cv

Ph.D. (Advisors: Prof. Jungbeom Lee and Prof. Jinkyu Kim)

Korea University

Major: Artificial Intelligence Applications

B.S.

Yonsei University

Major: Computer Science

• Operating Systems
• Computer Architecture
• Algorithms
• Data Structures
• Computer Graphics
• Object-Oriented Programming
• Artificial Intelligence
• Theory of Computation

Planner Developer – Autonomous Driving Development Team

KakaoMobility

Developing planning algorithms for real-vehicle autonomous driving.

• Developing planning algorithms that transform perception, prediction, map, and vehicle-state inputs into safe and natural autonomous driving behavior and trajectories.
• Analyzing scenario-level failures and improving behavior decision, trajectory generation, real-time integration, and verification logic for real-vehicle deployment.

Research Intern – Robot & AD Localization

NAVER LABS

Led a streaming 3D occupancy prediction project for camera-based 3D scene understanding (StreamOcc, ECCV 2026).

• Defined the latency, memory, and coordinate-alignment issues in multi-frame fusion.
• Designed stream-based voxel aggregation and query-guided refinement to preserve dense spatial representations while improving real-time efficiency and dynamic-object modeling.

Teaching Assistant

Korea University

Teaching assistant for undergraduate computer science courses.

• Algorithm (COSE214), Fall 2022 and Fall 2023
• Mathematics for Computer Science (COSE111), Spring 2024

Research Intern – Motion Prediction Team

42dot

Developed interaction-aware trajectory prediction models for multi-agent intersection scenarios.

• Designed a future-aware teacher-student distillation framework that transfers interaction knowledge from privileged future observations to an efficient inference-time model.
• Published at ICPR 2024 and the CVPR 2023 Workshop on Vision-Centric Autonomous Driving.

Research Intern

Korea Univ. Vision & AI Lab

Worked on multi-camera 3D object detection and occupancy prediction.

• Worked on ORA3D and Waymo occupancy-challenge baselines.
• Built perception data processing and evaluation pipelines for 3D detection and occupancy prediction experiments.

Intern – Face Detection Project

Samsung Electronics

Implemented an attention-based few-shot face detection algorithm for limited-data settings.

• Built and curated a face dataset and designed augmentation pipelines covering scale, crop, color, and blur variations.

Streaming Dense Voxel Representations for 3D Occupancy Prediction

Seokha Moon, Janghyun Baek, Yujin Jeong, Daewon Chae, Giseop Kim, Jungbeom Lee, Jinkyu Kim, Sunwook Choi

European Conference on Computer Vision (ECCV)

StreamOcc, a dense voxel streaming framework for accurate and efficient 3D occupancy prediction, reducing warping artifacts while selectively injecting instance-level query features into occupied voxel regions. Work done during an internship at NAVER LABS.

Image-Guided Semantic Pseudo-LiDAR Point Generation for 3D Object Detection

Minseung Lee, Seokha Moon, Seung Joon Lee, Reza Mahjourian, Jinkyu Kim

IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

An image-guided pseudo-LiDAR point generation method that improves 3D object detection for sparse and distant objects while reducing false positives.

ALIGN: Advanced Query Initialization with LiDAR-Image Guidance for Occlusion-Robust 3D Object Detection

Janghyun Baek, Mincheol Chang, Seokha Moon, Seung Joon Lee, Jinkyu Kim

arXiv Preprint

An object-aware, LiDAR–image guided query initialization strategy that significantly improves detection of occluded and crowded objects without modifying detector architectures.

Causality-Aware End-to-End Autonomous Driving via Ego-Centric Joint Scene Modeling

Seokha Moon, Minseung Lee, Joon Seo, Jinkyu Kim, Jungbeom Lee

arXiv Preprint

An end-to-end autonomous driving framework that jointly learns the ego vehicle and interaction-relevant agents, then improves the ego driving policy with closed-loop reinforcement learning.

RECAST: Recasting Vision-Language Semantics into an Actionable Cost Map for Robot Navigation

Incheol Cho*, Jintae Park*, Jinkyu Kim, Jungbeom Lee, Jaegul Choo†, Seokha Moon†

arXiv Preprint

A robot navigation framework that combines the reasoning of a vision-language model with the spatial grounding of vision foundation models to build an actionable cost map, judging which surfaces are traversable, which objects pose a risk, which heading to prefer, and which gaps are passable.

SUPER-AD: Semantic Uncertainty-aware Planning for End-to-End Robust Autonomous Driving

Wonjeong Ryu, Seungjun Yu, Seokha Moon, Hojun Choi, Junsung Park, Jinkyu Kim, Hyunjung Shim

arXiv Preprint

Uncertainty-aware BEV planning with lane-following priors for safer and more robust camera-only end-to-end autonomous driving.

BEVMap: Map-Aware BEV Modeling for 3D Perception

Mincheol Chang, Seokha Moon, Reza Mahjourian, Jinkyu Kim

IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

Map-aware BEV modeling that incorporates map priors into 3D perception.

Learning Temporal Cues by Predicting Objects Move for Multi-camera 3D Object Detection

Seokha Moon, Hongbeen Park, Jungphil Kwon, Jaekoo Lee, Jinkyu Kim

IEEE International Conference on Robotics and Automation (ICRA)

A Detection After Prediction (DAP) method that explicitly learns temporal cues for multi-camera 3D object detection.

VisionTrap: Vision-Augmented Trajectory Prediction Guided by Textual Descriptions

Seokha Moon, Hyun Woo, Hongbeen Park, Haeji Jung, Reza Mahjourian, Hyung-gun Chi, Hyerin Lim, Sangpil Kim, Jinkyu Kim

European Conference on Computer Vision (ECCV)

Incorporates visual features from surround-view cameras and VLM-generated textual descriptions as supervision for trajectory prediction.

Who Should Have Been Focused: Transferring Attention-based Knowledge from Future Observations for Trajectory Prediction

Seokha Moon, Kyuhwan Yeon, Hayoung Kim, Seong-Gyun Jeong, Jinkyu Kim

International Conference on Pattern Recognition (ICPR)

An attention-based teacher–student method that distills future-aware interaction knowledge to improve multi-agent trajectory prediction.

RUFI: Reducing Uncertainty in Behavior Prediction with Future Information

Seokha Moon, Sejeong Lee, Hyun Woo, Kyuhwan Yeon, Hayoung Kim, Seong-Gyun Jeong, Jinkyu Kim

CVPR Workshop on Vision-Centric Autonomous Driving (VCAD)

Reduces uncertainty in behavior prediction by leveraging future information during training.

Enhancing Trajectory Prediction Accuracy with Goal Location and Lane Information

Seokha Moon, Jinkyu Kim

Conference on The Institute of Electronics and Information Engineers (IEIE), Oral

Predicts the goal location and interacts with lane information to navigate towards the predicted goal.

ORA3D: Overlap Region Aware Multi-view 3D Object Detection

Wonseok Roh, Gyusam Chang, Seokha Moon, Giljoo Nam, Chanyoung Kim, Younghyun Kim, Jinkyu Kim, Sangpil Kim

British Machine Vision Conference (BMVC)

Overlap-region-aware multi-view 3D object detection.

First Prize in Computer Science

Yonsei University

Awarded first place in the Department of Computer Science for academic excellence.

HUAWEI Scholarship

Talent Development Foundation

Awarded for academic excellence and potential in technology research.