data

CompenNet++: End-to-end Full Projector Compensation Dataset

This folder should contain CompenNet++ dataset. You can download and extract our dataset (~10G) here.

Folder Structure

├─init                     # CompenNet initialization image, i.e., \dot{s} in Eq 13.
├─light1                   # lighting level
│  ├─pos1                  # cam, prj, surface pose
│  │  ├─cloud_np           # surface texture, _np stands for nonplanar
│  │  │  ├─cam             # camera-captured images or in images camera's view
│  │  │  │  ├─desire       # desired effect, i.e., Fig. 1 (f)
│  │  │  │  │  └─test      # testing images affine warped to displayable area, i.e., z' Fig. 5.
│  │  │  │  ├─raw          # camera-captured raw images (w/o warping)
│  │  │  │  │  ├─ref       # plain color images, ref/img_0126.png is \tilde(s)
│  │  │  │  │  ├─sl        # SL images to convert raw->warpSL (two-step methods)
│  │  │  │  │  ├─test      # CompenNet++ validation images \tilde(y)
│  │  │  │  │  └─train     # CompenNet++ training images, \tilde(x)
│  │  │  │  └─warpSL       # SL warped images, only used by two-step methods
│  │  │  │      ├─ref      # SL warped raw/ref
│  │  │  │      ├─test     # SL warped raw/test
│  │  │  │      └─train    # SL warped raw/train
│  │  │  └─prj             # compensation images (z*) of cam/desire/test (z') are saved here
│  .  └─lavender_np        # another setup with another surface texture
│  .  .
│  .  .
├─light2                   # another lighting level
.  .
.  .
.  .
├─ref                      # projector input plain color images, ref/img_gray.png is x0
├─sl                       # projector input SL images (see captured in cam/raw/sl) 
├─test                     # projector input validation images, i.e., y
└─train                    # projector input training images, i.e., x

Citation

@inproceedings{huang2019compennet++,
    author = {Huang, Bingyao and Ling, Haibin},
    title = {CompenNet++: End-to-end Full Projector Compensation},
    booktitle = {IEEE International Conference on Computer Vision (ICCV)},
    month = {October},
    year = {2019} }

@inproceedings{huang2019compennet,
    author = {Huang, Bingyao and Ling, Haibin},
    title = {End-To-End Projector Photometric Compensation},
    booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    month = {June},
    year = {2019} }

Acknowledgments

We thank the anonymous reviewers for valuable and inspiring comments and suggestions. We thank the authors of the colorful textured sampling images.