/MAANet

MAANet: Multi-view Aware Attention Networks for Image Super-Resolution

Primary LanguagePythonMIT LicenseMIT

MAANet

MAANet: Multi-view Aware Attention Networks for Image Super-Resolution implementation in Pytorch.

Deep Convolutional Neural Networks (DCNNs) based Image Super-Resolution (SR) with Local and Global Aware Attention modules.
More details can be found in the the paper.

Requirements

  • Pytorch 1.7 +
  • Torchvision 0.8.2 +
  • Numpy 1.14 +
  • Scikit-Image 0.16 +

To Do

  • Train model for 106 Epochs.

Dataset

DIV2K: DIVerse 2K resolution high quality images.
Dataset can be downloaded from this link.

@InProceedings{Agustsson_2017_CVPR_Workshops,
  author = {Agustsson, Eirikur and Timofte, Radu},
  title = {NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study},
  booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
  month = {July},
  year = {2017}
}

Citation

Original Paper and Architecture details can be found in the url below.

@article{DBLP:journals/corr/abs-1904-06252,
  author    = {Jingcai Guo and
               Shiheng Ma and
               Song Guo},
  title     = {MAANet: Multi-view Aware Attention Networks for Image Super-Resolution},
  journal   = {CoRR},
  volume    = {abs/1904.06252},
  year      = {2019},
  url       = {http://arxiv.org/abs/1904.06252},
  archivePrefix = {arXiv},
  eprint    = {1904.06252},
  timestamp = {Sat, 23 Jan 2021 01:19:38 +0100},
  biburl    = {https://dblp.org/rec/journals/corr/abs-1904-06252.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}