Taeho Kang

Computer Science and Engineering Ph.D. Student, Seoul National University.

prof_pic.jpg

I am interested in creating more expressive and interactive XR experiences. Toward this goal, I build computer vision and graphics systems for human understanding, 3D scene reconstruction, and generative 3D content. My work has explored egocentric human pose estimation, dynamic 3D scene reconstruction, and Gaussian Splatting-based representations, with the broader goal of enabling XR worlds that can understand user actions and respond in flexible, realistic ways.

news

Jun 20, 2026 LAYS: “Cross-View Yaw Estimation in Location Uncertainty with Line-Aligning Yaw Scoring” is accepted to ECCV 2026!
Aug 11, 2025 CEM-4DGS: “Clustered Error Correction with Grouped 4D Gaussian Splatting” is accepted to SIGGRAPH Asia 2025!
Jul 15, 2024 I am working with Amazon’s Last Mile team as an Applied Scientist from July 15 to Oct 18!
Feb 27, 2024 “Attention-Propagation Network for Egocentric Heatmap to 3D Pose Lifting” is accepted to CVPR 2024!
Sep 22, 2023 “SAME: Skeleton-Agnostic Motion Embedding for Character Animation” is accepted to SIGGRAPH Asia 2023, congrats to Sunmin Lee and co-authors!