M.S. in Image Science at Chung-Ang University (IRIS Lab), 2026. Dynamic 3D reconstruction, speech synthesis pipelines, and time-series restoration — research that has to survive contact with a real system.
- Motion–Surface–Appearance Dynamic Gaussian Splatting for Accurate Geometry and High-Fidelity Rendering — IEEE TCSVT, under review (JCR Top 5.5%)
- Enhancing 3D Scene Representation with Structural Dissimilarity-Aware Learning — IEEE ICIP 2025
- 3D Head and Body Pose Estimation-Based Rehabilitation Exercise Assessment for Thyroid Patients — IEIE 2025 · 🏆 Best Student Paper
- Deep Learning-Based Interpolation for Time-Series Data with Continuous Missing Values — KICS 2023
| Project | The problem | Result |
|---|---|---|
| MSA-DGS | Motion, geometry and appearance are entangled in one representation — improving one breaks the others | +11.25 dB PSNR · −0.024 CD |
| neck-rehab-pose-estimation | Absolute head angle reads 45° even when the neck barely moved | 6.5° MAE · 🏆 |
| llm-tts-talking-face | Latency was the sum of every stage instead of the max | Sentence-level pipelining |
| mallang-voice-storybook | Voice cloning needed 300 recorded sentences and a model per user | 10–20s audio · 4s → <1s |
| imputation-project | 23% missing, 4-day gaps, and no ground truth to score against | R² 0.94 |
2024–2026 M.S. Image Science, Chung-Ang University · IRIS Lab GPA 4.3/4.5
2023 Researcher, Korea Electronics Technology Institute (KETI) · SKT FLY AI 4th
–2024 B.S. Myongji University GPA 4.0/4.5 · valedictorian
🏆 Best Student Paper, IEIE 2025 · 🥇 Valedictorian, 2024 · Azure AI-900 · ADsP
Open to opportunities in AI research and engineering.