Flow-SfM
Flow-Gated Structure from Motion with Dense UFM Correspondences (submitted)
Flow-gated Structure-from-Motion using dense UFM correspondences for robust camera pose estimation and dense 3D structure recovery.
3D Computer Vision Student Researcher
M.S. in Artificial Intelligence, Robotics and Computer Vision Lab
University of Seoul
3D Computer Vision · 3D Reconstruction · Gaussian Splatting · Structure-from-Motion · Neural Rendering
Flow-Gated Structure from Motion with Dense UFM Correspondences (submitted)
Flow-gated Structure-from-Motion using dense UFM correspondences for robust camera pose estimation and dense 3D structure recovery.
To be updated
This section will be updated as new experiences are added.
University of Seoul · 2023–2025
Jetson-based lab setup and maintenance, TensorRT examples, and technical support for students in the Intelligent Robots course.
1st Place · 2023
Developed a winning computer vision pipeline including data preprocessing, model design, and ensemble strategies.