Ready-to-use artistic deep learning algorithms
Project description
Neurartist
A ready-to-use implementation of various Artistic Deep Learning Algorithms.
Image Style Transfer Using Convolutional Neural Networks, Gatys et. al, 2016
Installation
# It is recommended to install torch/torchvision manually before this command, according to your hardware configuration (see below) pip install neurartist
Please note that the use of a GPU is recommended, as CNN computations are pretty slow on a CPU.
NB for GPU users: pip ships torch/torchvision with the Cuda Toolkit 9.0. If you use a more recent version of the Cuda Toolkit, see the PyTorch website for instructions on PyTorch installation with another version of the toolkit.
Usage
Console entrypoint
# Then see the builtin help for usage details neurartist --help
Library
import neurartist
To be added.
Development
Anaconda is strongly recommended:
conda create python=3.7 --name neurartist_env conda activate neurartist_env # with gpu conda install pytorch torchvision cudatoolkit=<your cudatoolkit version> -c pytorch conda install --file requirements.txt # with cpu conda install pytorch-cpu torchvision-cpu -c pytorch conda install --file requirements.txt
You can then run the main entrypoint directly using:
python -m neurartist --help
Or build and install the wheel file with the --editable flag.
TODO
Documentation.
Implement the remaining parts of the jupyter notebook.
Semantic segmentation as described in this article as to limit spillovers: different approach than guided gram matrices, but same idea of using spatial guidance channels that describe a semantic segmentation of our images.
More deep-artistic algorithms.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for neurartist-0.1-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 33dbf7d4fd275ac5e5f7edcfd1ccafbf4d0b657f555d5487d04be4ca14216cd3 |
|
MD5 | 541d8f947ffaeb8deef0ed571c3a2e63 |
|
BLAKE2b-256 | 3967d9c88660877875ca5e83997a1139520c0ad5bbb68d7be0982a246c0bc96a |