# Overview

This repository implements OSMIS, an unconditional GAN model for one-shot synthesis of images and segmentation masks (Link to WACV 2023 Paper). The code also contains the implementation of the one-shot image synthesis baseline SIV-GAN (also was presented as One-Shot GAN). The code allows the users to reproduce and extend the results reported in the studies. Please cite the papers when reporting, reproducing or extending the results.

The code is tested for Python 3.8 and the packages listed in environment.yml. You can simply install all the packages by running the following command:

conda env create --file environment.yml
conda activate osmis

Note: due to licencing issues, the code does not provide the implementation of differentiable augmentation (DA) from StyleGAN2-ADA. To fully reproduce the paper results, the users are encouraged to incorporate DA into this repository using quidelines from here. In this case, please install the additional packages using the instructions in the original DA repository and place the DA implementation in core/differentiable_augmentation/. Alternatively, the code can be run with a provided reduced DA implementation by using the --use_kornia_augm --prob_augm 0.7 options.

Just create a name for your input image-mask pair and copy the training data under datasets/$name/. If you don't have a segmentation mask, or you simply want to run the code without it, you can also have a single image or video without masks. In this case, or in case you activate the --no_masks option, the code will automatically launch SIV-GAN training only on images. Examples of dataset structures can be found in ./datasets.

To train the model, you can use the following command:

python train.py --exp_name test_run --dataset_name example_image --num_epochs 150000 --max_size 330

In this command, the --exp_name parameter gives each experiment a unique identifier. This way, the intermediate results will be tracked in the folder ./checkpoints/$exp_name. The --max_size parameter controls the output resolution. Based on this parameter and the original image aspect ratio, the code will automatically construct a recommended model configuration,. The full list of the configuration options can be found in config.py.

If your experiment was interrupted unexpectedly, you can continue training by running

python train.py --exp_name test_run --continue_train

After training the model, you can generate images with the following command:

python test.py --exp_name test_run --which_epoch 150000 

You can select the epoch to evaluate via --which_epoch. Usually, using later epochs leads to higher overal image quality, but decreased diversity (and vice versa). By default, 100 image-mask pairs will be generated and saved in ./checkpoints/$exp_name/evaluation/$which_epoch. The number of generated pairs can be controlled via the --num_generated parameter.

To compute metrics for the set of generated images, execute

python evaluate.py --exp_name test_run --epoch 150000 

The computed SIFID, LPIPS, mIoU, and Distance to train. metrics will be saved in ./checkpoints/$exp_name/metrics as numpy files. To evaluate SIFID at different InceptionV3 layers, you can call the script with a --sifid_all_layers option.

If you use this code please cite

@article{sushko2021generating,
  title={Generating Novel Scene Compositions from Single Images and Videos},
  author={Sushko, Vadim and Zhang, Dan and Gall, Juergen and Khoreva, Anna},
  journal={arXiv preprint:2103.13389},
  year={2021}
}

@article{sushko2022one,
  title={One-Shot Synthesis of Images and Segmentation Masks},
  author={Sushko, Vadim and Zhang, Dan and Gall, Juergen and Khoreva, Anna},
  journal={arXiv preprint:2209.07547},
  year={2022}
}

This project is open-sourced under the AGPL-3.0 license. See the LICENSE file for details.

For a list of other open source components included in this project, see the file 3rd-party-licenses.txt.

This software is a research prototype, solely developed for and published as part of the publication cited above. It will neither be maintained nor monitored in any way.

Please feel free to open an issue or contact personally via email, using
[email protected]
[email protected]