RADIAL
VOXELIZATION
VOXELIZATION
Radial Scaling Voxelization for Accurate Small Object 3D Detection
International Conference on Machine Learning, 2026. ICML 2026
Hao Liu
刘浩Research Professor (Zijiang Young Scholar) Assistant to the Dean of
School of Geospatial Artificial Intelligence, East China Normal University (ECNU), Shanghai, China
青年研究员（"双百人才计划"紫江青年学者），博导，院长助理
华东师范大学空间人工智能学院
Room B501, Estuary and Coastal Building
Minhang Campus, Shanghai 200241
hliu at geoai.ecnu.edu.cn
I am a Zijiang Young Scholar, Principal Investigator, and Ph.D. Supervisor at the School of Geospatial Artificial Intelligence, East China Normal University (ECNU). My research interests lie in spatial intelligence, 3D vision, point cloud understanding, 3D object detection and tracking, scene understanding, and 3D reconstruction. The overarching goal of my research is to enable machines to perceive, understand, and reconstruct complex three-dimensional environments reliably.
Previously, I was a Postdoctoral Research Fellow at Nanyang Technological University, working with Prof. Ying He. I received my Ph.D. from Sun Yat-sen University in 2023, supervised by Prof. Yulan Guo, my M.E. from the National University of Defense Technology in 2018, and my B.E. from the University of Electronic Science and Technology of China in 2016.
OPENINGS
Ph.D. / Master / RA — I am always looking for self-motivated students and research interns to work on 3D vision and spatial intelligence, covering topics such as point cloud understanding, 3D detection and tracking, scene understanding, and reconstruction.
Interested? Send me your resume and transcripts.
News
Our paper Radial Scaling Voxelization for Accurate Small Object 3D Detection was accepted to ICML 2026.
Our monograph Point Cloud Intelligence was published.
Our paper Edge-Centric Relational Reasoning for 3D Scene Graph Prediction was accepted to AAAI 2026.
Our paper on geometry and semantics for camera-based 3D semantic scene completion was accepted to IEEE TIP.
I joined East China Normal University as a Zijiang Young Scholar.
Our paper SF-City was accepted to IEEE TMM.
Our semantic scene completion paper was accepted to IEEE TCSVT.
Two papers were accepted to ECCV 2024 and IEEE TCSVT.
Our multi-frame monocular depth estimation paper was accepted to IEEE TCSVT.
AnchorPoint was accepted to IEEE TITS.
CenterTube was accepted to IEEE TMM.
Publications
2026
Edge-Centric Relational Reasoning for 3D Scene Graph Prediction
AAAI Conference on Artificial Intelligence, 2026. AAAI 2026
2025
3D SEMANTIC
COMPLETION
COMPLETION
Enhanced Geometry and Semantics for Camera-Based 3D Semantic Scene Completion
IEEE Transactions on Image Processing, 2025. TIP
SF-CITY
SF-City: A Source-Free Domain Adaptation Method for City-scale Point Cloud Semantic Segmentation
IEEE Transactions on Multimedia, 2025. TMM
2024
SEMANTIC-AWARE
GUIDANCE
GUIDANCE
Semantic Scene Completion via Semantic-aware Guidance and Interactive Refinement Transformer
IEEE Transactions on Circuits and Systems for Video Technology, 2024. TCSVT
POINT CLOUD
COMPLETION
COMPLETION
Point Cloud Completion via Self-projected View Augmentation and Implicit Field Constraint
IEEE Transactions on Circuits and Systems for Video Technology, 2024. TCSVT
Heterogeneous Graph Learning for Scene Graph Prediction in 3D Point Clouds
European Conference on Computer Vision, 2024. ECCV 2024
2023
MONOCULAR
DEPTH
DEPTH
Self-Supervised Multi-Frame Monocular Depth Estimation for Dynamic Scenes
IEEE Transactions on Circuits and Systems for Video Technology, 2023. TCSVT
AnchorPoint: Query Design for Transformer-Based 3D Object Detection and Tracking
IEEE Transactions on Intelligent Transportation Systems, 2023. TITS
CENTERTUBE
CenterTube: Tracking Multiple 3D Objects with 4D Tubelets in Dynamic Point Clouds
IEEE Transactions on Multimedia, 2023. TMM
Earlier publications
- DeepPCT: Single Object Tracking in Dynamic Point Cloud Sequences.
- Spatial-Temporal Transformer for 3D Point Cloud Sequences.
- Detecting and Tracking Small and Dense Moving Objects in Satellite Videos: A Benchmark.
- Semantic Context Encoding for Accurate 3D Point Cloud Segmentation.
- Deep Learning for 3D Point Clouds: A Survey.
- Robust Long-Term Tracking via Instance Specific Proposals.
- Global Context Reasoning for Semantic Segmentation of 3D Point Clouds.
- Long-Term Object Tracking with Instance Specific Proposals.
About
Appointments & Education
- 2025–present: Zijiang Young Scholar, ECNU
- 2023–2025: Postdoctoral Research Fellow, NTU
- 2023: Ph.D., Sun Yat-sen University
- 2018: M.E., National University of Defense Technology
- 2016: B.E., University of Electronic Science and Technology of China
Selected Honors
- Excellent Graduate, Sun Yat-sen University, 2023
- Wang Shujun Donation Scholarship, 2023
- National Scholarship, Ministry of Education, China, 2020
- Third Prize, National Post-Graduate Mathematical Contest in Modeling, 2018