├── imgs └── non-hair-FFHQ.png └── README.md /imgs/non-hair-FFHQ.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/oneThousand1000/non-hair-FFHQ/HEAD/imgs/non-hair-FFHQ.png -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # non-hair-FFHQ 2 | The non-hair-FFHQ dataset is a high-quality image dataset that contains 6,000 non-hair FFHQ portraits, based on [ stylegan2-ada](https://github.com/NVlabs/stylegan2-ada-pytorch) and [ffhq-dataset](https://github.com/NVlabs/ffhq-dataset). 3 | 4 | ![non-hair-FFHQ](./imgs/non-hair-FFHQ.png) 5 | 6 | The dataset is built by our HairMapper method. 7 | 8 | > **HairMapper: Removing Hair from Portraits Using GANs**
9 | > [Yiqian Wu](https://onethousandwu.com/), [Yongliang Yang](http://www.yongliangyang.net/), [Xiaogang Jin](http://www.cad.zju.edu.cn/home/jin)*.
2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. 10 | 11 |
12 | 13 | [![Project](https://img.shields.io/badge/HairMapper-1?label=Project&color=8B93FF)](https://onethousandwu.com/HairMapper.github.io/) 14 | [![Paper](https://img.shields.io/badge/Main%20Paper-1?color=58A399)](https://openaccess.thecvf.com/content/CVPR2022/html/Wu_HairMapper_Removing_Hair_From_Portraits_Using_GANs_CVPR_2022_paper.html) 15 | [![Suppl](https://img.shields.io/badge/Supplementary-1?color=378CE7)](http://www.cad.zju.edu.cn/home/jin/cvpr2022/Supplementary_Materials.pdf) 16 | [![Video](https://img.shields.io/badge/Video-1?color=FDA403)](https://youtu.be/UNtgpphVR2w) 17 | [![Dataset](https://img.shields.io/badge/Dataset-1?color=FC819E)](https://github.com/oneThousand1000/non-hair-FFHQ) 18 | [![Github](https://img.shields.io/github/stars/oneThousand1000/HairMapper)](https://github.com/oneThousand1000/HairMapper) 19 | 20 |
21 | 22 | 23 | We apply our method on FFHQ images (all images have licenses that allow **free use, redistribution, and adaptation for non-commercial purposes**) and present a [non-hair-FFHQ](https://github.com/oneThousand1000/non-hair-FFHQ) dataset that contains 6,000 non-hair portraits to inspire and facilitate more works in the future. 24 | 25 | ## Overview 26 | 27 | Google drive link of the dataset : https://drive.google.com/drive/folders/1CbyFYDTUqWRneyuDlVznY4XG-8pLhoAS?usp=sharing. 28 | 29 | | dir | information | 30 | | ------------------------------------------------------------ | ------------------------------- | 31 | | ├ [hair](https://drive.google.com/drive/folders/1ItALK5S9vYY6pwG_hN3sV1rrA7puivuM?usp=sharing) | original images, `{img_id}.png` | 32 | | └ [non-hair](https://drive.google.com/drive/folders/1hOp-ulk11_FismlQ8nr57HdJfJTxen_D?usp=sharing) | results images , `{img_id}.png` | 33 | 34 | ## Code 35 | 36 | https://github.com/oneThousand1000/HairMapper 37 | 38 | ## Agreement 39 | 40 | The non-hair-FFHQ dataset is available for **non-commercial research purposes** only. 41 | 42 | 43 | 44 | ## Related Works 45 | 46 | > **A Style-Based Generator Architecture for Generative Adversarial Networks** 47 | > Tero Karras (NVIDIA), Samuli Laine (NVIDIA), Timo Aila (NVIDIA) 48 | > https://arxiv.org/abs/1812.04948 49 | 50 | > **Training Generative Adversarial Networks with Limited Data** 51 | > Tero Karras, Miika Aittala, Janne Hellsten, Samuli Laine, Jaakko Lehtinen, Timo Aila 52 | > https://arxiv.org/abs/2006.06676 53 | 54 | 55 | 56 | ## Citation 57 | 58 | ``` 59 | @InProceedings{Wu_2022_CVPR, 60 | author = {Wu, Yiqian and Yang, Yong-Liang and Jin, Xiaogang}, 61 | title = {HairMapper: Removing Hair From Portraits Using GANs}, 62 | booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, 63 | month = {June}, 64 | year = {2022}, 65 | pages = {4227-4236} 66 | } 67 | ``` 68 | 69 | 70 | ## Contact 71 | 72 | [jin@cad.zju.edu.cn](mailto:jin@cad.zju.edu.cn) 73 | 74 | onethousand@zju.edu.cn 75 | 76 | onethousand1250@gmail.com 77 | 78 | --------------------------------------------------------------------------------