Code for the paper: Image-to-Voxel Model Translation for 3D Scene Reconstruction and Segmentation This is the PyTorch implementation of the color-to-voxel model translation presented on ECCV 2018.

The code is based on the PyTorch implementation of the pix2pix and CycleGAN.

If you use this code for your research, please cite:

@InProceedings{Kniaz2020,
author="Kniaz, Vladimir A. and
Knyaz, Vladimir V. and Fabio Remondino and
Artem Bordodymov and Petr Moshkantsev
title={SSZ: Image-to-Voxel Model Translation for 3D Scene Reconstruction and Segmentation},
booktitle={{Computer Vision -- ECCV 2020 Workshops",
year="2020}},
publisher={Springer International Publishing},
}

• Linux or macOS
• Python 2 or 3
• CPU or NVIDIA GPU + CUDA CuDNN

• Install PyTorch and dependencies from http:https://pytorch.org
• Install Torch vision from the source.

git clone https://github.com/pytorch/vision
cd vision
python setup.py install

• Install python libraries visdom and dominate.

pip install visdom
pip install dominate

• Clone this repo:

git clone https://github.com/vlkniaz/SSZ

• Go to the repo directory

cd SSZ

• Download a SSZ dataset:

bash ./datasets/download_ssz_dataset.sh mini

• Train a model:

bash scripts/train_ssz.sh

• To view training results and loss plots, run python -m visdom.server and click the URL http:https://localhost:8097. To see more intermediate results, check out ./checkpoints/thermal_gan_rel/web/index.html
• Test the model:

bash scripts/test_ssz.sh

The test results will be saved to a html file here: ./results/ssz/test_latest/index.html.

Download a pre-trained model with ./pretrained_models/download_ssz_model.sh.

• For example, if you would like to download SSZ model on the mini dataset,

bash pretrained_models/download_ssz_model.sh SSZ

• Download the mini datasets

bash ./datasets/download_ssz_dataset.sh mini

• Then generate the results using

bash scripts/test_ssz_pretrained.sh

• The test results will be saved to a html file here: ./results/SSZ_pretrained/test_latest/index.html.