🎓 PhD Researcher in the Perception and Intelligent Systems (PercInS) group at Munich Institute of Robotics and Machine Intelligence (TUM MIRMI), supervised by Prof. Achim J. Lilienthal under the MSCA ENGAGE Project
🎓 Master of Science in Computer Vision (Soongsil University, South Korea)
💼 Previously AI Researcher at DeltaX (Seoul), a startup focused on cutting-edge vision-based AI systems
🌏 Passionate about bringing intelligence to the real world using deep learning, robotics, and computer vision
💡 Key areas of interest:
3D Object Detection, Point Cloud Processing, Scene Understanding, Panoptic Segmentation, Point Cloud Reconstruction, Robot Perception
- Member of the Perception and Intelligent Systems (PercInS) group at TUM MIRMI
- Supervised by Prof. Achim J. Lilienthal under the MSCA ENGAGE doctoral network
- Research focus on robot perception, 3D scene understanding, and intelligent systems for real-world robotics
- Worked on real-time AI perception systems for autonomous platforms and smart infrastructure
- Designed and optimized high-performance deep learning pipelines for vision-based tasks
- Leveraged frameworks like PyTorch, TensorRT, OpenCV, MMDetection3D, MMDet, Detectron2, OpenPCDet, and others for production-ready AI
- Focused on both image-based and LiDAR-based solutions, with deployment on embedded devices (Jetson Orin, Xavier)
- Conducted research on 3D object detection and multi-object tracking for autonomous driving applications
- Explored architectures involving transformers, sparse convolution, and panoptic segmentation
- Wrote and published multiple peer-reviewed papers on top-ranking benchmark results (KITTI, etc.)
- Developed deep learning models for visual inspection and damage detection
- Participated in the end-to-end system deployment including cloud integration and model scaling
- Simulated robotic motion planning and trajectory optimization on quadruped robots
- Utilized tools such as ROS, Gazebo, and MATLAB for robotics experiments
- Built real-time object recognition systems using deep learning for edge platforms
- Worked on projects involving facial recognition and vehicle detection with lightweight neural networks
- Languages: Python, C++, Matlab, C#
- Frameworks: PyTorch, TensorFlow, Keras, Detectron2, OpenPCDet, Ultralytics
- Libraries: OpenCV, Open3D, TorchVision, NumPy, Matplotlib, Mayavi
- Edge & Acceleration: TensorRT, ONNX, CUDA, OpenVINO
- Tools: Git, Docker, VSCode, Jupyter, PyCharm
- OS: Ubuntu, Jetson Linux
- 🏅 TSSTDet: Transformation-based 3D Object Detection via Spatial Shape Transformer — IEEE Sensors Journal (Q1) – First Author
- 🏅 3ONet: 3-D Detector for Occluded Object Under Obstructed Conditions — IEEE Sensors Journal (Q1) – First Author
- 🧠 AFMtrack: Attention-Based Feature Matching for MOT — IEEE Access (Q1) – Second Author
- 📦 Shape-Aware 3D Detection — KICS Conference – First Author
- 🧩 ESSDet: Enhancing Spatial Shape for 3D Detection — ICOIN Conference – First Author
- 🧠 CAMTrack: a combined appearance-motion method for multiple-object tracking — Machine Vision & Applications Journal (Q2) – Second Author
- 💻 C++ – SoloLearn
- 🐍 Python Core – SoloLearn
- 📊 Machine Learning – SoloLearn
- 🤖 Machine Learning with Python – IBM (Coursera)
- 🚗 Introduction to Self-Driving Cars – University of Toronto (Coursera)
- 🧠 Neural Networks and Deep Learning – DeepLearning.AI
- 🔍 Deep Neural Networks with PyTorch – IBM (Coursera)
- 👁️ Visual Perception for Self-Driving Cars – University of Toronto (Coursera)
“Driven by curiosity and guided by data, I build efficient, scalable, and real-time computer vision systems to empower intelligent machines.”