Flickr1024

Flickr1024: A Large-Scale Dataset for Stereo Image Super-Resolution

Yingqian Wang  Longguang Wang  Jungang Yang  Wei An  Yulan Guo

Flickr1024 is a large-scale stereo image dataset which consists of 1024 high-quality image pairs and covers diverse senarios. Details of this dataset can be found in our published paper. Although the Flickr1024 dataset was originally developed for stereo image SR (click here for an overview), it was also used for many other tasks such as reference-based SR, stereo matching, and stereo image denoising.

Sample Images

Downloads

• The Flickr1024 dataset can be downloaded via Baidu Drive or Google Drive

Notations

• The Flickr1024 dataset is available for non-commercial use only. Therefore, You agree NOT to reproduce, duplicate, copy, sell, trade, or resell any portion of the images and any portion of derived data.
• All images on the Flickr1024 dataset are obtained from Flickr and they are not the property of our laboratory.
• We reserve the right to terminate your access to the Flickr1024 dataset at any time.

Acknowledgement

We would like to thank Sascha Becher and Tom Bentz for the approval of using their cross-eye stereo photographs.

Citiations

  @InProceedings{Flickr1024,
  author    = {Wang, Yingqian and Wang, Longguang and Yang, Jungang and An, Wei and Guo, Yulan},
  title     = {Flickr1024: A Large-Scale Dataset for Stereo Image Super-Resolution},
  booktitle = {International Conference on Computer Vision Workshops},
  pages     = {3852-3857},
  month     = {Oct},
  year      = {2019}
  }
  
  @Article{PAM,
  author  = {Wang, Longguang and Guo, Yulan and Wang, Yingqian and Liang, Zhengfa and Lin, Zaiping and Yang, Jungang and An, Wei},
  title   = {Parallax Attention for Unsupervised Stereo Correspondence Learning},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year    = {2020},
  }
  
  @inproceedings{PASSRnet,
  title     = {Learning parallax attention for stereo image super-resolution},
  author    = {Wang, Longguang and Wang, Yingqian and Liang, Zhengfa and Lin, Zaiping and Yang, Jungang and An, Wei and Guo, Yulan},
  booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages     = {12250--12259},
  year      = {2019}
  }

Related Work

The Flickr1024 dataset was used by the following works for different tasks:

Stereo Image Super-Resolution:

• Symmetric Parallax Attention for Stereo Image Super-Resolution, arXiv 2020. [pdf], [code].
• Deep Stereoscopic Image Super-Resolution via Interaction Module, TCSVT 2020.
• Parallax Attention for Unsupervised Stereo Correspondence Learning, TPAMI 2020, [pdf], [code].
• Non-Local Nested Residual Attention Network for Stereo Image Super-Resolution, ICASSP 2020, [pdf].
• Parallax-based Spatial and Channel Attention for Stereo Image Super-resolution, IEEE Access 2019. [pdf].
• Stereoscopic Image Super-Resolution with Stereo Consistent Feature, AAAI 2020. [pdf].
• ProPaCoL-Net: A Novel Recursive Stereo Image SR Network with Progressive Parallax Coherency Learning, Electronic Imaging 2020. [pdf].
• A Stereo Attention Module for Stereo Image Super-Resolution, IEEE Signal Processing Letters 2020. [pdf], [code].
• Learning Parallax Attention for Stereo Image Super-resolution, CVPR 2019. [pdf], [code].

Stereo Image Denoising:

• Self-Convolution: A Highly-Efficient Operator for Non-Local Image Restoration, arXiv 2020, [pdf].

Reference-based Image Super-Resolution:

• Feature Representation Matters: End-to-End Learning for Reference-based Image Super-resolution, ECCV 2020, [pdf].

Stereo Matching:

• Learning Stereo from Single Images, ECCV 2020, [pdf].

Other Tasks:

• Mononizing Binocular Videos, ACM Transactions on Graphics 2020, [pdf].
• Convolutional Neural Networks: A Binocular Vision Perspective, arXiv 2019. [pdf].
• Holopix50k: A Large-Scale In-the-wild Stereo Image Dataset, arXiv 2020, [pdf].
  
  

Contact

Any question regarding this work can be addressed to wangyingqian16@nudt.edu.cn.