Seokhun Choi

I'm a research scientist at LG Electronics, currently working as a member of the Generative AI team. I received B.S. and M.S. degrees in automotive engineering at Hanyang University in 2020 and 2022, respectively.

My research interests lie in the fields of computer vision and generative AI. I'm particularly interested in 2D/3D content generation and manipulation using diffusion models, Neural Radiance Fields (NeRF), 3D Gaussian Splatting (3DGS), and other related methods.

	PoseSyn: Synthesizing Diverse 3D Pose Data from In-the-Wild 2D Data ChangHee Yang, Hyeonseop Song, Seokhun Choi, Seungwoo Lee, Jaechul Kim, Hoseok Do Equal contribution ICCV 2025 - arXiv / Paper We present PoseSyn, a novel data synthesis framework that transforms abundant in-the-wild 2D pose datasets into diverse 3D pose image pairs, boosting the accuracy of various 3D pose estimators by up to 14% across real-world benchmarks.
	Click-Gaussian: Interactive Segmentation to Any 3D Gaussians Seokhun Choi, Hyeonseop Song, Jaechul Kim, Taehyeong Kim, Hoseok Do Equal contribution ECCV* 2024 - arXiv / Project / Video We present Click-Gaussian, a swift and precise method for interactive segmentation of 3D Gaussians using two-level granularity feature fields derived from 2D segmentation masks.
	Blending-NeRF: Text-Driven Localized Editing in Neural Radiance Fields Hyeonseop Song, Seokhun Choi, Hoseok Do, Chul Lee, Taehyeong Kim Equal contribution ICCV* 2023 - arXiv / Project / Video / Paper / Poster Blending-NeRF performs localized 3D editing to a source object by text prompts with explicitly predefined three types of editing operations: color change, density addition, and density removal.
	Vehicle Speed Optimization Based on Predicted Traction Torque Using Machine Learning Byunggun Kim, Gihoon Kim, Yoonyong Ahn, Jihoon Sung, Seokhun Choi, Youngho Jun, Kunsoo Huh Transactions of the Korean Society of Automotive Engineers, 2022 - Paper We propose vehicle speed optimization method using a network that predicts traction torque, which is trained with real-world vehicle driving data.

Source code for this page was taken from Jon Barron's website.