diff --git a/_pages/about.md b/_pages/about.md index b913823bbfaea..9c1b27c8fa721 100644 --- a/_pages/about.md +++ b/_pages/about.md @@ -27,13 +27,19 @@ I'm Yiqian Wu (吴奕谦), a second-year (2021-now) Ph.D. student in State Key L ## 2023 -1. [LPFF: A Portrait Dataset for Face Generators Across Large Poses](https://onethousandwu.com/publication/lpff-dataset) +1. [3DPortraitGAN: Learning One-Quarter Headshot 3D GANs from a Single-View Portrait Dataset with Diverse Body Poses](https://onethousandwu.com/publication/3DPortraitGAN) + + Preprint + + **Yiqian Wu**, Hao Xu, Xiangjun Tang, Hongbo Fu, Xiaogang Jin* + +2. [LPFF: A Portrait Dataset for Face Generators Across Large Poses](https://onethousandwu.com/publication/lpff-dataset) 2023 IEEE/CVF International Conference on Computer Vision (ICCV) **Yiqian Wu**, Jing Zhang, Hongbo Fu, Xiaogang Jin. -2. [Deep Real-time Volumetric Rendering Using Multi-feature Fusion](https://onethousandwu.com/publication/mrpnn) +3. [Deep Real-time Volumetric Rendering Using Multi-feature Fusion](https://onethousandwu.com/publication/mrpnn) ACM SIGGRAPH 2023 Conference Proceedings (SIGGRAPH '23). Association for Computing Machinery, New York, NY, USA, Article 61, 1–10. diff --git a/_publications/3DPortraitGAN.md b/_publications/3DPortraitGAN.md new file mode 100644 index 0000000000000..bf344a26613c8 --- /dev/null +++ b/_publications/3DPortraitGAN.md @@ -0,0 +1,35 @@ +--- +title: "3DPortraitGAN: Learning Canonical Full-Head 3D GANs from a Single-View Portrait Dataset with Diverse Body Poses" +collection: publications +permalink: /publication/3DPortraitGAN +excerpt: '**Yiqian Wu**, [Hao Xu](https://xh38.github.io/), [Xiangjun Tang](https://yuyujunjun.github.io/), [Hongbo Fu](http://sweb.cityu.edu.hk/hongbofu/publications.html), [Xiaogang Jin*](http://www.cad.zju.edu.cn/home/jin)' +date: 2023-08-22 +venue: 'Preprints' +paperurl: 'coming soon' +citation: 'coming soon' +code: 'https://github.com/oneThousand1000/3DPortraitGAN' +video: 'coming soon' +supplementary_materials: 'https://drive.google.com/file/d/16aNE5USZ0U32bgGJS1G5xWrY0oIMTfre/view?usp=sharing' +project_page: 'https://github.com/oneThousand1000/3DPortraitGAN' +year: '2023' +--- +![coarse2fine](http://oneThousand1000.github.io/images/publications/3DPortraitGAN.png) + +Abstract: + +3D-aware face generators are typically trained on 2D real-life face image datasets that primarily consist of near-frontal face data, and as such, they are unable to construct [one-quarter headshot](https://www.backstage.com/magazine/article/types-of-headshots-75557/) 3D portraits with complete head, neck, and shoulder geometry. Two reasons account for this issue: First, existing facial recognition methods struggle with extracting facial data captured from large camera angles or back views. Second, it is challenging to learn a distribution of 3D portraits covering the one-quarter headshot region from single-view data due to significant geometric deformation caused by diverse body poses. To this end, we first create the dataset 360°-Portrait-HQ (360°PHQ for short) which consists of high-quality single-view real portraits annotated with a variety of camera parameters (the yaw angles span the entire 360° range) and body poses. We then propose 3DPortraitGAN, the first 3D-aware one-quarter headshot portrait generator that learns a canonical 3D avatar distribution from the 360°PHQ dataset with body pose self-learning. Our model can generate view-consistent portrait images from all camera angles with a canonical one-quarter headshot 3D representation. Our experiments show that the proposed framework can accurately predict portrait body poses and generate view-consistent, realistic portrait images with complete geometry from all camera angles. + +[Paper]() + +[Video]() + +[Suppl](https://drive.google.com/file/d/16aNE5USZ0U32bgGJS1G5xWrY0oIMTfre/view?usp=sharing) + +[Project Page](https://github.com/oneThousand1000/3DPortraitGAN) + + + +Recommended citation: +``` +coming soon +``` diff --git a/images/publications/3DPortraitGAN.png b/images/publications/3DPortraitGAN.png new file mode 100644 index 0000000000000..dce35e0c03e0d Binary files /dev/null and b/images/publications/3DPortraitGAN.png differ