|
Hojun Song My research interests are 3D perception, 3D reconstruction, and model compression, and I am a first-year PhD student in Video Intelligence Lab (VILAB) in Department of Computer Science and Engineering at Kyungpook National University, Korea. Email: hojunsong@knu.ac.kr / CV / Scholar / GitHub / Linkedin |
|
Research |
|
G2P: Gaussian-to-Point Attribute Alignment for Boundary-Aware 3D Segmentation
Hojun Song*, Chae-yeong Song*, Jeong-hun Hong, Chaewon Moon, Soo Ye Kim, Yiyi Liao, Jaehyup Lee, and Sang-hyo Park† Under Review, 2026 project page / paper / code G2P aligns 3D Gaussian attributes with point clouds to enhance appearance-aware learning and boundary localization, improving segmentation of geometrically ambiguous 3D scenes. |
|
CompSplat: Compression-aware 3D Gaussian Splatting for Real-world Video
Hojun Song*, Heejung Choi*, Chae-yeong Song, Gahyeon Kim, Soo Ye Kim†, Jaehyup Lee†, and Sang-hyo Park† Under Review, 2026 project page / paper A compression-aware 3D Gaussian Splatting framework for real-world video novel view synthesis that exploits frame-wise compression information to improve rendering quality and pose stability under compressed video conditions. |
|
Compression Framework for Light 3D Scene Graph Generation via Pruning-As-Search and Distillation
Hojun Song*, Chae-yeong Song*, Dong-hun Lee, Heejung Choi, Jinwoo Jeong, Sungjei Kim, and Sang-hyo Park† IEEE Transactions on Multimedia, 2026 project page / paper / code A lightweight compression framework for GNN-based 3D scene graph generation that integrates pruning-as-search and knowledge distillation to reduce computational complexity while preserving classification performance. |
|
Condition-based Synthetic Dataset for Amodal Segmentation of Occluded Cucumbers in Agricultural Images
Jin-Ho Son*, Hojun Song, Chae-yeong Song, Minse Ha, Dabin Kang, and Yu-Shin Ha† Computers and Electronics in Agriculture, 2025 paper / code A condition-based synthetic dataset generation framework for amodal segmentation that models realistic occlusion and illumination variations to improve robust crop segmentation in agricultural environments. |
|
Automatic Classification of Disaster Images Based on Deep Learning
Hojun Song*, Dong-hun Lee, Han-Gyul Baek, Byungjun Bae, and Sang-hyo Park† Journal of Korean Institute of Communications and Information, 2023 paper |
|
|
Template adapted from Jon Barron's public academic website. |