Donghwan Seo, Abhijith Punnappurath, Luxi Zhao, Abdelrahman Abdelhamed, Sai Kiran Tedla, Sanguk Park, Jihwan Choe, and Michael S. Brown
- The code was tested on Ubuntu 18.04, Python3.7, CUDA 11.3, cuDNN 8.2, Pytorch 1.10
- The code may work in other environments
- Install requirements
pip install -r requirements.txt
Our graphics dataset can be downloaded from here.
In addition to the full dataset, a small set of sample images is provided for a sneak peek.
python3 -m data_generation.package_exr_to_dng_graphics2raw --exr_folder_path path/to/graphics2RAW-dataset-samples --save_path path/to/outputs
Data generation, training and testing instructions for the downstream tasks:
We would like to express our special thanks to SiDi collaborating with Samsung for the graphics data.
If you use this code or the associated data, please cite the paper:
@InProceedings{Seo_2023_ICCV,
author = {Seo, Donghwan and Punnappurath, Abhijith and Zhao, Luxi and Abdelhamed, Abdelrahman and Tedla, Sai Kiran and Park, Sanguk and Choe, Jihwan and Brown, Michael S.},
title = {Graphics2RAW: Mapping Computer Graphics Images to Sensor RAW Images},
booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
month = {October},
year = {2023},
pages = {12622-12631}
}