Kim Youwang
youwang.kim@postech.ac.kr

I am a Ph.D. student at AMILab in Electrical Engineering at POSTECH, supervised by Tae-Hyun Oh. Currently, I am also a visiting researcher at Real Virtual Humans Group at the Univ. of Tübingen, working with Gerard Pons-Moll. I am also an incoming research scientist intern at Meta Reality Labs, hosted by Yaser Sheikh.

I received my bachelor's degree in Electrical Engineering from POSTECH. During my bachelor, I spent a semester at Nanyang Technological University (NTU) as an exchange student.

CV  |  GitHub  |  Google Scholar  |  LinkedIn  |  Twitter

headshot
News
  • 02/2024 (New!) Our paper on text-driven PBR texture synthesis is accepted to CVPR 2024.
  • 02/2024 (New!) I'll be joining Meta Reality Labs - Codec Avatars team as a research scientist intern, this fall.
  • 12/2023 Our paper on efficient 3D scene stylization is accepted to AAAI 2024.
  • 11/2023 I won the Grand Prize (1st place, Minister's award, $12,000) at ICT paper awards.
  • 10/2023 I am selected as an outstanding reviewer for ICCV 2023.
  • 10/2023 I am joining Real Virtual Humans Group as a visiting Ph.D. student.
  • ---- show more ----
Research

I am interested in 3D/4D virtual humans, animals, and objects. My goal is to enrich a virtual world through realistic and creative 3D/4D content generation. Currently, I'm focusing on the topics below, but not limited to.

  • Realistic 3D/4D human, animal and object generation for virtual world.
  • Accurate 3D pose and shape reconstruction of humans and animals.
🎨 Paint-it: Text-to-Texture Synthesis via Deep Convolutional Texture Map Optimization and Physically-Based Rendering

Kim Youwang, Tae-Hyun Oh, Gerard Pons-Moll
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024
[Project page] [Paper] [Code]

• Synthesizing realistic 3D mesh PBR texture from a text.
CLIP-Actor-X: Text-driven 4D Human Avatar Generation via Cross-modal Synthesis-through-Optimization

Kim Youwang*, Taehyun Byun*, Kim Ji-Yeon, Sungjoon Choi, Tae-Hyun Oh
Journal under review

• An improved text-to-4D textured human avatar synthesis.
ObjectDR: Object-Centric Domain Randomization for 3D Shape Reconstruction in the Wild

Junhyeong Cho, Kim Youwang, Hunmin Yang, Tae-Hyun Oh
arXiv preprint, 2024
[Project page] [Paper]

• Learns to capture a domain-invariant geometry prior via synthetic 3D dataset.
(Ongoing) Monocular 3D object mesh reconstruction

• Reconstructing 3D textured mesh from a single image.

Feed-Forward Photorealistic Style Transfer for Large-Scale 3D Neural Radiance Fields

GeonU Kim, Kim Youwang, Tae-Hyun Oh
AAAI Conference on Artificial Intelligence (AAAI), 2024
[Project page] [Paper] [Code]

• Efficient feed-forward stylization method for 3D scenes.
A Large-Scale 3D Face Mesh Video Dataset via Neural Re-parameterized Optimization

Kim Youwang, Lee Hyun*, Kim Sung-Bin*, Suekyeong Nam, Janghoon Ju, Tae-Hyun Oh
arXiv preprint, 2023
[Project page] [Paper]

• Reliable 3D face mesh labels for large-scale facial video datasets.
Text-driven Human Avatar Generation by Neural Re-parameterized Texture Optimization

Kim Youwang, Tae-Hyun Oh
IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2023

• Generating high quality 4D human avatars from a single text.

STREAM: Spatio-Temporally Consistent Face Mesh Reconstruction on Videos

Kim Youwang, Lee Hyun*, Kim Sung-Bin*, Suekyeong Nam, Janghoon Ju, Tae-Hyun Oh
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2023

• Reconstructing multi-view & temporally consistent 3D face meshes on videos.
Multi-stage Adaptive Rank Statistic Pruning for Lightweight Human 3D Mesh Recovery Model

Dong Hun Ryou, Kim Youwang, Tae-Hyun Oh
The Visual Computer Journal (TVCJ), 2023
[Paper]

• Neural network pruning for efficient 3D human mesh reconstruction.
CLIP-Actor: Text-Driven Recommendation and Stylization for Animating Human Meshes

Kim Youwang*, Kim Ji-Yeon*, Tae-Hyun Oh
European Conference on Computer Vision (ECCV), 2022
[Project page] [Paper] [Video] [Code] [Poster]

• Text-to-4D textured human avatar synthesis.
- Grand Prize (1st place, Minister's award, $12,000) at ICT paper awards 2023
- Qualcomm Innovation Award Winner 2022
FastMETRO: Cross-Attention of Disentangled Modalities for 3D Human Mesh Recovery with Transformers

Junhyeong Cho, Kim Youwang, Tae-Hyun Oh
European Conference on Computer Vision (ECCV), 2022
[Project page] [Paper] [Code] [Poster]

• A lightweight and fast transformer for human mesh reconstruction.
Unified 3D Mesh Recovery of Humans and Animals by Learning Animal Exercise

Kim Youwang, Kim Ji-Yeon, Kyungdon Joo, Tae-Hyun Oh
British Machine Vision Conference (BMVC), 2021
[Project page] [Paper]

• An efficient multi-task model for multi-class mesh reconstruction.
Invited Talks
  • Towards Efficient & Realistic Virtual World Communication, INNERVERZ, Korea, 2023
Academic Services
  • Conference Reviewer: ICCV (Outstanding reviewer in 2023), CVPR (2024), ECCV (2024)
  • Journal Reviewer: TMM (2023)
Misc.
    +) As a guitarist of the band STEELER, I also upload some music covers or gigs.

template adapted from this awesome website