Yangyi Huang (黄洋逸)

I am a third-year Master candidate(2021.09 - ) at State Key Lab of CAD&CG, College of Computer Science and Technology at Zhejiang University, advised by Prof. Deng Cai.

Recently, I have had the opportunity to work closely with Hongwei Yi, Yuliang Xiu and Justus Thies on exciting research projects involving the generation and reconstruction of digital humans.

Before that, I received my B.Eng(2017.09 - 2021.06) in Computer Science from the same department with an honor degree at Mixed Class, Chu Kochen Honor College. During my undergraduate studies, I enjoyed the challenges of competitive programming and participated in events such as ACM-ICPC and CCPC.

Email  /  CV  /  Google Scholar  /  Twitter  /  Github

profile photo



My research interest lies in computer vision, neural rendering and generative modeling. Currently I mainly focus on creating 3D contents and modelling 3D humans. Representative papers/projects are highlighted.

TeCH: Text-guided Reconstruction of Lifelike Clothed Humans
Yangyi Huang*, Hongwei Yi*, Yuliang Xiu*, Tingting Liao, Jiaxiang Tang, Deng Cai, Justus Thies (* denotes equal contribution)
3DV, 2024
code / arXiv / project / video

We reconstruct high-resolution textured meshes for clothed humans from single images, with textual guidance from a VQA model and a few-shot finetuned T2I model.

TADA! Text to Animatable Digital Avatars
Tingting Liao*, Hongwei Yi*, Yuliang Xiu, Jiaxiang Tang, Yangyi Huang, Justus Thies, Michael J. Black
3DV, 2024
code / arXiv / project / video

We create expressive 3D avatars from text, based on T2I (I: image; T: text) model.

One-shot Implicit Animatable Avatars with Model-based Priors
Yangyi Huang*, Hongwei Yi*, Weiyang Liu, Haofan Wang, Boxi Wu, Wenxiao Wang, Binbin Lin, Debing Zhang, Deng Cai(* denotes equal contribution)
ICCV, 2023
code / arXiv / project

We create a animatable NeRF-based avatar reconstruction from a single image with model-based prior.

BEVFormer++: Improving BEVFormer for 3D Camera-only Object Detection
Zhiqi Li*, Hanming Deng*, Tianyu Li*, Yangyi Huang*, Chonghao Sima*, Xiangwei Geng*, Yulu Gao*, Wenhai Wang*, Yang Li, Lewei Lu
1st place solution for Waymo Open Dataset Challenge 2022

We enhanced BEVFormer, a DETR-based 3D detection model, with efficient techniques.

FuseFormer: Fusing Fine-Grained Information in Transformers for Video Inpainting
Rui Liu*, Hanming Deng*, Yangyi Huang*, Xiaoyu Shi, Lewei Lu Wenxiu Sun, Xiaogang Wang, Jifeng Dai, Hongsheng Li (* denotes equal contribution)
ICCV, 2021
code / arXiv

We proposed a Transformer-based model to inpaint videos with sharp and realistic results.

This site is based on Jon Barron's template.