BYEONG GWON LEE / 이병권

3D Computer
Vision Researcher

3D Reconstruction · Gaussian Splatting · SLAM · Multi-View Stereo

Turning images into geometry.
I work on depth, mapping, and 3D reconstruction.

KETI Researcher · Sep. 2026 – Present

IJCAI 2025 · First Author Online 3D Gaussian Splatting Mobility One Inc. · Research project Drone-based 3D reconstruction
Explore research CV (PDF)
Email Google Scholar GitHub LinkedIn
GEOMETRY LAB / 01 Illustrative reconstruction
Illustrative building point cloud with camera frustums following a curved capture trajectory
Scene geometry Camera trajectory

01 / RESEARCH FOCUS

Understanding the world in 3D.

Connecting learned visual priors with
the structure of multi-view geometry.

Geometry & depth

Multi-view stereo and monocular depth priors for complete, consistent scene geometry.

Mapping & localization

SLAM and camera pose estimation connecting image sequences to spatial understanding.

Neural reconstruction

3D Gaussian Splatting and view selection for detailed, online scene reconstruction.

THE RESEARCH PIPELINE

Images → spatial understanding

  1. 01 / Observe

    Multi-view RGB

  2. 02 / Localize

    SLAM / Camera pose

  3. 03 / Estimate

    Depth / Multi-View Stereo

  4. 04 / Represent

    Gaussian Splatting

  5. 05 / Reconstruct

    3D Reconstruction

02 / SELECTED PUBLICATIONS

From ideas to published research.

All publications

Published / Accepted

Research overview: MVS-GS: High-Quality 3D Gaussian Splatting Mapping via Online Multi-View Stereo
IEEE Access · 2025 Published

MVS-GS: High-Quality 3D Gaussian Splatting Mapping via Online Multi-View Stereo

First Author

Bringing dense multi-view geometry into online Gaussian Splatting for high-quality 3D mapping.

Paper Demo video
Research overview: Stereo-GS: Online 3D Gaussian Splatting Mapping Using Stereo Depth Estimation
MDPI Electronics · 2025 Published

Stereo-GS: Online 3D Gaussian Splatting Mapping Using Stereo Depth Estimation

Co-first Author

Integrating stereo depth estimation into online Gaussian Splatting mapping.

Paper

ONGOING RESEARCH

Under Review

M2Depth IEEE Transactions on Multimedia (TMM) Under Review · First Author

03 / FEATURED PROJECTS

Research, out in the world.

All projects
Drone imagery and reconstructed scene from the online and offline mapping project

Mobility One Inc. · Oct. 2024 – Apr. 2025

From drone imagery to a 3D map.

SLAM / Multi-View Stereo / Gaussian Splatting

Combine SLAM camera tracking with multi-view stereo in an online/offline reconstruction pipeline.

Explore the project
Diffusion-assisted completion pipeline for drone-based 3D reconstruction

Mobility One Inc. · Jun. 2025 – Dec. 2025

Reconstructing what the camera missed.

COLMAP / OpenMVS / Diffusion

Use COLMAP and OpenMVS with diffusion-based image interpolation and inpainting.

Explore the project

EXPERIENCE

Research in practice.

  1. Sep. 2026 – Present

    KETICurrent

    Researcher

  2. Jul. 2026 – Jul. 2026

    StradVision

    Researcher

  3. May 2026 – Jul. 2026

    MotifDrive

    Researcher

  4. Sep. 2022 – Feb. 2024

    RTM AI Image

    2D Computer Vision Researcher

    Manufacturing defect detection, image segmentation, model optimization, and few-shot auto labeling. Progressed from research intern to data scientist.

EDUCATION

A foundation in vision.

  1. Mar. 2024 – Feb. 2026

    Dongguk University

    M.S. in Computer Science and Artificial Intelligence

    Advisor: Prof. Soohwan Song

    Thesis: Online 3D Gaussian Splatting Modeling with Novel View Selection

  2. Mar. 2020 – Feb. 2023

    Hanyang University ERICA

    B.S. in Software Engineering

LET’S CONNECT

Better geometry.
New perspectives.

For research discussions and collaborations in 3D vision.

CV (PDF) Google Scholar GitHub LinkedIn