Byeonghyun Pak

Hi! I'm Byeonghyunpronounced "Byung-hyun", a first-year Ph.D. student in Computer Science at the University of Maryland, advised by Prof. Furong Huang. Previously, I served as a research officer at the Agency for Defense Development (ADD), working on vision-language models and visual representation learning. I received my bachelor’s degree from Daegu Gyeongbuk Institute of Science and Technology (DGIST).

My research goal is to build robots that can reason over long-horizon, complex tasks and generalize robustly to real-world environments. To this end, I develop structured state representations for robotic perception, policy learning, and long-horizon planning.

Please feel free to reach out to me at bhpak@umd.edu.

Byeonghyun Pak

Recent News

view all
  • [Sep 2026]Our paper was accepted to CoRL 2026!
  • [Aug 2026]I started my PhD at UMD!
  • [Jun 2026]I completed three years of mandatory military service at ADD. Grateful for my time there! 🥳
  • [Jun 2026]Our paper received the Best Paper Award at the ICRA 2026 Workshop on MM-SpatialAI! 🎉

Selected Publications

view all

SERF: Spatiotemporal Environment and Robot Feature Map for Long-Horizon Mobile Manipulation

Sunghwan Kim*, Byeonghyun Pak*, Kehan Long, Yulun Tian, Nikolay Atanasov

Conference on Robot Learning (CoRL), 2026

ICRA Workshop on Multi-Modal Spatial AI, 2026 (Best Paper Award)

paperwebsitecode

Pixel-level Scene Understanding in One Token: Visual States Need What-is-Where Composition

Seokmin Lee, Yunghee Lee, Byeonghyun Pak, Byeongju Woo

CVPR Workshop on Pixel-level Video Understanding in the Wild, 2026

paperwebsitecode

Textual Query-Driven Mask Transformer for Domain Generalized Segmentation

Byeonghyun Pak*, Byeongju Woo*, Sunghwan Kim*, Dae-hwan Kim, Hoseong Kim

European Conference on Computer Vision (ECCV), 2024

paperwebsitecode
B-spline Texture Coefficients Estimator for Screen Content Image Super-Resolution demo

B-spline Texture Coefficients Estimator for Screen Content Image Super-Resolution

Byeonghyun Pak*, Jaewon Lee*, Kyong Hwan Jin

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023

Selected as a highlight paper (Top 2.5%)

paperwebsitecode