Depth Completion of Transparent Objects using Augmented Unpaired Data

ICRA 2023 release of depth completion of transparent objects using augmented unpaired data.

Written in Python and using TensorFlow 2+.

To install, clone this repository and run pip install -r requirements.txt. Probably best to run it in a conda, virtual or docker environment :)

Models that convert from RGBD to RGBD can be found in the folder RGBD2RGBD. Models that convert from depth to depth can be found in the folder Depth2Depth.

Some notebooks for evaluating the results are also included, but these notebooks are quite bare.

For more details (and datasets), please check our web site.

This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike (CC BY-NC-SA) license: "This license lets others remix, adapt, and build upon your work non-commercially, as long as they credit you and license their new creations under the identical terms." https://creativecommons.org/licenses/by-nc-sa/4.0/

If you want to cite this work:

@inproceedings{erich2023depth,
    title={Learning Depth Completion of Transparent Objects using Augmented Unpaired Data},
    author={Erich, Floris and Leme, Bruno and Ando, Noriaki and Hanai, Ryo and Domae, Yukiyasu},
    year=2023,
    booktitle  = {2023 IEEE International Conference on Robotics and Automation (ICRA)},
    publisher = {IEEE}
}