FUNIT: Few-Shot Unsupervised Image-to-Image Translation
Project page | Paper | FUNIT Explained | PetSwap Demo Video | Have fun with PetSwap
Few-shot Unsueprvised Image-to-Image Translation
Ming-Yu Liu, Xun Huang, Arun Mallya, Tero Karras, Timo Aila, Jaakko Lehtinen, and Jan Kautz.
In arXiv 2019.
License
Copyright (C) 2019 NVIDIA Corporation.
All rights reserved. Licensed under the CC BY-NC-SA 4.0 (Attribution-NonCommercial-ShareAlike 4.0 International)
The code is released for academic research use only. For commercial use, please contact researchinquiries@nvidia.com.
Installation
- Clone this repo
git clone https://github.com/NVlabs/FUNIT.git
- Install CUDA10.1+
- Install cuDNN7.5
- Install Anaconda3
- Install required python pakcages
conda install -y pytorch torchvision cudatoolkit=10.0 -c pytorch
To reproduce the results reported in the paper, you would need an NVIDIA DGX1 machine with 8 V100 GPUs.
Dataset Preparation
Animal Face Dataset
We are releasing the Animal Face dataset. If you use this dataset in your publication, please cite the FUNIT paper.
- The dataset consists of image crops of the ImageNet ILSVRC2012 training set. Download the dataset and untar the files
cd dataset
wget http://www.image-net.org/challenges/LSVRC/2012/nnoupb/ILSVRC2012_img_train.tar
tar xvf ILSVRC2012_img_train.tar
- The training images should be in
datasets/ILSVRC/Data/CLS-LOC/train
. Now, extract the animal face images by running
python tools/extract_animal_faces.py datasets/ILSVRC/Data/CLS-LOC/train --output_folder datasets/animals --coor_file datasets/animal_face_coordinates.txt
- The animal face images should be in
datasets/animals
. Note there are 149 folders. Each folder contains images of one animal kind. The number of images of the dataset is 117,484. - We use 119 animal kinds for training and the ramining 30 animal kinds for evaluation.
Training New Models
Once the animal face dataset is prepared, you can train an animal face translation model by running
python train.py --config configs/funit_animals.yaml
For training a model for a different task, please create a new config file based on the example config.
Citation
If you use this code for your research, please cite our papers.
@inproceedings{liu2019few,
title={Few-shot Unsueprvised Image-to-Image Translation},
author={Ming-Yu Liu and Xun Huang and Arun Mallya and Tero Karras and Timo Aila and Jaakko Lehtinen and Jan Kautz.},
booktitle={arxiv},
year={2019}
}