SEP 2024 — PRESENT
Postdoctoral Researcher
Korea Institute of Science and Technology (KIST)
- Vision-based system that reconstructs game states from broadcast video through multi-object tracking, player identification, and geometric mapping for dynamic scene understanding
- Multimodal healthcare AI with IMU sensors and pose keypoints
Computer visionMulti-object trackingHomography estimationRe-IDVLM
FEB 2021 — AUG 2024
Doctoral Researcher
Korea Advanced Institute of Science and Technology (KAIST)
- Multimodal representation learning across text, audio, and video for emotion and sentiment understanding, with robust cross-modal alignment under noisy modalities and missing labels
- LLM-based task planning
Cross-modal alignmentSensor fusionDomain generalizationMultimodal learningAffective computingLLM
FEB 2019 — FEB 2021
Master’s Researcher
Korea Advanced Institute of Science and Technology (KAIST)
- Contactless text input from finger movements and mobile input systems
- Vision-based industrial defect classification
HCIDefect detection
