Jihyun Lee

CV | Google Scholar | GitHub
LinkedIn | X

Hello! 👋 I am a Postdoctoral AI Research Scientist at Meta (Codec Avatars Lab), where I am fortunate to work with Shunsuke Saito. Previously, I obtained my Ph.D. at KAIST advised by Tae-Kyun (T-K) Kim and Minhyuk Sung. In 2024, I visited Meta Reality Labs as a research scientist intern, hosted by Jason Saragih.

The goal of my research is to enable artificial intelligence to visually understand and simulate ourselves and our interactions with the environment. Lately, my focus has been on reconstructing and generating hand interactions, but my research interests are not limited to that. 🙂

Contact

jihyunlee [at] meta.com

Education

Integrated Master's & Ph.D in Computer Science, KAIST

Sep. 2020 - Aug. 2025

News

Sep 2025: I joined Meta as a Postdoctoral AI Research Scientist.
Sep 2025: MPMAvatar and ORIGEN are accepted to NeurIPS 2025. Both papers were led by talented new MSc students, whom I helped as a mentor.
Aug 2025: I’ll be giving an invited talk at the HANDS Workshop at ICCV 2025. Thank you to the workshop organizers for this great opportunity!
Jul 2025: I’ll be giving a talk at the Doctoral Consortium and presenting an invited poster on REWIND at the KCCV 2025.
Jul 2025: SCGen is accepted to ICCV 2025. Congratulations to the lead author, Take!
Jun 2025: I’m attending CVPR 2025 to present my recent work, REWIND, done during my internship at Meta. I’ll also be presenting an invited poster at the EgoVis Workshop.
Jun 2025: Successfully defended my Ph.D.! 🎓 I sincerely thank Prof. Tae-Kyun Kim, Prof. Minhyuk Sung, Prof. Seunghoon Hong, Prof. Hanbyul Joo, and Prof. Gyeongsik Moon for serving on my thesis committee. Please stay tuned for what’s next!

Publications

CV

MPMAvatar: Learning 3D Gaussian Avatars with Accurate and Robust Physics-Based Dynamics

Changmin Lee, Jihyun Lee, Tae-Kyun Kim

NeurIPS 2025

[Project] [Paper] [Code]
ORIGEN: Zero-Shot 3D Orientation Grounding in Text-to-Image Generation

Yunhong Min*, Daehyeon Choi*, Kyeongmin Yeo, Jihyun Lee, Minhyuk Sung (* equal contributions)

NeurIPS 2025

[Project] [Paper] [Code]
Generative Modeling of Shape-Dependent Self-Contact Human Poses

Takehiko Ohkawa, Jihyun Lee, Shunsuke Saito, Jason Saragih, Fabian Prada, Yichen Xu, Shoou-I Yu, Ryosuke Furuta,

Yoichi Sato, Takaaki Shiratori

ICCV 2025

[Project] [Paper]
Affordance-Guided Diffusion Prior for 3D Hand Reconstruction

Naru Suzuki, Takehiko Ohkawa, Tatsuro Banno, Jihyun Lee, Ryosuke Furuta, Yoichi Sato

ICCVW 2025

[Paper]
REWIND: Real-Time Egocentric Whole-Body Motion Diffusion with Exemplar-Based Identity Conditioning

Jihyun Lee, Weipeng Xu, Alexander Richard, Shih-En Wei, Shunsuke Saito, Shaojie Bai, Te-Li Wang, Minhyuk Sung,

Tae-Kyun Kim, Jason Saragih

CVPR 2025

[Project] [Paper]
Dense Hand-Object(HO) GraspNet with Full Grasping Taxonomy and Dynamics

Woojin Cho, Jihyun Lee, Minjae Yi, Minje Kim, Taeyun Woo, Donghwan Kim, Taewook Ha, Hyokeun Lee, Je-Hwan Ryu,

Woontack Woo, Tae-Kyun Kim

ECCV 2024

[Project] [Paper] [Code]
InterHandGen: Two-Hand Interaction Generation via Cascaded Reverse Diffusion

Jihyun Lee, Shunsuke Saito, Giljoo Nam, Minhyuk Sung, Tae-Kyun Kim

CVPR 2024

[Project] [Paper] [Code]
FourierHandFlow: Neural 4D Hand Representation Using Fourier Query Flow

Jihyun Lee, Junbong Jang, Donghwan Kim, Minhyuk Sung, Tae-Kyun Kim

NeurIPS 2023

[Project] [Paper] [Code]
Im2Hands: Learning Attentive Implicit Representation of Interacting Two-Hand Shapes

Jihyun Lee, Minhyuk Sung, Honggyu Choi, Tae-Kyun Kim

CVPR 2023

[Project] [Paper] [Code] [Presentation] [Poster]
Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images

Hangil Park*, Jihyun Lee*, Yongmin Seo, Taewon Min, Joodong Yun, Jaewon Kim, Tae-Kyun Kim (* equal contributions)

MVA 2023 (oral)
Writen as a report of an industrial research project with Samsung Display

[Paper]
Pop-Out Motion: 3D-Aware Image Deformation via Learning Shape Laplacian

Jihyun Lee*, Minhyuk Sung*, Hyunjin Kim, Tae-Kyun Kim (* equal contributions)

CVPR 2022

[Project] [Paper] [Code] [Presentation] [Poster]
Face Parsing from RGB and Depth Using Cross-Domain Mutual Learning

Jihyun Lee, Binod Bhattarai, Tae-Kyun Kim

CVPRW 2021 (oral) - IEEE AMFG acceptance rate: 27%

[Paper] [Dataset]
Fast DCTTS: Efficient Deep Convolutional Text-to-Speech

Minsu Kang, Jihyun Lee, Simin Kim, Injung Kim

ICASSP 2021

[Paper]

Awards & Honors

Outstanding Reviewer 2024, ECCV
Outstanding TA Award 2022, KAIST

Awarded to TAs who earned the highest ratings on KAIST CS student survey

Excellence Award (Mayoral Award) at Korea Software Convergence Hackathon 2019, Ministry of Science and ICT of Korea

Developed an automatic traffic light control system based on reinforcement learning

Honorable Mention Award at Undergraduate Student Paper Competition 2019, Korea Software Congress (KSC) and Korea Computer Congress (KCC)

Proposed a method to reduce the blocking effects of DCT-based X-ray grid line suppression techniques

Silver Prize at Undergraduate Student Paper Competition 2018, Korea Information Processing Society (KIPS)

Proposed a DCT-based X-ray grid line suppression method using dynamic segmentation

Academic Activities

Reviewer

CVPR, CVPR Workshops
NeurIPS
ICCV
ECCV
ICML
AAAI
IJCV
Pacific Graphics
Image and Vision Computing (Elsevier Journal)

Student Organizer

Google exploreCSR Workshops @ KAIST

Teaching Experiences

Teaching Assistant, Artificial Intelligence and Machine Learning, 2021-2023, KAIST
Teaching Assistant, Machine Learning for Computer Vision, 2021-2022, KAIST
Teaching Assistant, Operating System, 2020, Handong Global University
Teaching Assistant, C++ Programming, 2019, Handong Global University
Teaching Assistant, Data Structure, 2018, Handong Global University