Benchmarking Single-Image Reflection Removal Algorithms

Renjie Wan1      Boxin Shi2      Haoliang Li3    Yucheng Hong2      Ling-Yu Duan2      Alex C. Kot4     

1Hong Kong Baptist University      2Peking University      3City University of Hong Kong     
4Nanyang Technological University     

We propose the 'SIR2+' benchmark dataset with a large number and a great diversity of mixture images, and ground truth of background and reflection. Our dataset includes the controlled scenes taken indoor and wild scenes taken outdoor.

One part of the controlled scene is composed by a set of solid objects, which uses commonly available daily-life objects (e.g., ceramix mugs, plush toys, fruits, etc.) for both the background and the reflected scenes. The other parts of the controlled scenes use five different postcards and combines them in a pair-wise manner by using each card as background and reflection, respectively.

The wild scenes are with real-world objects of complex reflectance (car, tree leaves, glass windows, etc), various distances and scales (residential halls, gardens, and lecture room, etc), and different illuminations (direct sunlight, cloudy sky light and twilight, etc.).

News

  • April 10, 2022: The dataset (RID) of our work "CoRRN: Cooperative Reflection Removal Network" is released. This is a dataset for the reflection image captured by putting a piece of black sheet behind glass.
  • April 8, 2022: The dataset (CID) of our work "Background Scene Recovery from an Image Looking through Colored Glass" is released. This is a specific dataset for colored glass.
  • April 7, 2022: The first version of 'SIR2+' is released. We also release the original file for postcard and solid object dataset. The original file for other parts will be released.
  • April 6, 2022: The reflection image dataset (RID) of our work "CoRRN: Cooperative Reflection Removal Network" will also be released in this page.
  • April 6, 2022: The dataset (CID) of our work "Background Scene Recovery from an Image Looking through Colored Glass" will also be released in this page.
  • April 6, 2022: The dataset (FIRR) of our work "Face Image Reflection Removal" will also be released in this page.
  • March 18, 2022: The new dataset 'SIR2+' will be released soon. Besides, the complete `SIR2' dataset can be directly downloaded from this page.
  • July 21, 2017: The webpage is online now. The dataset will come soon!

    • Dataset

    'SIR2' dataset       'SIR2+' dataset       'FIRR' dataset       'CID' dataset       'RID' dataset

    Materials



    Paper


    Supplemental

    'SIR2' Citation

    @inproceedings{SIR2-iccv17,
    author = {Renjie Wan and Boxin Shi and Ling-Yu Duan and Ah-Hwee Tan and Alex C. Kot},
    title = {Benchmarking Single-Image Reflection Removal Algorithms},
    booktitle = {International Conference on Computer Vision (ICCV)},
    year = {2017}
}

    'SIR2+' Citation

    @inproceedings{SIR2pami,
    author = {Renjie Wan and Boxin Shi and Haoliang Li and Yucheng Hong and Ling-Yu Duan and Alex C. Kot},
    title = {Benchmarking Single-Image Reflection Removal Algorithms},
    booktitle = {IEEE Transacations on Pattern Analysis and Machine Intelligence},
    year = {2022}
}

    'CID' Citation

    @article{wang2022background,
  title={Background Scene Recovery from an Image Looking through Colored Glass},
  author={Wang, Ce and Xu, Dejia and Wan, Renjie and He, Bin and Shi, Boxin and Duan, Ling-Yu},
  journal={IEEE Transactions on Multimedia},
  year={2022},
  publisher={IEEE}
}

    'RID' Citation

    @article{wan2019corrn,
  title={CoRRN: Cooperative reflection removal network},
  author={Wan, Renjie and Shi, Boxin and Li, Haoliang and Duan, Ling-Yu and Tan, Ah-Hwee and Kot, Alex C},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  volume={42},
  number={12},
  pages={2969--2982},
  year={2019},
  publisher={IEEE}
}

    Discussion

    Please email us at wanpeoplejie [at] gmail.com for any inquiries.