Curriculum Vitae
📄 Download CV (PDF) | Google Scholar | GitHub | LinkedIn
Research summary
Master’s researcher at UNIST working on 3D human avatar reconstruction, 3D Gaussian splatting, and vision–language reasoning for hand understanding. Recent first-author work (ECCV 2026, under review) proposes a visibility-aware 3D Gaussian avatar framework that reconstructs animatable avatars from monocular video across full-body, upper-body, and head-only settings within a single pipeline — integrating occlusion-robust SMPL-X tracking, part-specific residual refinement, and diffusion-based texture completion. Additional contributions span hand pose benchmarking (CVPR 2026), hand–object mesh generation (CVPR Findings 2026), real-time two-hand manipulation (AAAI 2025), and generative replay for continual detection (CVPR 2024 Highlight).
Research interests
3D Gaussian Splatting · 3D Human Avatar Reconstruction & Animation · Visibility-Aware Optimization · Diffusion Models · SMPL-X / FLAME Parametric Body Models · Hand Pose Estimation & 3D Hand–Object Interaction · Vision–Language Models.
Education
- M.S. in Computer Science and Engineering, Ulsan National Institute of Science & Technology (UNIST), Sep 2024 – Aug 2026 (expected). GPA: 4.1 / 4.3 (98 / 100).
- B.S. in Computer Science and Engineering, Ulsan National Institute of Science & Technology (UNIST), Sep 2020 – Jun 2024. Cum Laude.
Selected research highlights
- First author, ECCV 2026 (under review): Visibility-aware 3D Gaussian avatar framework — state-of-the-art across full-body, upper-body, and head-only settings, with ~3% PSNR gain and up to 50% memory reduction.
- Co-author, CVPR 2026: HandVQA — 1.6M-scale benchmark for diagnosing spatial reasoning failures in vision–language models via hand pose question answering.
- Co-author, CVPR 2026 Findings: THOM — text-conditioned generative framework for physically plausible hand–object mesh interactions.
- Co-author, AAAI 2025: QORT-Former — real-time 3D two-hand manipulation modeling (53.5 FPS; +27.2% H2O; +10.4% FPHA).
- Co-author, CVPR 2024 Highlight (top 2.8%): SDDGR — Stable Diffusion-based generative replay for class-incremental object detection.
Publications
The full list is available on the Publications page.
Research experience
Vision & Learning Lab, UNIST — Graduate Research Assistant (previously Research Assistant) Jan 2023 – Present · Ulsan, South Korea · Advisors: Prof. Seungryul Baek, Prof. Binod Bhattarai.
- Led development of a first-author visibility-aware 3D Gaussian avatar reconstruction framework (ECCV 2026, under review).
- Contributed to HandVQA (CVPR 2026) and THOM (CVPR Findings 2026) on 3D hand understanding and hand–object interaction generation.
- Co-authored SDDGR (CVPR 2024 Highlight), a Stable-Diffusion-based generative replay method for class-incremental object detection, and QORT-Former (AAAI 2025), a real-time transformer for 3D two-hand manipulation.
- Implemented transformer-based approaches for high-fidelity 3D facial texture refinement and designed Trimesh-based pipelines for hand–object interaction visualization.
- Optimized MobileNetV3-Small for 3D face reconstruction, achieving 50% faster inference and 25% accuracy improvement.
Machine Learning, Vision & Language Lab, UNIST — Research Assistant Sep 2022 – Dec 2022 · Mentor: Prof. Taehwan Kim.
- Led a CV research initiative on astronomical image clarity enhancement using GANs, improving image detail by 60%.
- Evaluated deep learning models using accuracy, sensitivity, specificity, and precision metrics.
Bio-Optics & Computational Imaging Lab, UNIST — Research Assistant Sep 2021 – Dec 2021 · Mentor: Prof. Jung-Hoon Park.
- Conducted research on non-line-of-sight imaging using Ghost Imaging techniques.
- Applied machine learning algorithms to extract scattering resistance modes, achieving 95% accuracy.
Selected research projects
- Visibility-Aware 3D Gaussian Human Avatar Reconstruction — Vision & Learning Lab, UNIST (Jan 2025 – Present). Unified Gaussian-splatting pipeline for animatable avatars across full-body, upper-body, and head-only inputs; visibility-aware optimization (–50% memory, +34% FPS); occlusion-robust SMPL-X co-registration; diffusion-based 360° view synthesis. Code.
- Accelerating Inference Speed for 3D Face Reconstruction — Vision & Learning Lab, UNIST (Aug 2023 – Aug 2024). Sparsity-aware quantization, ViT backbone optimization, and CNN student networks via knowledge distillation. Code.
- Mesh Reconstruction & Visualization on the ARCTIC Dataset — Vision & Learning Lab, UNIST (Feb 2023 – Jul 2023). FastInst implementation and rendering pipelines for hand–object manipulation evaluation. Code.
Technical skills
- Core: PyTorch, CUDA, Distributed / Accelerated Training, Linux, Git, LaTeX.
- Generative & Multimodal Models: Diffusion Models, HuggingFace Transformers.
- 3D Vision & Rendering: 3D Gaussian Splatting, NeRF, SMPL-X / FLAME, differentiable rendering, multi-view geometry, pose estimation, human/hand modeling.
- Data & Tooling: NumPy, Pandas, OpenCV, Trimesh, experiment scripting and evaluation pipelines.
Teaching experience
Honors & awards
- Graduate School Scholarship (Korean Government) — full tuition + stipend for M.S.
- UNIST Dream Scholarship (UNIST) — full tuition + stipend for B.S.
References
- Prof. Seungryul Baek — Supervisor (M.S.). Associate Professor, Artificial Intelligence Graduate School (AIGS) and Department of Computer Science and Engineering, UNIST. Website · srbaek@unist.ac.kr
- Prof. Binod Bhattarai — Co-supervisor (M.S.). Lecturer, University of Aberdeen, UK; Honorary Lecturer, University College London, UK. Website · b.bhattarai@ucl.ac.uk