GRE

Environment setup

conda create -n gre python=3.9
conda activate gre
pip install torch==2.0.0+cu117 torchvision==0.15.1+cu117 -f https://download.pytorch.org/whl/torch_stable.html
pip install -r requirements.txt

Download pre-trained models

Download the data folder as a zip file and unzip.

Download dataset

Download the dataset, unzip the files, and place it to the same path as well. Put the images into the test folder if they are split into test1.zip and test2.zip. Run link.sh to create soft links for the dataset.

Testing

Make sure the configuration in test.sh is correct and run

sh test.sh

Training

Make sure the configuration in train.sh is correct and run

sh train.sh

Acknowledgement

Our code is developed based on DIRE. The dataset is generated by 3D Gaussian Splatting, mip-splatting, Compact-3DGS, ges-splatting, Scaffold-GS, and GS-IR.