, head over to the documentation.

Citation

If you use this software, please cite it as

@Article{ML2020,
  author  = {Meier, Philip and Lohweg, Volker},
  journal = {Journal of Open Source Software {JOSS}},
  title   = {pystiche: A Framework for Neural Style Transfer},
  year    = {2020},
  doi     = {10.21105/joss.02761},
}