/SR-DARTS

An implementation of DARTS on Super-Resolution Task.

Primary LanguagePython

SRdarts: An implementation of DARTS on Super-Resolution Task

Description

SRdarts is an implementation of DARTS on Super-Resolution Task, we use DARTS as the method baseline.

Our code is built on EDSR and Meta-SR-Pytorch.

Download

Code

Clone the repository into your place and let's get started!

git clone https://github.com/GeniusGaryant/SRdarts.git
cd SRdarts

Datasets

We use DIV2K dataset as our train/valid dataset during the search.

For the test dataset, we use widely-used benchmark datasets as follows:

Set5、Set14、B100、Urban100

Installation

Requirements

  • Python 3.6

  • PyTorch 1.1 from a nightly release. Installation instructions can be found in PyTorch.

  • torchvision

  • cuda9.0

  • numpy

  • scikit-image

  • imageio

  • matplotlib

  • tb-nightly

  • future

Step by step installation

# first, make sure that your conda is setup properly with the right environment
# for that, check that `which conda`, `which pip` and `which python` points to the
# right path. From a clean conda env, this is what you need to do
conda create -n srdarts python=3.6
source activate srdarts

# install torchvision and pytorch1.1
pip install numpy torchvision_nightly
pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cu90/torch_nightly.html

conda install scikit-image
conda install imageio

# install tb-nightly for tensorboard part
pip install tb-nightly
pip install future

Contributing to the Project

Any pull requests or issues are welcome. The repository is still updating.

License

For academic use, this project is licensed under the 2-clause BSD License - see the LICENSE file for details. For commercial use, please contact the authors.