Sungjae Lee (이성재)

I recently received my M.S. in Artificial Intelligence POSTECH Graduate School of AI (POSTECH GSAI), where I was a member of the Machine Learning and Vision Group under the supervision of Prof. Kwang-In Kim. I previously received my B.S. in Biomedical Engineering from Gachon University.

My current research focuses on the intersection of computer vision and robotic manipulation, with particular emphasis on robotic grasping and the application of commonsense knowledge. I am also open to exploring other emerging areas within artificial intelligence and robotics.

Email  /  CV  /  Scholar  /  Github  /  Linkedin

profile photo

Publications

mass
Physically Guided Visual Mass Estimation from a Single RGB Image
Sungjae Lee, Junhan Jeong, Yeonjoo Hong, Kwang In Kim
arXiv preprint, 2026

A physically structured framework for ill-posed single-image mass estimation that aligns visual cues with the physical factors governing mass.

zosah
Subspace-based Approximate Hessian Method for Zeroth-Order Optimization
Dongyoon Kim, Sungjae Lee, Wonjin Lee, Kwang In Kim
arXiv preprint, 2025

A zeroth-order optimization algorithm that mitigates the cost of function evaluations by focusing on randomly selected two-dimensional subspaces.

graspcorrect
GraspCorrect: Robotic Grasp Correction via Vision-Language Model-Guided Feedback
Sungjae Lee*, Yeonjoo Hong*, Kwang In Kim
arXiv preprint, 2025

A policy-agnostic add-on module designed to enhance grasp performance through vision-language model-guided feedback.

table
Piece of Table: A Divide-and-Conquer Approach for Selecting Sub-Tables in Table Question Answering
Wonjin Lee, Kyumin Kim, Sungjae Lee, Jihun Lee, Kwang In Kim
arXiv preprint, 2024

A sub-table-based question answering framework using divide, conquer, and combination steps.

tonguecancer
Enhancing deep learning classification performance of tongue lesions in imbalanced data: mosaic-based soft labeling with curriculum learning
Sungjae Lee*, Hyun Jun Oh*, Young-Don Son, Jong-Hoon Kim, Ik-Jae Kwon, Bongju Kim, Jong-Ho Lee, Hang-Keun Kim
BMC Oral Health, 2024

A mosaic-based soft labeling augmentation method for class-imbalanced tongue lesion classification.

2022-IEEE-NSS-MIC
Image Reconstruction Network with Sinogram-to-Image Transform Layer for Projection Data
Sungjae Lee, Yong-Gi Hong, Tae-Jin Kim, M. Kim, Hang-Keun Kim, Kyungsang Kim, Young-Don Son
Poster (on-site presentation), 2022

An image reconstruction from positron emission tomography (PET) sinogram data.

Education

Pohang University of Science and Technology  |  Pohang, South Korea
Mar. 2024 - Feb. 2026
M.S. in Artifical Intelligence, advised by Prof. Kwang-In Kim.
Gachon University  |  Seongnam, South Korea
Mar. 2018 - Feb. 2024
B.S. in Biomedical Engineering (Summa Cum Laude).

Work experiences

Machine Learning and Vision Group (MLV Group)  |  Research student
Mar. 2024 - Feb. 2026
Vision-based robot interaction.
Neurophet AI Team  |  Research intern
Aug. 2023 - Dec. 2023
Development of a deep-learning model for automated cerebral microbleed segmentation.
Multi-Modality Medical Imaging Lab (MMMIL)  |  Research student
Apr. 2021 - Dec. 2023
Tongue cancer classification.
Image reconstruction from positron emission tomography (PET) sinogram data.
Also worked on malocclusion classification and brain vessel segmentation from MR angiography.